The long tail theory states that there is a shift in today’s culture and economy from a relatively small number of mainstream products called “hits” towards a huge number of niches. The e-commerce has the tremendous advantage of the so called infinite shelf space giving less popular products the possibility to arise and attract consumers as distribution costs especially shelf… Read more
The long tail theory states that there is a shift in today’s culture and economy from a relatively small number of mainstream products called “hits” towards a huge number of niches. The e-commerce has the tremendous advantage of the so called infinite shelf space giving less popular products the possibility to arise and attract consumers as distribution costs especially shelf space costs are disappearing. Niche products have the ability to better target interests and thus, are said to represent a considering competition for the hits. (1)
In the context of the growing digital market, consumers are facing an unlimited range of products which why is deciding on one product often represents an overwhelming task for the consumer. This is where recommendation systems have come into existence. Indeed, recommendation systems, often based upon past purchases, ease the product search for the consumer as they help him reducing its choice. Moreover, they reinforce the customer’s engagement towards the brand or the firm by offering personalized product information and suggestions. (2)
In the following part, we will discuss if recommendation systems lead to an increase in sales in the long tail and thus, support the theory of the long tail, or not.
To start with, in the context of the digital area, we are moving away from a world of scarcity to a world of abundance. As stated by Anderson’s theory of the the long tail, the digital market offers unlimited choice for the consumer as there is space for everyone. Thus, as everyone’s taste departs at some point from the mainstream, the online retail and distribution offers the consumer the possibility to discover new products, niches, related to his specific interests. Besides this, Anderson claims that there is a strong demand for these niche products. (3).
However, recent studies have observed that recommendation systems do not promote niche products. Rather than boosting sales diversity, recommendation systems lead to an enhancement of the best selling products. In fact, recommendation system do push consumers towards new products, but they often push different consumers towards the same new product. This results in a richer-get-richer effect for popular products and unpopular products often remain undiscovered. (4)
The previous argument is affirmed by Marcelllo Vena who states that the concentration of the digital market hinders diversity. Indeed, the market share of the global players is continuously growing leaving the small and independent retailers with only a negligible part. Hence, the more a market is concentrated, the more it tends to promote best selling products. Recommendation system seem to follow this pattern as they often suggest items that are the most shared or the most sold favouring sales rather than diversity. Although Vena’s study is mainly focused on the digital book market, we can easily understand the concept behind and apply it to other sectors. (5)
Following these studies, recommendation systems shorten rather than lengthen the long tail.
To conclude, it seems clear that recommendation systems contribute to an increase in the total sales. However, opinions differ regarding whether recommendation system enhance sales diversity or strengthen the purchase of best selling products. Therefore, the impact of recommendation systems on the long tail can be evaluated as limited.
As far as I am concerned, Anderson’s theory of the long tail should be considered with precaution. Indeed, being a consumer in the digital market myself and using reviews for my purchases, I believe that recommendations have made me discovered items that otherwise I would not have found, but I have also observed that the same items are proposed again and again which confirms the richer-get-richer effect.
Along with the digital evolution, new phenomenons have arisen. This is the case for the so called ‘Long tail theory’ by Chris Anderson. This theory focuses on the strategy of selling a broad spectrum of products, each in small quantities.
It is commonly known that shops have a limited offer and prefer thus selling best sellers or hits as they… Read more
Along with the digital evolution, new phenomenons have arisen. This is the case for the so called ‘Long tail theory’ by Chris Anderson. This theory focuses on the strategy of selling a broad spectrum of products, each in small quantities.
It is commonly known that shops have a limited offer and prefer thus selling best sellers or hits as they face the issue of having a limited stock. Nevertheless, digital markets make it nowadays possible for online retailers to offer unlimited choice as their stock is virtual. Therefore, C. Anderson states that the easier the disappearance of physical shelf place is, the higher profits for unpopular goods will be. As a result, niche products’ profits can surpass mainstream products’ profits by satisfying narrow interests. In other words, everything is available on online selling platforms and unpopular products can remain in the tail forever with a chance to become popular one day. (2)
Bearing in mind online retailers have unlimited offer, they need to assist the customer in its researches by means enabling him to discover and explore that extra available data. This is the reason why they have set up recommender systems putting forward suggestions usually based on predictions of customers’ interests, past history of purchases, product searches or products’ rating.
The question we should ask ourselves is if recommendation systems rest on the fact that the success of business is to sell less in quantity but more in diversity ?
From a customer’s point of view, I do not have any idea of how large the available offer of the online retailer is. Therefore, thanks to the recommendation system, I will be suggested movies, songs or books I had never heard about before. Indeed, as a human being I do not have the knowledge of the broad offer of the sector. Up to this point, I would tend to state that recommendation systems insists on diversification and makes thus the tail longer.
However, we can notice that those suggestions are backed up by ratings of customers or number of views. Consequently, although I was not aware of the products’ existence, the product was already more or less popular on the platform. Therefore, I see in the recommendation system a kind of vicious spiral leading to the fact that we finally end up watching the same movies, listening to the same music or reading the same book. Indeed, we contribute to the fame of the same products being influenced by the suggestions based on certain factors.
Studies have observed that our economies are not as diverse as C. Anderson would affirm. But, what are the factors supporting that statement ?
According to Marcello Vena, the digital market concentration is to blame for the lack of diversity of sales in the book sector; big worldwide actors are increasing their market share becoming numeric monopolies and making it difficult for independent retailers. Therefore, as the market share for monopolies increase, the one of independent retailers decrease.(4)
Besides this, recommender systems mislead the customer by favouring the chances of future purchases instead of the discovery. (5) This phenomenon is understandable as retailers’ first aim is to boost their sales. Consequently, they use the recommendation tool to seek their interest.
For instance, in the book industry, e-commerce platform will tend to suggest another book of an author the customer had already bought from beforehand. In fact, the customer is more likely to like it and this will undeniably boost the volume of sales.
In the music industry, it has been observed that only 3 millions out of the 13 millions available-for-sale titles are being purchased. Besides this, 3% of the sold title represent 80% of the revenue. (6)
Moreover, another striking fact is that recommendations are usually based on social facts and so, it is always the better rated or most liked product that will be recommended. (5) Consequently, popular products being are first suggested to the customer. A s a result, it is difficult to state that recommendation tools lead to a higher diversity.
To cut a long story short, I would state that the diversity of the offer does not absolutely lead to the diversity of purchased products. In fact, recommendation systems bring forward products based on algorithms leading to the reinforcement and enhancement of hits’ sales. Therefore, I believe the tail becomes shorter due to recommendation systems. However, if e-commerce platform’s first aim was not driven by profit, the tail could become longer.
It is no big news that we’ve entered a new economical era. We’re two feet well within the digital era and this has a ton of economical consequences. This paper focuses on recommendation systems. Those may look like a simple add-on of websites and online services but they have actually deep consequences.
One aspect to keep in mind is the… Read more
It is no big news that we’ve entered a new economical era. We’re two feet well within the digital era and this has a ton of economical consequences. This paper focuses on recommendation systems. Those may look like a simple add-on of websites and online services but they have actually deep consequences.
One aspect to keep in mind is the current change of our culture. Our hit-driven culture is slowly but surely shifting toward a niche-driven culture, where the trend is to like what’s not trendy, as paradoxical as it may sound. This is of course in direct link with the Long-Tail theory. Whereas, earlier, people had access only to certain amount of big hits, the numerous online platforms gives them access to a much bigger market, allowing them to purchase more of the less popular items/articles/hits/… . This new accessibility combined with the change of culture have for a consequence that people start to buy more of the previously unknown items (the niches) and less of the big hits, making in this sense the tail longer. As studies show that recommendation systems increase the sales of the niches, it would seem like those systems are working in this direction as well. But a much more difficult question to answer is have they created or pioneered this trend ? If I was to give my opinion, I would say no. I don’t see such systems being able to influence human behaviors that much. I’d rather say that recommendation systems were set up in order to answer to this change in our culture, but that’s just my opinion.
But we can also find arguments to say that recommendation systems don’t make the tail longer but rather smaller. The current generation is composed of three big families : content-based, collaborative and hybrid approaches. Those systems still have a few limitations. I would name for example the problem of “over-specialization” for content-based recommendation. This means that systems can’t offer anything else than articles similar to those already bought/rated, preventing therefore the diversification of the user. In this way, we could say that recommendation systems are actually making the tail shorter, as they confine users in the market they already know and prevent them from going to something new (and therefore making the tail longer). As far as collaborative systems are concerned, we can name the “new item” problem. New articles are added regularly on most platforms. Collaborative systems base their recommendations on users’ ratings. Therefore, a new item won’t appear in the recommendation until a certain (often large) amount of users have rated it. In this way, those systems prevent new items from being known by only recommending the already popular items. This could also be seen as a “tail-shortening” characteristic. Of course, those problems are being addressed by the various platforms and new improvements such as multi-criteria ratings, incorporation of contextual information, etc. are being set up.
To conclude my comment I would like to give a little thought to what’s going to happen next. According to the Harvard Business Review, the tail does become longer and longer but what’s interesting is that it’s also becoming flatter. This means that even though the array of products being sold is increasing, the demand for those less popular items remains low. They also predict that there will be less blockbusters but that those will be bigger than nowadays. This could be kind of contradictory to trend of our culture shifting from the hit-driven to the niche-driven. I’m therefore wondering what the tail will look like 10 years from now ? Will we see the niche keep on growing, transforming the tail into a bulker one ? Or will the blockbusters capture an even bigger part of the market, leaving a tail extremely long but flat ?
I personally have some interrogations about recommendations and the long tail theory.
Firstly, the big theory is that recommendations are made about previous choices of the consumers. But how to be sure that big firms don’t pay the website to be in the recommendations? We see this kind of thing for the Search Engine Advertising, which are “advertisements shown on… Read more
I personally have some interrogations about recommendations and the long tail theory.
Firstly, the big theory is that recommendations are made about previous choices of the consumers. But how to be sure that big firms don’t pay the website to be in the recommendations? We see this kind of thing for the Search Engine Advertising, which are “advertisements shown on search engines pages when specific keywords are typed” (1). The process explained in the article “Do recommendation systems make the ‘tail’ longer or shorter? “ is based on this “free and magic” peace of information but there is written nowhere that this information is truthful. And thus the tail will not be taller because the biggest companies will pay to have their mainstream products in the recommendations.
Secondly, if we agree that these recommendations are really previous choices, I am not sure that it will really enlarge the tail because some choices are logical. Indeed, if I am buying the first “Game of thrones” book, I will probably look after the second one and so the choices for these kinds of saga are distorted. The same reasoning is valid for the people who buy book (or Dvd and so on) from the same author or actor. And the principle to “enlarge the tail” is here not meet because there is no surprised (2).
Thridly, some companies have difficulties to have a database large enough and thus they have to buy one (so it is not a base on previous choices) or they ask consumers to rank some products on a “top 10”(2). But I think it will not enlarge the base. Let’s take an example and before buy a movie online you have to make “a top 10 of your favorite movies”. You will do it quickly because you don’t have time to waste and you will write the first 10 movies you have in minds which are most of the time really mainstream and that have connections between them. And there is a difference between products I like or book I really bought. So the database will not be fair and the tail will not be very taller.
In conclusion, there are some interrogations but if the data base is true, based on previous choices and not distorted, we can say that the tail is enlarged.
Nowadays with the growth of online retailers with less production costs, we have the possibility to buy a wider range of products. The traditional retail was focused on hits and blockbusters because of the limited storage space.
The theory of Long Tail is the strategy of selling a large amount of different small quantities of unique products (the tail… Read more
Nowadays with the growth of online retailers with less production costs, we have the possibility to buy a wider range of products. The traditional retail was focused on hits and blockbusters because of the limited storage space.
The theory of Long Tail is the strategy of selling a large amount of different small quantities of unique products (the tail of the demand curve) additionally of selling popular products (the head). (1) & (2)
Relating to the online retailers, lots of recommendation systems have emerged. Filters could help to provide products who better reflect the customer’s needs. If the customers could easily find new products they like using recommendation system, this could lead to an increase of sales in the tail of the demand curve.
But what is exactly the effect of these filters on the general shape of the curve?
Some studies such as these of Bart Van Looy and Annelies Geerts from the KU Leuven argue that the curve becomes longer but flatter. (3) The blockbusters products are still selling but in smaller quantities than before, they are less popular.
If we are shifting from a retailing of hits to niche products, does it mean that the total demand would be bigger?
In my opinion, the customer’s satisfaction increase with filters and they will therefore be more inclined to buy more items. Besides, knowing some hits will still remain, they could lead to an increase of the sales of similar unknown products with the recommandations like it was the case with the Simpson’s book “Touching the Void”. (4)
Moreover, it has been demonstrated that consumers in the movie sector who rent obscure movies are more likely to be heaviest customers than those who focus on blockbusters. (5) The study shows that they rented 50 movies against 20 for those chosen hits products in the six-month period under review.
In conclusion, presenting a large choice of products in addition to performant recommendation systems will lead to more sales and more satisfaction from the customers.
There is no doubt that recommendation systems can and have had very successful results for companies like amazon and Netflix, there ability to increase the size of and profit from long tail has been impressive. For Netflix the recommendation system serves to increase switching costs for there service, and Amazon is able to increase their long tail by recommending similar… Read more
There is no doubt that recommendation systems can and have had very successful results for companies like amazon and Netflix, there ability to increase the size of and profit from long tail has been impressive. For Netflix the recommendation system serves to increase switching costs for there service, and Amazon is able to increase their long tail by recommending similar or linked products.
Recommendation systems are a natural result of the world that we now live in. the sheer quantity of information, media, products and services that are available online is staggering. So naturally people developed ways to reduce searching costs from the internet to bring the exact content you are searching for. These systems are becoming more and more relevant and being applied to more and more products.
Despite all the positive effects recommendation systems can have I thought it would be interesting to look at some potential negative effects from recommendation systems. Specifically, what social medias effect has been on the news and its role in creating Eco chambers (or information bubbles). Facebook news allows people to customize the news the see and while opposing views can still get through. However, people create there own Eco chambers based on the things they like and click on. A study from Facebook also found that “liberals tend to be connected to fewer friends who share conservative content than conservatives who tend to be linked to more friends who share liberal content”. And with this and confirmation bias these Eco chambers have done a lot to cause divided political views. While sites like Facebook are not specifically for news a PEW study found that 61% of millennials obtain their news from social media (Facebook and twitter). Obviously it is not Facebook’s intent to create these Eco chambers however it is a side effect of their services. It is important to note that there are many other problems with the news including sensationalism, lack of trust for media, the effects of citizen journalism, accuracy of different news sources, etc.
This effect has not gone unnoticed. The director of product at Facebook Mike Hudack had a rant on the state of the media recently with the quote “It’s hard to tell who’s to blame. But someone should fix this shit.”. While the effects of the long tail have been observed for many recommendation systems, social medias effect has been far more polarizing.
Overall I believe that effective recommendation systems for things that we want are important to sift through the vast amount of content that comes with a digital world. However, when we rely on a system for something we need like the news its easy to see how only seeing the news we want could cause some problems.
I suppose you mean ‘echo chambers’. That’s an interesting perspective, thanks.
Gauthier Seny
Nowadays, it becomes obvious that a lot of sales platforms use the recommendation system which is certainly in their interest from an economical point of view. Nevertheless, it is easier for big platforms as Amazon and Netflix to deliver niche product thanks to their “unlimited inventory” as opposed to “limited-inventory” competitors. Amazon and Netflix benefit significantly from the long tail… Read more
Nowadays, it becomes obvious that a lot of sales platforms use the recommendation system which is certainly in their interest from an economical point of view. Nevertheless, it is easier for big platforms as Amazon and Netflix to deliver niche product thanks to their “unlimited inventory” as opposed to “limited-inventory” competitors. Amazon and Netflix benefit significantly from the long tail because those niche products, influenced by the recommendation system, that are present in the tail represent an important share of their sales.
Now, to answer the question whether the recommendation systems make the tail longer or shorter I would say that they make it shorter or unchanged.
Why?
For instance, if you’re looking for a specific music genre, even though the songs are unlimited in this genre, recommendation systems will make some niche songs more popular and put them into the head of the tail. Following this logic, the head of the tail will become longer or you may also say the tail is becoming broader and shorter. In this situation, the recommendation systems will make the tail shorter.
Following an other logic, recommendation systems might have the effect of “popularizing” niche songs while in the mean time, previously “popular” songs might join the end of the tail. In this situation, there would be a replacement effect. “End of the tail” songs would replace “head of the tail” songs which would make the situation unchanged.
From my point of view, those situations would arise because there is still a limit to recommendation systems. What I mean is that there’s some stability point where the tail cannot get any longer. A song has a life cycle. Most of the popular songs stay popular for several years, then end in the tail and finally “completely disappear”. It’s the same with clothing, movies and other products. In general, I think the tail has a tendency to keep its length even if from time to time, some “shocks” might modify the length.
As an introduction, I will remind the long tail theory.
In the chart (1), we can clearly see what is the long tail: Seen by Chris Anderson, the long tail includes the products that are less popular than the head. The fact is that the popular products are sold in large quantities where the niche products are sold in low quantities… Read more
As an introduction, I will remind the long tail theory.
In the chart (1), we can clearly see what is the long tail: Seen by Chris Anderson, the long tail includes the products that are less popular than the head. The fact is that the popular products are sold in large quantities where the niche products are sold in low quantities but consist of a large variety of products. The long tail theory says that the (niche) products in the long tail outnumber those ones in the head. (1)
Indeed, the Long Tail theory asserts that “the products with a low demand, or a low sales volume can collectively make up a market share that rivals or exceeds the relatively few current bestsellers and blockbusters, but only if the store or distribution channel is large enough.”. (2)
It seems to me that this long tail can be a potential market share that Internet made available with the fall of the costs of production and distribution.
As explained in the comment, the long tail is becoming thicker (and longer) as the niche products are doing better in the digital age and become more popular. Hence, those niche products are interesting and big retailers understood it.
A little example may illustrate the long tail theory: An Amazon employee said that “We sold more books today that didn’t sell at all yesterday than we sold today of all the books that did sell yesterday.” (3)
After this remind, we can ask: How the recommendation systems impact the long tail and the niche products?
There exist more and more recommendation systems. “The top more popular in this category” but also “People who bought this also bought…”. They appear on the big retailer website and on the mobile platform such as Google Play and Apple. That can be a problem because the recommendation systems are based on the sales and rating. Hence, those systems will redirect the customer to popular products again and again. As said in this article: “Because common recommendation systems are based on sales and ratings — for example, people who bought this also bought this — they’re unable to surface truly novel items that have not been discovered by many other people.”. (4)
I think that as long as customers will use the recommendation systems based on sales and ratings, they won’t likely discover niche products, as explained in this article: “This tends to create a “rich gets richer” effect for popular items, and it might also prevent consumers from finding better product matches because of this bias for items that have been purchased by others or that have been rated well by others.”.
However, a little further in the article, they made two conclusions: the first one is that however, the recommendation systems were supposed to help the niche products to obtain opportunities, recommendations systems doesn’t necessarily help to discover niche products
The other conclusion is that when a product with low rating is recommended, people may give the benefit of the doubt and give them more attention. (4)
At first glance, it may seem clear that recommendation systems foster niche markets. Indeed, we may think that it opens the gates between people and less known products, that it increases the chances of SMEs etc… It is partly true. Undeniably, thanks to the recommendation systems, people discover new items that they did not know before, they have access to… Read more
At first glance, it may seem clear that recommendation systems foster niche markets. Indeed, we may think that it opens the gates between people and less known products, that it increases the chances of SMEs etc… It is partly true. Undeniably, thanks to the recommendation systems, people discover new items that they did not know before, they have access to more products and they can directly find products that perfectly match their tastes. For example, here is how the Google News’s recommendation system works: «The recommendation system builds profiles of user’s news interests based on user’s click behavior on the website. This system helps users to find new articles that are interesting to read” (1)
However, even if recommendation systems may help niche markets to be known, it is less obvious that those systems make the ‘tail’ longer. Effectively, it diversifies the recommendations for individuals, but if we aggregate the recommendations, the recommendation systems often drive people to the same items and thus there is no real diversity. This happens because “most of existing recommender systems struggle to recommend tail products because of the data sparsity issue”. Thus those systems only recommend popular items which creates a “rich gets richer” effect. (2) Users don’t really discover truly novel items that has not been searched or bought by many other people because common recommendation systems are based on sales and ratings. (3) To return to the example of online news reading, we have to be cautious because algorithms and big data used by recommendation systems often have bias. As a result we may miss the big picture of the news. (4)
To conclude, I think that recommendation systems are sure a great opportunity for niche markets but we must continue to improve their process, remove biases from algorithms and big data they use in order to really make the ‘tail’ longer.
(1)Liu, J., Pedersen, E., Dolan, P. (2010). 2010 International Conference on Intelligent User Interfaces.
(2)Yin, H., Cui, B., Li, J., Yao, J., Chen, C. (2012). Challenging the Long Tail Recommendation. Department of Computer Science & Key Lab of High Confidence Software Technologies (Ministry of Education), Peking University.
(3)Hosanagar, K. (2015). ‘Recommended for You’: How Well Does Personnalized Marketing Work? http://knowledge.wharton.upenn.edu/article/recommended-for-you-how-well-does-personalized-marketing-work/
(4)ibid.
Let’s start this case analysis by trying to make a clear definition of the longer tail “effect”: If all the goods from a market were normally distributed, the tail of the distribution will represent a niche market (sells unique goods in small quantities) when the center of the distribution will represent the main products (blockbuster, best-seller, …). In other words,… Read more
Let’s start this case analysis by trying to make a clear definition of the longer tail “effect”: If all the goods from a market were normally distributed, the tail of the distribution will represent a niche market (sells unique goods in small quantities) when the center of the distribution will represent the main products (blockbuster, best-seller, …). In other words, the well-known products have a more powerful sales-force than the unknown-one: logic. But the interesting thing is that with the internet revolution and indirectly, the e-commerce, companies are not physically constrained anymore: they can put in plain view all the products they want. It doesn’t sound that significant, but back in the time, a bookseller for example, has no interest to a book which has not any guarantee of success in his store. Indeed, it is most likely that a very small amount people will buy it. Especially when the last book of a famous writer just came out. As he is physically constrained (limited capacity), he is more or less obliged to only propose mainstream books. This small example allows us to realize some important effects.
In today digital world, anyone can put anything anywhere on the internet and this includes big companies like Netflix and Amazon. They transformed this new power in a marketing tool which allows them to adapt the content of a website page according to the user’s interest (recommendation system). Therefore, nowadays, they are able to offer us a larger ranged of products, including the less known one which nobody is aware of. Economically speaking, on one side, consumers will be satisfied as they discover new products which are potentially going to interest them, and on the other side, selling these products is going to be more profitable for the niche market producers. Both consumer and producer welfare is enhanced. To me, it is definitely something crucial and important. This effect is lowering the normal distribution and rebalance the situation.
Furthermore, the recommendation system (personalized marketing) has evolved with an extension to the initial concept: the online reviews. The figures speak for themselves: 90% of consumers read online review before visiting a business, 88% of consumers trust online reviews as much as personal recommendations. Here again, we can see the strength of a personal recommendation system as a personalized marketing tool.
To conclude this discussion about the recommendation system, we can see that internet and the globalisation are making the long tail taller, increasing the numbers of available products on the market. Personally I think that this will progress again and again, and that marketers will never stop to innovate, in a good way I hope. This is a very interesting subject which is complex and has more impact than just a variation in quantity/price. I mean, we can question ourselves about the respect of privacy of companies using this tracking techniques. Most people using internet don’t even know that their data are collected. Anyway, it has a significant positive impact on consumers/supplier. In my opinion, I think this is a step forward for a better balanced economy where all economic players have more or less the same success chance. Ultimately, this further strengthens the competition instead that this time, it’s not just a harder fight between the main actors of the market…
References:
Su, N., Levina, N., & Ross, J. W. (2016). The long-tail strategy of IT outsourcing. MIT Sloan Management Review, 57(2), 81-89.
Firstly, I will start by giving you some explanations about the Long Tail phenomenon and illustrate them through the example of Amazon. Then, I will talk to you about the recommendation systems and their link with the Long Tail theory.
Anderson’s Long Tail theory is illustrated by a decreasing curve on a graphic: popular brands/services/products on the vertical axis (called… Read more
Firstly, I will start by giving you some explanations about the Long Tail phenomenon and illustrate them through the example of Amazon. Then, I will talk to you about the recommendation systems and their link with the Long Tail theory.
Anderson’s Long Tail theory is illustrated by a decreasing curve on a graphic: popular brands/services/products on the vertical axis (called the “Head”) are distributed through traditional retailers. On the horizontal axis (called the Long Tail), an infinite array of smaller and niche brands are spread and distributed through the Internet. By using the example of Amazon, Anderson explains that more and more firms have understood that Pareto Principle – which states that 80% of the effects come from 20% of the efforts – is not always the best choice. Effectively, if everyone focused only on the 20% of the market representing 80% of the benefits, competition would increase and profit per market share become limited. On the contrary, focusing on the remaining 80% of the market which are neglected could be very lucrative if you can entirely incorporate it. It is sometimes more beneficial to have 20% of the market that to share 80% with the rest of the world.
This was what Amazon understood by offering products called “normal” (represented by the 80% of earnings) and on which they met strong competition, as well as products called “niche” (very specific and difficult products to find, representing 20% of profit from the market). Some researchers have confirmed that a huge quantity of sales come from unpopular books that were not available in brick-and-mortar stores. This leads us to believe that if Amazon only focused on generic products, it would not be as successful as it is now, simply because the competition between these products is too tough to differentiate efficiently. Therefore, we observe that consumers’ primary value of the Internet “comes from releasing new sources of value by providing access to products in the long tail”.
Actually, to determine whether a sales distribution has a long tail or not is the cost of inventory storage and distribution. Before Internet, it was difficult to reach niche markets: since storage and distribution costs are high, only the bestseller products were sold and competed for physical shelf space in physical retailers such as Walmart or Target. But nowadays, with the emergence of the e-commerce, it becomes economically viable to sell unpopular products. Effectively, by centralizing their warehouses, online providers such as Amazon or Netflix see their storage costs significantly decrease. Moreover, whether it is a popular or an unpopular movie, distribution costs remain the same. That way, it allows e-companies to stock a wider range of movies compared to traditional physical retailers. And effects are felt: Netflix for instance has observed that in aggregate, unpopular movies are rented more than the popular ones.
According to MIT Sloan Management Review article titled “From Niches to Riches: Anatomy of the Long Tail”, tools such as recommendation systems are allowing customers to find products outside their geographic area without being confined only to local markets. No perimeter on market demands exists anymore since Internet provides an unlimited selection of products. On the supply side, the long tail has possible effects for companies’ culture and politics. As a matter of fact, it gives them opportunities to introduce products in the niche category and encourage the diversification and innovation of new products.
Economists in an article from “The Economist” (2015) stated that two benefits of e-business were provided to consumers: prices would decrease and become uniform and the selection available to consumers would increase. But the latter statement seems not to be correct. For instance, a rare book is more valued by consumers who have troubles to find it in physical stores. “By posting its inventory online, all those readers scattered around the world are now added to the potential demand for the store’s copy. Higher demand translates into higher prices which clearly makes the bookseller better off. But so is the reader since without the Internet he would not have found the book.” It seems obvious that some customers will be willing to pay a lot if they finally find on the Internet a rare product they have researched for a long time. “Higher prices are therefore a sign that buyers are being better matched to books they want. […] On the contrary, benefits are less likely to hold for easy-to-find, commoditised products; online prices of popular, usually in-print, books were less dispersed and closer to offline prices.” Nevertheless, the success of this Long Tail approach shows that you need both ends of the curve. Effectively, bestsellers still matter in attracting consumers in the first place.
My opinion about recommendation systems (as many other tools such as search engines, sampling tools etc.) is that they play an important role in the growth of the niche markets. We can quote an extract coming from The theory of the Long Tail by Chris Anderson which illustrate quite well those systems: “For instance, the front screen of Rhapsody features Britney Spears, unsurprisingly. Next to the listings of her work is a box of “similar artists.” Among them is Pink. If you click on that and are pleased with what you hear, you may do the same for Pink’s similar artists, which include No Doubt. And on No Doubt’s page, the list includes a few “followers” and “influencers”, the last of which includes the Selecter, a 1980s ska band from Coventry, England. In three clicks, Rhapsody may have enticed a Britney Spears fan to try an album that can hardly be found in a record store”. Amazon and Netflix also do the same thing through other recommendation techniques (“Customers who bought this also bought …”). But with all the same goal: drive demand down the Long Tail.
Recommendation systems could be therefore very efficient for small businesses with a small budget that try to innovate, promote and launch new things.
Introduction
Recommendation systems « identify recommendations autonomously for individual users based on past purchases and searches, and on other users' behavior ». (1) These systems have become more and more common in recent years and I often face with these one while I’m visiting a panel of different applications such like music on Spotify, books on Amazon or movies on Netflix.
Toward… Read more
Introduction
Recommendation systems « identify recommendations autonomously for individual users based on past purchases and searches, and on other users’ behavior ». (1) These systems have become more and more common in recent years and I often face with these one while I’m visiting a panel of different applications such like music on Spotify, books on Amazon or movies on Netflix.
Toward the realization of this comment, I have focused myself on the music application called Spotify I’m using everyday in order to understand the evolution of the recommendation system and how this technique of recommendations could make the tail longer or shorter.
Collaborative filtering system
Spotify recommender system was mainly based on collaborative filtering techniques. These approaches are defined as follows « collaborative filtering is the process of filtering for information or patterns using techniques involving collaboration among multiple agents, viewpoints, data sources, etc. » (2) In other words, « the idea of collaborative filtering is to determine the users’ preferences from historical usage data » (3). For example if two users listen to largely the same set of songs, their tastes are probably similar.
Limits of the system
But this system seems to have certain limits. In fact, two main issues called « content-agnostic » problem and the « cold-start » problem were encountered by Spotify. The first one is met because the technique of collaborative filtering don’t use any kind of information about the items that are being recommended. Given this, « recommendation systems become predictable and boring in the sense that popular items will be much easier to recommend than unpopular items, as there is more usage data available for them ». (4) The second one, called « cold-start » problem, is encountered when songs are new or unpopular. In fact, in this case, there is no data to analyze and then the collaborative filtering system breaks down : recommendations don’t work for these songs. A last problem from the collaborative filtering system, which can be mentioned here, is the heterogeneity of content with similar usage patterns. In fact, when I’m listening a playlist or the entire album from an artist, these one may contain intro tracks, cover songs and remixes. The collaborative filtering algorithms are only coded to take the original song and won’t take all other forms of songs to draw recommendations.
The solution called « Deep Learning »
In order to solve these different issues, Spotify has had to take different actions to incorporate other sources of information into its recommendation system. Spotify found another approach called « the content-based recommendation ». In this system, the firm is taking other sources of information as tags, lyrics, text mined from the web (reviews, interviews,…) or the audio signal itself for example.
It’s precisely this last source of information (the audio signal) that Spotify began to analyze in order to predict what songs users will enjoy. The algorithm created by Spotify is able now to learn the specific content of the songs following different factors as specific pitches, chords, sounds, and others. With this, filters are set up to cluster songs in different categories following these last factors. For instance, if a new song comes on the platform or if this song comes from an artist that’s not popular, it would be hard to recommend that to users if Spotify is using more traditional methods because no one else is listening this song. But with this new method, the algorithm is relying on the acoustics of the song (for example), it can make better recommendations (lesser-known artists i.e.) for users.
Conclusion
As a conclusion, people in the past were discovering products which were already rather popular in the whole population. However, Spotify managed to overcome and resolve this problem. In fact, the firm created a new algorithm which allows them to recommend new and unpopular musics what wasn’t possible with collaborative filtering systems.
In general and based on the Spotify case, we can see that firms are in the process of finding systems which are able to recommend new products or unknown products to customers. In the past, one would have thought that recommendations systems were making the tail shorter. But in reality and thanks to new technologies, we see that the tail is becoming more and more longer…
Thanks! Your comment should contribute to a ‘deep learning’ of the topic.
Verschoore Quentin
According to my research and in order to submit a comment relevant as possible, I decided to divide the content of my analysis into 3 parts, as follows:
1. Definition of the concept of “recommendation system” and areas of application:
Commonly used on e-commerce, social media and content-based websites, this information filtering technology analyses available data about the registered user’s profile in… Read more
According to my research and in order to submit a comment relevant as possible, I decided to divide the content of my analysis into 3 parts, as follows:
1. Definition of the concept of “recommendation system” and areas of application:
Commonly used on e-commerce, social media and content-based websites, this information filtering technology analyses available data about the registered user’s profile in order to provide him suggestions in which he might be interested.
Initially, this recommender system was used by Amazon, over more than 10 years ago, when the core business of the company was essentially based on online bookstore. Their idea was revolutionary and intended to suggest books to customers, using the gathered data about their previous activity. But since then, even if this technology remains mainly used by online information goods retailers (books for Amazon, songs for Spotify, movies for Netflix, ect.), its scope has extended to diversified fields.
Indeed with his “jobs you might be interested in” section, “Linkedin” applies the available data about its members, such as background, locations, skills to adapted algorithms, in order to come up with suitable job suggestions. Furthermore, this technology even extends to the medical field, with for instance, the “knowledge-based dietary nutrition recommendations”.
2. From an analytical point of view:
Based on cutting edge algorithms, those recommender systems use a number of different technologies, which can be classified into 2 broad groups:
– Content-based systems also referred to as cognitive filtering, recommends items based on a comparison between the content of the items and a user profile. For instance, if a Netflix user has watched many cowboy movies, then recommend a movie classified in the database as having the “cowboy” genre
– Collaborative filtering systems recommend items based on similarity measures between users and/or items.
In fact, from a conceptual standpoint, those algorithms intend to assess and the unknown and not available rating value (representing the user’s preference for the item) with respect to a specific item the consumer hasn’t tried/doesn’t possess yet. This way, starting from the user’s profile or from similarities, a high predicted value would mean that the user could be really interested in the product in question.
3. Economic impact:
In order to introduce this part of my analysis, I wanted to highlight the fact that those recommendation systems were clearly profitable for online information goods retailers. The “Netflix challenge” which occurred a couple of years ago is a good example of this “profitability”. Indeed, in view of the great opportunity it represented, the company offered a prize of $1,000,000 to the first person or team implementing an algorithm beating their own recommendation one of 10%.
This above-mentioned opportunity could be explained by the “Long Tail theory”.
Initially encouraged by the emergence of the e-commerce, this theory is based on the fact that traditional physical delivery systems are limited by the resources they have at hand. For instance, physical bookstores, restricted by their shelf space, can show only a small fraction of all the choices that exists, while Amazon proposes millions of them. Therefore, it allow such online information goods retailers to sell small quantities of special goods, located in niche markets and that would not be available in the shelves of traditional physical retailers. Those aggregated quantities represent the so-called “Long Tail” on the graph on which items are ordered on the horizontal axis, according to their popularity (most popular items on the left side and niches market on the right side of the decreasing curve). This way physical retailers offer only the most popular items while on-line institutions can provide the entire range of items (the tail as well as the popular items).
Considering the size of the “Long Tail”, composed by the aggregated niche markets, and the size of market dedicated to popular products, a remaining question concerns the generated impact of such a system on sales and profit. Does it simply increase the benefits through adding a share of special products to the amount of popular items sold, or does it reallocate demands from those popular items to the niche markets?
In my opinion, even if a share of the items previously bought will switch from the popular side to more specific products, this technology still increases the number of product sold. This way it remains profitable for online retailers through increasing the accessibility to specific items and therefore through inciting users to buy more.
Personally when I go on the internet looking for clothes, shoes, books, any item I need, I appreciate the recommendation system of some websites. It may lead you to discover new products, to which you would never have thought, related to the product you are looking for. For example when you look for clothes, it can give you some ideas… Read more
Personally when I go on the internet looking for clothes, shoes, books, any item I need, I appreciate the recommendation system of some websites. It may lead you to discover new products, to which you would never have thought, related to the product you are looking for. For example when you look for clothes, it can give you some ideas of others clothes you can wear with the piece of cloth you want to buy. And this make sure customers’ search costs will decrease as people will gain time while looking for the right piece they want. (1)
Recommendation leads to increase the purchases of niche products and so supports the theory of the long tail but recommendation systems cannot be only based on previous sales otherwise it will only highlight popular sales as limited sales won’t appear in the recommendations. And it will lead to a “rich gets richer” effect. (2)
Customers pay also attention to reviews of previous buyers. It helps increase the conversion rate of users. It has been shown that 63% of customers are more likely to make a purchase if there are reviews. (3) This will thus boost the sales and make the tail longer.
But this system remains really subjective as, like for reputation systems, it is based on other customers’ previous sales or on the type of product you already looked at. If you liked one product, you won’t especially like another product recommended by the website just because someone else also looked at it or bought it.
Although I find recommendation systems a great adviser, I don’t like remarketing which is based on your personal website visits or sales. I can be annoying being remarketed as, as soon as you leave a specific website, you will get an ad for the product you just looked for on every website you will visit afterwards. It is a great marketing tool as it was already shown that it can increase sales, but for the customer it is disturbing. Therefore, it needs to be controlled with publication rates. The brand can’t post an ad on every website you visit every day. It needs to be spread over time. Otherwise the customer will get angry about it.
Into the world of big data, recommendations systems are without disputes huge useful tools for the e-commerce in every industries, whatsoever in goods industry, like for music or clothes but also in services industry. These systems are so many effects on the customers and the e-tailers.
At the point where the customers prefer web sites with recommendations systems to make… Read more
Into the world of big data, recommendations systems are without disputes huge useful tools for the e-commerce in every industries, whatsoever in goods industry, like for music or clothes but also in services industry. These systems are so many effects on the customers and the e-tailers.
At the point where the customers prefer web sites with recommendations systems to make their shopping (1). Hosanagar K. and Lee D. (2014) say that with the long tail phenomenon the web site offer a large range of items and the recommendation systems help the customers by reducing the choice, with items based on their preferences. By using recommendation systems, the customers can be sure that their search costs will be reducing, they will find faster products that they want and like. This is on line with the personalization of the web and the need for speed of the society (3).
Moreover, by allowing to discover items towards which we didn’t go usually and by using filters more personalized, the recommendation systems allow to developed the consumption because people buy more than without these systems. They allow also to aim more the consumer segments. In fact, with products proposed and recommended, it attracts more customers with needs and cultures different (2).
For the e-tailers (the biggest who use it are Amazon and Netflix), recommenders are useful in the management of huge inventory present in the supply-side. Popular items ask an important part of the space of stocking to meet demand and the less popular items are in competition for the limited space remaining (2).
There are also others positive impacts: on the sales volumes and the revenue. Hosanagar K. and Lee D. (2014) say that “according a recent study, the recommendation engines were reported to increase revenue up to 300 percent, conversion rates up to 150 percent, and consumers’ average order value up to 50 percent” (1).
In addition to have impacts on the sales of the company, recommenders help also to provide a better image of the brand, offer a feeling of trust at the customers (3). This enhances also the online experience provides to customers.
The long tail theory states that the volume of sales of few bestselling products are comparable to the volume of the many products left sold in small quantities (often niche products). The tail symbolize then these leftover products sold and available in reduced quantities.
Due to the recommendation system in the digital world, consumers are easily set in relation with very… Read more
The long tail theory states that the volume of sales of few bestselling products are comparable to the volume of the many products left sold in small quantities (often niche products). The tail symbolize then these leftover products sold and available in reduced quantities.
Due to the recommendation system in the digital world, consumers are easily set in relation with very specialized products which fulfill their needs with more precision. The key question is thus to know if this system makes the tail longer or shorter than before.
My first thought is that the recommendation system makes this tail shorter. On internet, there is (almost) no capacity restrictions anymore. Therefore, much more products are available and this will make the tail eventually longer. But on the other hand, due the more global availability of the goods, the tail is also thicker as more people will buy them. Globally, the tail will become stronger than before, with more volume traded through internet.
With the recommendation systems, people will have more choice than before, according to their needs. They will thus be more likely to buy from this recommendations. This creates a certain loop, as the recommendation system becomes “smarter” by analyzing the buyers’ behaviors.
The recommendation system will present to the next customers more accurate data fulfilling their needs and the loop will grow bigger and bigger. Eventually, what was categorized as “niche products” will become more popular and become bestseller with the others “older” bestselling products.
This will make the tail smaller and thicker, and the head of the sales, the bestsellers, will also become thicker and a bit longer, according to me. The wholes sales will then become more even, with less volume differentiation between the head and the tail.
On the whole, I believe it is important to consider the question of whether internet recommendation systems is making the media content tails longer or shorter from multiple angles and under the lens of multiple different media types. On the whole I believe it is undeniable that complex algorithm based recommendation systems certainly do expose users to different media sources… Read more
On the whole, I believe it is important to consider the question of whether internet recommendation systems is making the media content tails longer or shorter from multiple angles and under the lens of multiple different media types. On the whole I believe it is undeniable that complex algorithm based recommendation systems certainly do expose users to different media sources they might not otherwise be aware of, however, the design of these systems can vary greatly in their effectiveness in lengthening out the tail.
Within music, it seems that the most relevant music delivery services (Music Streaming) are noticing the consumer demand for lengthening the tails of the content they can offer and are reorganizing their recommendation systems to accommodate this. The most prominent firm utilizing this is Spotify. This service, dubbed “Fresh Finds” seeks to complement Spotify’s current popularity based recommendation system (Shortens the tails), discover weekly. The key difference with Fresh Finds intends to populate its suggested playlists by focusing on lesser known artists by scouring 50,000 blogs, critics and other taste influences. This should have the effect of greatly widening the curve for many artists as the selection is going to be based off of little known artists that don’t have the large marketing budgets that a major record label could provide. This also implies that Spotify will need to work with the artists that this newly designed discovery system uncovers to upload their work to the platform, allowing the work to be consumed by new, much larger audiences and indeed monetized. [1]
With that being said, there is no credible evidence to suggest that the proliferation of online music content delivery systems has not helped out the lesser known or wider tail arists, in fact, quite the opposite. According to The Economist, Independent Record Labels (“Indie”)’s share of the Bilboard top 200 albums has grown to 35% in 2010 from 13% in 2001. This lends credence to the popularly romanticized ideal of the internet democratizing the creative media space.[2]
Another key content viewing platform, Youtube, has been at the center of a lot of controversy over whether the algorithms that determine which videos get selected to be “Reccomended” or on the front page is unfair to the smaller or lesser known content creators. The current system uses a time spent viewing weighted approach to identifying the best, most relevant content. [3] this problem has led to popular, early adopter youtube content creator Philip Defranco to state that he does not believe it would be possible for him to recreate the sucess of his channels in the current environment as it is much, much harder for great content to get noticed under the current system.
To know if the Anderson’s theory of the long tail is well reflecting the facts, I would suggest we have an “a posteriori” look at the reality of business today.
The theory implies that the consumers, thanks to the internet and the recommendation systems it offers, will find a new way of consuming: retailers should now offer a larger number… Read more
To know if the Anderson’s theory of the long tail is well reflecting the facts, I would suggest we have an “a posteriori” look at the reality of business today.
The theory implies that the consumers, thanks to the internet and the recommendation systems it offers, will find a new way of consuming: retailers should now offer a larger number of less mainstream products (and focus more on the tail of the demand curve, which should become longer and thicker) where they used to offer a few “blockbuster” products (they used to focus on the left/head part of the demand function). However, is it really the case today ?
Of course, the recommendation systems could give the chance to smaller production as the internet provides indeed an unlimited space on the “shelves” to offer less mainstream products. Let’s take the example of Amazon VS library : for a traditional library, their total number of shelves can be seen as a rent cost so, to maximize their profits, they have to offer as many best-sellers as they can to sell as much as possible. On the other side, Amazon does not have this problem of room size and they can offer thousands of books, CD’s,… that a traditional library could never hold in one room.
Besides, the rising of the internet and recommendation systems can also give the chance to good products to be known even if they have not enough budget to make an advertising campaign as the internet can be seen as a free showcase.
These are the arguments provided by the long tail theory advocates and by companies (like Netflix or Amazon) coming from this “long tail era”.
However, I personally share the opinion of Harvard Professor Anita Elberse, which is very well explained in the articles [1][2] linked below. I think that many benefits of the recommendation systems are parts of an utopia.
As showed in these articles, we are indeed more than ever in a blockbusters era where taking risks by investing a lot of money in one project is often the best way to success and where being wise by spreading the investment is counter-intuitively synonym of risk taking. As an example of this blockbusters’ supremacy, the articles compare two different strategies of these last years : Warner Bros VS NBC (TV channel).
The Warner Bros philosophy is a legacy of the strategy which was first used by the studio producing “Jaws” in 1976 : money spending by millions in advertising and trying to have as many movie theater partners as possible at the beginning to avoid word-of-mouth, so, “take the money and run”, even if the movie is bad. This strategy is obviously successful as we can see with the huge success of some of the giant productions of these last years (Harry Potter saga, Batman, Ocean Eleven,…).
On the other side, we have the NBC vision. A few years ago, they decided to use the opposite strategy by avoiding risks and produce series with not mainstream actors, trying to spend less money on ONE production to be able to produce other series and multiply chances of success. Despite a few successful productions, the decline of audience of the channel can be seen as a proof that this strategy was not the winning one.
A few other examples are given by A. Elberse such as having stars in a sport team (to sell more tickets, whatever the result of the match) or the Apple products (which can be seen as stars or blockbusters).
So, today we can say that we are still in a blockbusters’ dictatorship era. It is obviously the “winner takes all” marketing strategy which is working and companies that used to be heirs of the long tail philosophy (and used to be very proud of that status!) such as Netflix have now become producers to follow the laws of business. They have to promote their series (e.g. House of cars) with mainstream actors (Kevin Spacey) and by using their recommendation systems to present their own productions.
Finally, to conclude, I would like to take another example, coming from my personal experience. As a user of YouTube, I have never seen an independent video with only a few hundred views in the “Recommended for you” section. The algorithm clearly focuses on VEVO channels (which are partially YouTube/Google produced channels [3]) and mainstream videos with already millions of views.
As a consumer, I think that these mechanisms are a brake to inventiveness and diversity as they force studios and artists to be part of a recurrent successful model in the name of business.
The long tail theory focuses on the market of niche products: low volume sales of a large number of product . Thanks to the internet channel, the theory argues that it is possible to target this market more efficiently, so as to compete with the current sales of few blockbuster and even to sell more of these niche products… Read more
The long tail theory focuses on the market of niche products: low volume sales of a large number of product . Thanks to the internet channel, the theory argues that it is possible to target this market more efficiently, so as to compete with the current sales of few blockbuster and even to sell more of these niche products (Anderson 2015).
The “hit-driven business model” has been challenged by the increased competition between firms. Indeed, the internet offers the possibility for online stores to offer such a large range of products that it becomes nearly impossible for consumers to chose among them, and more importantly to know about every product. Moreover, the rise of internet has changes the consumer behavior: consumers want a dialogue rather than just an overload of information and want to build a trust-relationship with firms. “Engagement” is key for e-marketing. Firms have then to develop strategies to get the consumer’s attention.
In the growing e-commerce market, one of these strategies is the online recommendation system. “We are leaving the age of information and entering the age of recommendation” said Chris Anderson. Recommendation systems have not only a direct impact on sales (by providing a review and so an incentive to buy or not), but more importantly an indirect impact on it, thanks to an increase un consumer satisfaction. The system responds to what the current consumer wants, by increasing brand loyalty and consumer engagement trough personalized offers which build this trust-relationship mentioned before. A personalized service is possible because people that are leaving reviews often use key words that other consumer will use too when searching for the same kind of product.
A recent study (2015) proves that in the pre-purchase phase, 90% of consumers read reviews,
88% of consumers trust online reviews as much as personal recommendations, 72% say that positive reviews make them trust the business more, 31% spend a larger amount on good reviewed sites and 86% will hesitate to purchase from bad reviewed businesses.
While recommendations clearly are important to consumers, a study of the University of Wharton (December 2015) shows empirically that “because common recommendation systems are based on sales and ratings — for example, people who bought this also bought this — they’re unable to surface truly novel items that have not been discovered by many other people. This tends to create a “rich gets richer” effect for popular items, and it might also prevent consumers from finding better product matches because of this bias for items that have been purchased by others or that have been rated well by others. ». This is because even if recommendations let us find new products, it does with all consumers so that the “new” products are the same for everyone. So the total effect isn’t an increase in diversity of purchases, but an expansion of best-seller purchases. So, recommendations partly “benefit the long tail” but don’t permit us to discover real novel niche products. This analysis confirms the argument of Belleflamme and Peitz.
Another study (University of Cambridge, November 2015) « estimates the effect of product saliency, affected via recommendation sets, on user choice in online markets. We find a sharp and robust 6% increase in the sales of a product when it is recommended by a highly popular new product. This effect is however short-lived, lasting for approximately only four days. On average the daily increase in sales attributable to saliency is around 5%. We also find that products recommended in smaller sets experience larger effects of saliency as they have to compete less for user attention. » « We show that such recommendations can temporarily affect the sales and appeal for products by shaping and expanding consumers’ consideration sets. »
Finally, a third study (2016) goes further and argues that recommendation systems (word of mouth) and the concentration of sales depend on consumer preferences. « When consumers with more prevalent preferences are the main beneficiaries of these improvements, the concentration of sales increases, generating a “superstar” effect. This increases the performance of bestselling products. When consumers with less prevalent preferences, such as niche items, are the main beneficiaries, the concentration of sales is reduced, generating a long tail effect. »
These three studies highlight that various aspects have to be taken into account to formulate a complete analysis. However, some statements can be done. As shown by the figures, there is a positive link between recommendations and sales, trough both direct and indirect effects. But recommendation only partly increase sales in the long tail. Niche products, as defined by Anderson, still have to find their way to the consumers.
First of all, I will give a clear definition of what the theory of the long tail involves. The theory assumes that our society and its customers are changing their buying habits. Until now the customers were mainly focused on a smaller number of mainstream products that can be considered as “hits”. Those products are situated in the head of… Read more
First of all, I will give a clear definition of what the theory of the long tail involves. The theory assumes that our society and its customers are changing their buying habits. Until now the customers were mainly focused on a smaller number of mainstream products that can be considered as “hits”. Those products are situated in the head of the demand curve, the familiar market. This theory tells us that the customers are changing their habits towards a big number of products that are then situated in the tail of the curve, the niche market.
The theory is explained by the fact that overall costs decreased highly because of the new possibilities that is offered by the internet. Indeed, the internet gives access to all of the information and at any time. One simple example for this are the songs which are online which are not played on the radio.
“When consumers are offered infinite choice, the true shape of demand is revealed. And it turns out to be less hit-centric than we thought. People gravitate towards niches because they satisfy narrow interests better, and in one aspect of our life or another we all have some narrow interest (whether we think of it that way or not).” (Anderson, 2015)
In addition, we can ask ourselves if this theory is affected by the recommendations options that several platforms have put in place. The answer to this question is positive. Chris Anderson gave two conditions in order to make the niche products exploitable. The first condition is that everything has been made available and the second one is that the customers are helped to find them. The first condition is satisfied thanks to the internet’s characteristics. Those characteristics eliminated the overall access that can limit the access. The second one can be facilitated with the recommendation options. However it is important that those functions really fulfill this condition. In other words, the recommendation should not only put forward the hits products but also the niche products. Indeed, we can see that the current recommendation options work on a popularity basis and the customer’s own interest. The popularity basis could potentially erase the positive effect of the internet on the niche markets. Since we return to a system where the most classic product (the one that interest the majority of the people) has a better visibility.
Let’s explain the steps in recommandation systems.
First, you have to collect data and filter them in order to have an idea of what the customers think and feel. You have two choices for that. First, explicit data gathering and active filter is the fact that the user indicates his interests to the system clearly. It’s for example a rating scale… Read more
Let’s explain the steps in recommandation systems.
First, you have to collect data and filter them in order to have an idea of what the customers think and feel. You have two choices for that. First, explicit data gathering and active filter is the fact that the user indicates his interests to the system clearly. It’s for example a rating scale (1 to 5 stars), liking, comments. (1)
Second, the implicit data gathering and passive filter. This is the fact of looking at the customer’s behaviour without asking him clearly. For example, it’s the fact of getting the list of products that the customer bought, the music he has been listening to, etc.Both these methods have their advantages and disadvantages.
According to the website of the Carleton College : « In passive filtering, every user will be given the same predictions for a particular item. In active filtering, the system takes into account your specific history in order to make a recommendation. To put this distinction in a solid example, the news site Digg.com uses passive filtering, showing all users the articles which have received the most votes, whereas the online sales site Amazon.com uses active filtering, trying to recommend products based upon a user’s specific actions. » (2)
But I don’t think Amazon uses only active filtering, as they push informations about what other customers bought. (3)
According to Cassidy (2006), «Elaborate “filters,” such as search engines, blogs, and online reviews, which help to match supply and demand» (4) are indeed ways to make the long tail possible. But Fleder and Hosanagar (2009) said: « Recommenders can push each person to new products, but they often push users toward the same products ».(5)
Personnaly, it depends on whether the system is passive or active. I often buy make up online and seeing the opinions of different users is helpful. But I can’t help but to think that some of the consumers might be people within the company, trying to make us buy their product. That’s why I’m always paranoid about active filtering, especially when there is something in return for the user that rate, share, post of pictures of the products, etc.
My opinion is that the « length » of the tail depends on the filtering. To sum up very roughly in a sentence, I think active filtering will lend to a short tail as it will make customer aware of popular products, while passive filtering allows a long tail as it allow the customers to discover new products.
As Chris Anderson stated “the theory of the Long Tail is that our culture and economy is increasingly shifting away from a focus on a relatively small number of "hits" (mainstream products and markets) at the head of the demand curve and toward a huge number of niches in the tail”. Firstly, I would like to evaluate the relevance of… Read more
As Chris Anderson stated “the theory of the Long Tail is that our culture and economy is increasingly shifting away from a focus on a relatively small number of “hits” (mainstream products and markets) at the head of the demand curve and toward a huge number of niches in the tail”. Firstly, I would like to evaluate the relevance of recommendation systems in people’s daily life because Anderson seems to assume that his theory applies in any case. Secondly, I will try to expand on the possible market niches a consumer might discover thanks to the theory of the long tail. Finally, I will discuss the threat of false reviews that could represent a real threat to companies using the long tail in their business strategy.
My first reaction after reading this article was to look back and to estimate my use of online recommendations. I have to admit that I do not read often the recommendations’ part when I’m thinking about buying something. However, it depends heavily on the amount I’m willing to spend on the product. In a second time, I realized that most of the time when I read comments of others on a movie it is after watching myself the movie to compare impressions and see if I missed something. Last week’s class analysed the impact of ads on people and advertisers. For my part, I hate ads and downloaded as quickly as possible adblock and still avoid websites using cookies because they upset me. This hate might result from the significant flow of ads users encounter by surfing on the Internet. Likewise, people might dislike recommendations because they are also overwhelming.
Now let’s move on to the possible market niches consumers might discover thanks to the recommendation systems of popular products. In this part, I would like to make a link with the class about net neutrality because in my opinion, Google and even YouTube are mostly not proposing the perfect match. However, we might suppose that they could because of the data and information they gather on consumers’ preferences. Nonetheless, we learned from previous scandals that Google chooses which companies are suggested first when a consumer searches for a specific product, depending on how much the firm is willing to pay to ensure its visibility online. Similarly, YouTube could propose music tailored to the consumers’ preferences but keeps to suggest the same videos with the most viewers. Therefore, I really wonder about this better access to market niches.
Finally, many companies on the Internet depend heavily on the theory of the long tail and some even build all their business around this specific part as for instance AirBnB. Firms using this kind of business model need to use it very carefully. Some of the comments from the last year already highlighted the problem of false reviews i.e. firms putting glowing comments about their service/products online. Some firms are even going further by paying some kid in Bengladesh to let a false review to make themselves known and stand out of the crowd, what makes them liable for prosecution in the US and in Europe (under consumer protection regulations). According to me, this kind of malpractice will just push users to disregard the possible comments and just generate bad advertisement to the concerned companies or individuals.
As far as I’m concerned, I prefer the last trend I observed in ‘real’ libraries (that might be existing for a long time) i.e. people letting post-its on the books with their impressions, feedbacks and books tackling the same subject than to learn how to differentiate the right matches from the pseudo suggested ones.
An interesting recent development in the area of "recommender systems" is most probably the emergence of "personal assistants" such as Siri, Google Now, Cartona, S-Voice, etc. These softwares allow users to perform different tasks through a natural language user interface. Users can ask these assistants questions, get recommendations and make research on the internet.This sofwares even "proactively deliver information… Read more
An interesting recent development in the area of “recommender systems” is most probably the emergence of “personal assistants” such as Siri, Google Now, Cartona, S-Voice, etc. These softwares allow users to perform different tasks through a natural language user interface. Users can ask these assistants questions, get recommendations and make research on the internet.This sofwares even “proactively deliver information to the user that it predicts they will want, based on their search habits”.
For example Google Now can help you find a restaurant thanks to its recommendation system, then guide you there using Google Maps, taking into account the current evolution of the traffic on the roads, or providing you with the local subway schedules. It keeps you posted about eveything you might want to know, and sometimes even without you asking for it.
Personal assitants are developed as “machine learning” tools, which basically means the more they are used, the better they get. They base their answers and recommendations on “user input, location awareness, and the ability to access information from a variety of online sources (such as weather or traffic conditions, news, stock prices, user schedules, retail prices, etc.).”
What is very interesting with machine learning is that these tools try to emulate the human reasoning, and are developed in such a way they learn to think as the human brain. This means they dot not have “rigid” behaviours, and can actually produce recommendations which could easily be given by a human being. Machine learning is one of the many evolution of the Big Data revolution, which is currently changing the way many businessess are done.
Machine learning is still relatively young, but you can safely expect it to completely change the recommendation system business, as it would allow for more personnal, more individualised and more advanced results. Nowadays, recommendation system are based either on a collaborative filtering approach, either on a content filtering approach. This means that we either use user’s history and compare it with people who have similar tastes to make recommendations, or that we use similarity in contents to make these recommendations. With machine learning, we could one day be able to ak our smartphones for music recommendations just like if we were asking a friend. It would open the door to a lot of “obscure contents” which we couldn’t find on our own, but that the “artificial intelligence” of the personnal asssitant could find, and it would probably be very relevant.
Maybe one day we will all have a friend who knows everything about music and movies and that could make better suggestions than any other friend. And that friend would be in our pocket.
The long tail effect has become popular with the emergence of e-commerce and the birth of websites such as eBay and Amazon. Long tail represents a retailing strategy that sells a number of unique goods in a small quantity, in addition of the mass sale of some blockbusters. The large number of special goods that are located on a niche… Read more
The long tail effect has become popular with the emergence of e-commerce and the birth of websites such as eBay and Amazon. Long tail represents a retailing strategy that sells a number of unique goods in a small quantity, in addition of the mass sale of some blockbusters. The large number of special goods that are located on a niche market can be represented by the long tail on a graph. In other words, if you sum the size all the small niche markets and compare them to the size of the blockbuster markets, there is a possible competition between both. So, does the Long Tail grow the pie or does it simply reallocate demands from blockbusters to the niche market? According to Anderson’s blog, the first effect of Long Tail will be that our choices will be reallocated to more specific products. In the future, if we are more satisfied with what we have purchased, we could be tempted to buy more. An article from the Economist confirms this statement. When people finally find what they are looking for after a long time, they sometimes are willing to pay a strong price in order to become finally the processor of the so much desire good.
The long tail effect is possible because of an infinite available shelf space on the internet. Indeed, sellers don’t need to restrict them-self anymore at the physical space available in their shop, where the incentive was to present on the shelves only the products that sold the best. Goods no longer have to compete for physical space on shelves. Recommendation systems play a crucial role, highlighting to customers goods to which they maybe even didn’t dare to dream about.
My feeling is that niche markets get a chance to be known thanks to recommendation systems. While before, many worthwhile products were simply outlawed from the shelves, many products have now more chance to be known and also to be purchased. In fact, I am wondering about the effect that it must have on the market. It must certainly be positive, but in which extend? Do many SME emerge or have more chance to last thanks to the new opportunity their products have of being known on the web? Even more, has the Long Tail effect a positive influence on innovation? According to an article of the Economist, the rate of introduction of new products and services has exploded during this last years; leading to think that company will be producing for a “market of one”.
What about the consumers? Is their utility increasing thanks to the fact that they found exactly the rare product they were looking for? The behavior of customers has changed and suppliers must adapt to that change of behavior.
As a conclusion, I think I am not wrong with stating that the internet seems definitely something that has increased the efficiency in the responding to the demand by the suppliers. While before consumers had to be satisfied with goods that did not suit exactly their preferences, they now can find what they hope for on the internet and are ready to pay higher prices for it.
You raise a number of very interesting issues. Good!
Zhang Beili
Numerous studies have demonstrated that the positive impact of recommendations systems leaned, indeed, more towards the ‘’long tail business’’ by having a larger impact on niche consumers because they would have more interactions between themselves (Drane, 2007) (1). It was also justified by the facts that the profit margin would be higher for the niche products in comparison to ‘’best-sellers’’… Read more
Numerous studies have demonstrated that the positive impact of recommendations systems leaned, indeed, more towards the ‘’long tail business’’ by having a larger impact on niche consumers because they would have more interactions between themselves (Drane, 2007) (1). It was also justified by the facts that the profit margin would be higher for the niche products in comparison to ‘’best-sellers’’ products, which wouldn’t give as much profit : due to their popularity, the competitors would have to sell them at the same price. (Oktar, 2009) (2).
However, two conditions have to be completed in order for the long tail business to thrive. First, the products have to be readily available (no stock shortage) and second, the products can be easily found. (Anderson, 2006) (3). The later imperative is the one that causes problem. Indeed, the most popular existing recommendation algorithms are not efficient for helping ’’users to find long tail items with limited historical data due to data sparsity, even if they would be viewed favorably.’’(4)
As a matter of fact, one of the most popular recommendations model, the ‘’collaborative filtering’’ take into account our past usage and on a huge amount of usage data from other users (Youtube use this system for example) in order to recommend products to the users. The problem is that this method doesn’t benefit new products or less popular items situated at the end of the long tail. (Domingues et Al., 2014) (5).
Fortunately, these last three years, numerous studies, whose purpose is to design new recommendation algorithms, have been written in order to solve the problem and stimulate the ‘’whole’’ of the long tail. (6)
To conclude, even if the actual recommendation system does make the tail longer, the most used recommendation algorithms cannot help customers find lesser known items situated at the end of the tail, even if they would actually correspond to the users’ needs and wants. Thankfully, new studies have been done in order to optimize the recommendations algorithms and make the system more efficient so that the long tail business model can thrive as a whole.
Recommendations systems are increasingly present on many websites. We particularly cited non-exhaustively the most famous like Amazon, Netflix, Pandora, etc.
This topic is quite complex and involves many mathematical concepts and probability. I think that the data collect is one of the important topics to discuss when we talk about the recommendation system. The data collection can be done in two… Read more
Recommendations systems are increasingly present on many websites. We particularly cited non-exhaustively the most famous like Amazon, Netflix, Pandora, etc.
This topic is quite complex and involves many mathematical concepts and probability. I think that the data collect is one of the important topics to discuss when we talk about the recommendation system. The data collection can be done in two ways: implicit and explicit. Implicit data collect is rarely used for websites because it is more binding on the consumer. Indeed, the consumer must explicitly show these preferences in relation to a set of products or services. This may for example be through a ranking system where the user class products / services based on these preferences.
The implicit collect of data, meanwhile, is more complicated to implement because it uses more complicated algorithm but is much less restrictive over the consumers. It is this type of collect is most often used for websites. This may for example be based on the consumer objects seen on the website. Other criteria, less common, such as interests, the places of residence, the age of the consumer, etc. can also be took into account.
Of course, each of the data input to the algorithm must make a difference in importance according to criteria defined by the website. Example, we can easily imagine that the data “last purchase made” or “product in the basket” have far more important than ata “product page visited.”
To maximize the effectiveness of recommendation systems, a latent semantic models is often used. Professor Saerens said in his courses: “We say that semantically related words are used when they are in the same context “. For example, the words “baby” and “newborn” have semantic similarities. This allows, for example, that the recommendation system provides consumers with product recommendation with description as “baby pushchair” while the consumer was looking for “newborn pushchair.”
The long tail approach is a way to describe niche markets and how they work on the Internet. At the very beginning the physical limitation of stores didn’t allow them to expand their supply, hence they preferred to offer records, books, movies, and other items that would have been successful. Nowadays with the Internet revolution and online shopping development… Read more
The long tail approach is a way to describe niche markets and how they work on the Internet. At the very beginning the physical limitation of stores didn’t allow them to expand their supply, hence they preferred to offer records, books, movies, and other items that would have been successful. Nowadays with the Internet revolution and online shopping development this system is outdated because there’s no need of a geographic location. So Amazon can sell obscure books, Netflix can rent obscure movies, and iTunes can sell obscure songs. But the same concept can also be applied to Google, in the sense that as Anderson said they are “long tail advertisers”. Rather than focusing on being the most popular page, in the niche market it is better to appear in a lot of pages.
According to me this kind of approach has for some websites many limitations. For example the recommendation system of Youtube should be improved otherwise you find yourself in a loop, in which you feel annoyed by the same related music videos . So because I consider inefficient their selection I use another app called 8tracks. The main difference is that with some tags this app will always find some items that respect my preferences, at the same time are innovative and I haven’t ever listened before.
In conclusion, this approach has his limitation because from one side we have a broader range of selected products, but on the other side the community that ranks and evaluates the product is smaller (and the long tail approach is valuable only if you have a broad range of quality reviews). For example customers that evaluates Amazon.fr amounts only for 12% of total transactions. Hence we have biased recommendation. The trend is that many reviews are given by adolescents (self-selection because online reviewing is a social practice empowering reviewer’s self-estime and ego).
References:
Six Degrees of Reputation: The Use and Abuse of Online Review and Recommendation System, David & Pinch, http://papers.ssrn.com/sol3/papers.cfm?abstract_id=857505 http://www.fritscher.ch/hec/_media/si/reputation-systems-finale.pdf http://google.about.com/od/googleforbusiness/f/longtailfaq.htm https://www.techdirt.com/articles/20081123/1230162928.shtml
When the Internet was first created, its main purpose was to serve as a platform of information that people could just as well inject as collect. With the years passing by, the quantity of information the web contained grew tremendously.
Nowadays, it is possible to find thousands of answers to a simple question. In my opinion, we have reached a… Read more
When the Internet was first created, its main purpose was to serve as a platform of information that people could just as well inject as collect. With the years passing by, the quantity of information the web contained grew tremendously.
Nowadays, it is possible to find thousands of answers to a simple question. In my opinion, we have reached a level of knowledge so high that the efficiency of the search engines such as Google, Yahoo, etc. has been compromised. Bertram Gross defined the phenomenon of “ information overload”, in his book ” The Managing of Organizations”, the following way: Information overload (also known as infobesity or infoxication) refers to the difficulty a person can have understanding an issue and making decisions that can be caused by the presence of too much information.
Recommendation engines are therefore very useful tools, since they scan and classify the billions of data according to the desires and preferences of each individual.
However, it is not relevant to advertise products that may have already been easily discovered by the user, being famous products falling in the taste category of the consumer. A popular product does not need a boost to be visible to the consumer.
It can even be annoying to perceive the obvious individual advertising on some websites.
I believe the power and value of a recommendation system lies in its ability to introduce us to new things. Products that suit our interests, but that we would’ve never discovered without the recommendation data research engine digging it up for us. Indeed, it results in giving us the impression of finding a rare unknown pearl among the other products.
Therefore, a solution to make the tail longer could be to sensitize and boost the algorithms of recommendations systems. In order to accentuate their research performances and increase their capacity to dig deeper in the giant ocean of information the Internet is today, and brings new products to the consumers.
As conclusion, I think it is important to promote the application of the long tail theory. If we are using Internet platforms to purchase objects, it is to have access to a range of varied personalized proposals and not to be found in the case of a local business.
Making recommendations to the online consumer seems to take place in almost every industry present online. Even Walmart.com makes personalized item recommendations and keeps working on increasing the quality and frequency of its recommendations. It is interesting to note that Walmart uses information both from customers’ online purchases, as well as store purchases. Walmart wants to see its online business… Read more
Making recommendations to the online consumer seems to take place in almost every industry present online. Even Walmart.com makes personalized item recommendations and keeps working on increasing the quality and frequency of its recommendations. It is interesting to note that Walmart uses information both from customers’ online purchases, as well as store purchases. Walmart wants to see its online business do well, mainly because of Amazon’s increasing success, stealing away share from physical retailers (reference below).
Another interesting, and perhaps a little scary, example comes from the gaming industry. GSN Game is a company designing mobile games. The company collects a huge amount of information from its users and uses this information to a very wide extent. I found interesting to learn that they use this information to track customers’ preferences, but also to customize services and games accordingly.
The illustrate this, technologyreview.com says the following : « If two people were to download a game onto the same type of phone simultaneously, in as little as five minutes their games would begin to diverge—each one automatically tailored to its user’s style of play. » (reference below).
However, the company doesn’t limit its use of information to preferences and customization of services. GSN Games also induces its players to play longer and try out more games, by observing the player’s behavior, from the frequency to the speed of pressing on the screen. This way, the company is able to foresee players getting tired. The company will therefore suggest new games, in order to extend its customer’s « stay ».
Influencing people’s behavior and attitudes through « persuasive technology » has been around since the 90s, but today we can observe a new trend : companies start using data on customer behavior to design products that « are not just persuasive but specificially aimed at forging new habits », as explained by Ms. Byrnes (technologyreview.com).
Habit formation used to be the breadwinner of addiction-inducing product manufacturers like the tabacco industry and casinos. The great progress in behavioral economics and psychology of the last decade, allows many industries to create persuasive technologies.
Currently, companies seem to be focussing their efforts on understanding how people make choices, in order to stimulate activity (that is, encourage daily visits or induce longer visits as in the GSN case).
The use of online consumer data is subject to debate. Regarding this new trend in “persuasive technology”, a question to debate is the following : what are the appropriate limits of such persuasion and habit formation ? I might add to that, what will be the impact of such persuasion on our conception of freedom and freedom of choice?
According to Ms. Byrnes, there is a need for regulation. Transparency and disclosure, or even asking for a user’s permission to persuade him/her, could be a first step towards it (Nanette Byrnes, technologyreview.com).
I’m not sure to see the link with the long tail hypothesis.
Zhiyi Wang
In my opinion, recommendation systems make the “tail” longer.
According to Chris Anderson, the theory of the Long Tail is that our culture and economy is increasingly shifting away from a focus on a relatively small number of "hits" (mainstream products and markets) toward a huge number of niches in the tail.
With the fast growth of internet market, people get… Read more
In my opinion, recommendation systems make the “tail” longer.
According to Chris Anderson, the theory of the Long Tail is that our culture and economy is increasingly shifting away from a focus on a relatively small number of “hits” (mainstream products and markets) toward a huge number of niches in the tail.
With the fast growth of internet market, people get access to enormous amount of products. As the costs of production and distribution fall, especially online, there is now less need to lump products and consumers into one-size-fits-all containers. In an era without the constraints of physical shelf space and other bottlenecks of distribution, narrowly-targeted goods and services can be as economically attractive as mainstream fare. I, for an example, am an user of Netflix, and I find that the recommendation lists are especially useful. I love comedy shows, but I only watch the popular, or mainstream comedies; however, Netflix recommend many good comedies to me, some of them I would never thought about before. With the recommendation systems, people get more access to the niche products.
In today’s music industry, music consumption is biased towards a few popular artists. For instance, in 2007 only 1% of all digital tracks accounted for 80% of all sales. Similarly, 1,000 albums accounted for 50% of all album sales, and 80% of all albums sold were purchased less than 100 times. Current music recommendation systems try to accurately predict what people demand to listen to. Current music recommendation systems try to make accurate predictions about what a user could listen to, or buy next, independently of how useful to the user could be the provided recommendations. There is definitely a huge need to assist people to filter, discover, personalize and recommend from the huge amount of music content available along the Long Tail.
The client is an ambassador of the product. Collaborative marketing is largely based on the good opinion from satisfied customers. The marketer’s intervention is useful to organize the recommendations to make them have a greater impact and to moderate critics by answering them with moderation in order to give credibility to these consumer’s opinions. Publishing customer recommendations has become a… Read more
The client is an ambassador of the product. Collaborative marketing is largely based on the good opinion from satisfied customers. The marketer’s intervention is useful to organize the recommendations to make them have a greater impact and to moderate critics by answering them with moderation in order to give credibility to these consumer’s opinions. Publishing customer recommendations has become a popular manner to promote online businesses and establish good reputation. An ambassador represents the brand. Strategically speaking, the brand has to identify the key opinion leader, create a community and turn it to good account. (Bernard, Fèvre, Kovacs, 2012).
This function is taken by the community manager (Synthesio, 2010) who becomes the spokesperson of the brand on the web. His or her function consists in following the brand into its marketing strategy in building a real community of ambassadors.
A concrete example of community management is the Gatorade’s experience of Command Center. Led by a team of six community managers, the Command Center conducts the community of fans and meets the consumers’ demands or claims.
Unfortunately, this comments’ management system has also been applied negatively on some booking websites of, for example, restaurants or hotels. This opportunity has been taken by companies who offered positive comments in exchange for payment for these advertisers to give a fictitious web-reputation.
Manjul Gupta and Joey F. George (Iowa State University) analysed this issue in a round table presentation called ‘the effects of fictitious customer testimonials across different cultures’ and concluded that “relatively little is known about how customer testimonials influence perceived deception and trust among web users ” (Gupta, George, 2013). Studies still have to be launched in order to analyse more deeply the impact of fictitious recommendations on web-reputation and consumers.
However, when the recommendation is spontaneous, the client becomes a business partner who needs neither a gift nor a payment but only that the real product’s quality meets the promised quality. A satisfied customer with a great friends’ network allows to quickly convincing new targets with a minimal marketing effort.
I’m not sure to see the link with the long tail hypothesis.
Gérard Christophe
First of all, the recommendation system ables to "predict the rating or preference that user would give to an item thanks to the user’s past behaviour" (Wikipedia). This system permits to reveal the truth about what consumers want and in some cases, helps some items to be selled again (Touching the Void of Joe Simpson with his book that increases… Read more
First of all, the recommendation system ables to “predict the rating or preference that user would give to an item thanks to the user’s past behaviour” (Wikipedia). This system permits to reveal the truth about what consumers want and in some cases, helps some items to be selled again (Touching the Void of Joe Simpson with his book that increases their sales decades after his release). But is that recommendation system increases the long tail effect ?
I think it depends on the sector, the type of the product and the algorithm used by the website. I’ll take some example. If you use an algorithm which recommends you a book or a film with the same subject it could be very interesting and it makes the tail longer. With « Allocine », when you read details and the test of a film, the website provides other films that « you could like » on the same subject (and not only best-sellers or films from the same producer). From my own experience, I really like film of « prisons ». When I read the test of « The Green Mile », I discovered a lot of really interesting films on this subject and which was not really known by people because of the lower marketing resources of the producers.
On the contrary, Youtube recommends most of the time « videos » or « music clips » that are the most viewed (« Vevo » recommebds each time the most popular songs). We also realize that some « videos » are all the time recommended. Youtube favors the biggest creators with more money who pay to be seen. In this case, recommendation make the tail shorter for the small budgets.
In order to improve his recommendation system, Netflix launch a competition « the Netflix Prize », where the best improvement was rewarded of 1000000$. The winner of this competition improved the algorithm just 10% better but finally, they never implemented this solution in their company. WHY ??
Because the « additional accuracy gains that we measured did not seem to justify the engineering effort needed to bring them into a production environment » (Forbes.com).
To conclude, the approach of your recommender system will depends on your algorithm and what you want for your website. The first approach is the Collaborative filtering “which can be constructed from a single user’s behavior or also from the behavior of other users who have a similar behavior.” The second one approach is the “Content-based filtering which constructs a recommendation on the basis of a user’s behavior” (ex : historical browsing information). You also have hybrid approaches which is a mix of these both approaches and which could be the best thing according to your personal offer.
Many recommendation systems exist and evolve over time. Consider first the example of the digital music player Spotify. Spotify has decided to change its recommendation system into a more intelligent system. In fact, before it used a collaborative recommendation system. That is to say that if two people often listen to the same kind of music, they will say they… Read more
Many recommendation systems exist and evolve over time. Consider first the example of the digital music player Spotify. Spotify has decided to change its recommendation system into a more intelligent system. In fact, before it used a collaborative recommendation system. That is to say that if two people often listen to the same kind of music, they will say they have the same musical tastes. It is their behavior on the musical platform that are analyzed. But this technique has a big disadvantage it is that it often tends to recommended songs that are already known because they are often the ones that come first. But the music library likes to make the user discover new songs that he has never heard before.
Therefore, Sander Dieleman, a PhD student and intern Spotify, decided to completely change the system and use deep learning to recommend songs. This is to analyze the type of song, for example: gospel, French rap, etc and propose other songs in this same range of songs, new ones that are completely unknown to the user, but could turn into a beautiful discovery. I think this system is well done because it avoids to only offer songs already known and allows the user to expand its horizons.
Another model of recommendation is the “peer-to-peer”. A first user is given a list of movies, music, books, … and should see if he could possibly recommend this list to a friend. If this is the case, the friend will receive this list by saying that the first user recommends this selection. But this selection doesn’t only include the elements of the selected list, but also other elements that have been selected by a recommendation engine. The fact that a friend recommends a product, that can be a book, a song, a movie or something else, pushes us to pay more attention because we trust him. And the fact that there are additional products that are available is also a good idea because it can lead to discoveries.
Each recommendation system has its limits and this is why hybrid systems are often proposed to overcome the disadvantages of each system and try to take advantage of their benefits. For example, there is a hybrid system that consists of a combination of collaborative system and content-based methods.
Finally, I think that social networks can play a very big role in recommendation systems, especially in peer-to-peer. Social networks can help to improve recommendation systems. For example, if two people are friends, we can consider that they have some things in common and it is not necessary to try to look for other means of similarities. Finally, social networks put a lot of information available about users. This is a real gold mine for data collection. Thus for example through notices “like” a system can already identify potential interests of its user.
Sources:
Kanna Al Falahi, N. M. (2014). Computational Social Networks. Chapter 18: Social Networks and Recommender Systems: AWorld of Current and Future Synergies. Springer.
For many platforms recommender systems play a role as important as the Search, Advertising and Networking systems; for example, for Google News recommendations generate 38% more clickthrough, 35% of Amazons sales comes from recommendations, 2/3 of the movies watched at Netflix are recommended and 28% of the people at Choicestream would buy more music if they found what they liked.… Read more
For many platforms recommender systems play a role as important as the Search, Advertising and Networking systems; for example, for Google News recommendations generate 38% more clickthrough, 35% of Amazons sales comes from recommendations, 2/3 of the movies watched at Netflix are recommended and 28% of the people at Choicestream would buy more music if they found what they liked. As available data is become bigger and bigger the algorithms used to build Recommender systems are becoming more complex and accurate meaning that the importance of user and data mining is going to be a growing trend in many areas for such us platforms.
According to “Recommender Systems Handbook” authors, there are 2 “traditional” recommender systems: Collaborative Filtering (recommend items based only on the users past behavior ) which can be user-based if the recommendations depend on what similar users to me prefer or item-based if they depend on similar items to those that I have previously liked. The second one is Content-based Recommendations (based on items features).
There are also some “novel” methods including Personalized Learning to Rank (treating recommendation as a ranking problem), Context-aware Recommendations (recommendations based on user features, specially demographics) and Social recommendations based mostly on users similarity and trust. What most platforms put in practice is a Hybrid method that combines any of the above methodologies depending on the domain and particular problem the platform is face to or in order to improve results in specific cases such as the “cold-start” problem those platforms face when they don’t have yet an important users base. The most common Hybrid methods are Weighted where outputs from several techniques (in the form of scores or votes) are combined with different degrees of importance to offer final recommendations, Switching (depending on situation, the system changes from one technique to another), Mixed (several techniques at the same time), Feature Combination (features from different recommendation sources are combined as input to a single technique), Cascade (the output from one technique is used as input of another that refines the result), Feature augmentation (The output from one technique is used as input features for another) and Meta-level where the model learned by one recommender is used as input to another. According to Amatrianin (2014), in the general case it has been demonstrated that the best isolated approach is Collaborative Filtering.
As a practical example. In Netflix recommendation approach they go from the rating system (main input) with which they build a ranking and finally a “page generation” (grid with rows and column). Everything is personalized, for example the genre rows are generated in an implicit way (based on user´s recent plays, ratings, and other interactions), explicit way (taste preferences) and combining the both (hybrid way). Thanks to their methodology the go from a “page generation” of 10,000´s of possible rows and thousands of possible videos for each row to one personalized page of 10-4o rows. They try to have a balance between accuracy and diversity, discovery and continuation, depth and coverage and Freshness and stability; this is in line with the 3 rules described in The Long Tail, especially when having a balance between Hits and less well-known films.
From the Netflix case is interesting how they don’t put so much emphasis on the social side since they argue that if they know enough about their customers, social information becomes less useful; they track what users get shown and their interactions but they recognize that from an infrastructure perspective it is very costly. Also regarding the context they had to improve their recommender systems, Netflix own offline studies showed the wining models were too computationally intensive to scale so the expected improvement was not worth the engineering effort (they also acknowledged that their focus had shifted to other issues that had more impact than rating prediction).
What an extraordinary news for artists, writers or even designers who previously needed to reach a certain amount of potential consumers in order to be launched. This amount was the minimal number of copies that would be sold to make the producer and the distributor profitable. Now, it is still true but this number is way lower as the distributor’s… Read more
What an extraordinary news for artists, writers or even designers who previously needed to reach a certain amount of potential consumers in order to be launched. This amount was the minimal number of copies that would be sold to make the producer and the distributor profitable. Now, it is still true but this number is way lower as the distributor’s costs have become lower as well. Thereby, a huge creative potential has been unleashed. The diversity of products is exploding. To me, this phenomenon is the greater advantage of the long tail theory.
However, this amount of new products on the market is creating an subsequent problem: how to promote the not yet known creations?
From my point of view, various recommendation systems are the solution. It enables consumers to discover little known artists they will enjoy even more than very popular ones. In order to give a personal example, I will share my experience with Spotify, a music streaming platform. Traditionally, the recommendation system used by Spotify is a collaborative filtering approach [1]. This technique do not use any of the content characteristic. It is exclusively based on the consumption habits. However, this recommendation system is not very appropriate to make the tail as longer as possible. Indeed, since it is only providing recommendations of listened songs, it is more likely to be recommended famous songs than creations unknown from the general public. Therefore, Spotify has started to get involved in content-based approaches. There are numerous informations that can be learned from the song such as the artist previous featuring and music style, lyrics and especially the music signal in itself. The content-based approach enable the members to discover completely new songs since it doesn’t even need to be rated or previously listened [2]. Unfortunately, the use of the music signal is very complicated and needs to be improved in the coming years.
Even if the recommendation system of Spotify isn’t perfect and isn’t exploiting yet the content-based approach at its best, I think that it enables users like me to discover many more artists and thereby increases the diversity of discovered songs.
According to Chris Anderson « The theory of the Long Tail is that our culture and economy is increasingly shifting away from a focus on a relatively small number of "hits" (mainstream products and markets) at the head of the demand curve and toward a huge number of niches in the tail. As the costs of production and distribution fall,… Read more
According to Chris Anderson « The theory of the Long Tail is that our culture and economy is increasingly shifting away from a focus on a relatively small number of “hits” (mainstream products and markets) at the head of the demand curve and toward a huge number of niches in the tail. As the costs of production and distribution fall, especially online, there is now less need to lump products and consumers into one-size-fits-all containers. In an era without the constraints of physical shelf space and other bottlenecks of distribution, narrowly targeted goods and services can be as economically attractive as mainstream fare »(1)
In my opinion, recommendation systems can in certain cases make the “tail” longer but it depends on certain factors.
In order to understand how online product recommendations could influence consumers’ online choices, we should refer to the article of Sylvain Senecal and Jacques Nantel (2).
In their article, they aim to investigate consumers’ usage of online recommendation sources and their influence on online product choices.
In their study, they focus on three determinants that could influence the impact of computer-mediated recommendations on consumers’ online product choices.
The three determinant that they investigate in their study are: the nature of the product recommended, the nature of the website they are recommended on and the type of recommendation source.
The first determinant underlines the fact that goods can be classified as possessing either search qualities (those that the consumer can determine by inspection prior to purchase) or experience qualities (those that are not determined prior to purchase).
The second determinant underlines the fact that the nature of the website can also influence the impact of a given recommendation.
They are three kinds of websites: sellers, commercially linked third parties and non-commercially linked third parties.
The last determinant sorts the type of recommendation sources in three categories: other consumers, human experts and expert systems.
After testing several hypothesis on these assumptions, they were finally able to show that consumers would be more influenced by recommendations for experience products than for search products which seems logical as it is difficult or even impossible to evaluate experience products before purchase, consumers should then rely more on product recommendations for these products than for search products.
But also that no relationship exists between the type of website and subjects’ propensity to follow product recommendations
And finally, that the recommender system (expert system) is more influential than other consumers and human experts.
It appears that a recommendation source providing personalized information to consumers is more influential than a recommendation source providing non-personalized information.
To sump up, I think that the article of Senecal and Nantel help us understand that the theory of the long tail is not sufficient to itself and that other factors must be taken into account if we want to evaluate the influence of a recommendation system.
Indeed, I think that a recommendation system could make the tail longer but it will mostly depend of the type of product and the type of recommendation system.
Having said that, I think that the theory of the long tail doesn’t apply with the same intensity in every sectors or industries.
Then, we cannot generalize a positive impact of the recommendation systems.
Currently, all the recommendation systems are taking more and more importance on Internet. Indeed they are simply omnipresent. Here are some examples:
• Online purchasing website knows what are your preferences and give you suggestion once you bought a certain product.
• Social networks manage to find friend you probably know or “fan-pages” you probably like,
• Broadcast video website (like Youtube) propose related… Read more
Currently, all the recommendation systems are taking more and more importance on Internet. Indeed they are simply omnipresent. Here are some examples:
• Online purchasing website knows what are your preferences and give you suggestion once you bought a certain product.
• Social networks manage to find friend you probably know or “fan-pages” you probably like,
• Broadcast video website (like Youtube) propose related videos (based on the content of what we just have seen) .
It is clear that the main advantage of this is that it can be very helpful to make discoveries of new artists, persons … Besides, it’s not new that people refer to others in other to gather information. For instance, before watching a movie, many of us check on the net the critics on specialized website.
However, there’s something that really bother me and raise an important issue. By getting all those “suggested” products or services, I personally think that we inhibit our free will and choices. I have the feeling that, due to those recommendations system, it’s making us like robot that follow what the firms claim good to offer. It is sad that our choices rely that much on mathematical algorithms and are not especially based on the personal willingness to discover the unknown.
A first update worth mentioning is the distinction between user recommendations and system recommendations. This distinction is around for a wile now, however, only recently these different types are empirically investigated, moreover the difference in effects of both was studied.
On the one hand we have system recommendations, which are most commonly generated by the collection of all user preferences.… Read more
A first update worth mentioning is the distinction between user recommendations and system recommendations. This distinction is around for a wile now, however, only recently these different types are empirically investigated, moreover the difference in effects of both was studied.
On the one hand we have system recommendations, which are most commonly generated by the collection of all user preferences. With these preferences the firm will suggest products to consumers. A very well known example of a system recommendation is the co-purchase recommendation of Amazon. Here, recommended products will be listed as “customers who bought this item also bought”.
On the other hand we have the user recommendations or also called user-generated word-of-mouth (WOM). They display consumer reviews of products whereby the review is based on the volume and the rating of the product.
This division of recommendation systems into two different types is necessary according to the paper of Zhijie Lin because according to her, both recommendations will generate different impacts. According to Lin, user recommendations are more effective than system recommendations in driving product sales.
Recommendation systems are already very popular and in my opinion they will only increase during the years. I based this comment on some striking results I found in a recently conducted survey:
– 85% of customers consuls online reviews before purchasing
– 65% of consumers claim that there purchasing decision is influenced by online reviews of other customers.
To further optimize the recommender systems firms need to increase the amount of consumer feedback that they receive. This because customer feedback will improve accuracy and can lead to a learning curve for the firm to notice a pattern in consumer preferences.
Literature has extended its knowledge about the efficacy of the recommendation systems quite a lot. Literature found that there are some differences in efficacy in relation to the sector/ type of products sold but also in relation to the characteristics of beneficiaries and in firms.
A recent paper from Harvas-Drane (2015) found empirical prove for the fact that recommendation systems do not work as efficient in all sectors. He argues that these recommendation systems are very useful for the following products: music, films, books and video and game entertainment.
He also makes a distinction in efficacy of the recommendations according to the characteristics of beneficiaries. According to him, customers with less prevalent preferences will create a long tail effect and the concentration of sales will be reduced. Consequently products that have a smaller share of total sales, otherwise called niche products, will generate a long tail. This finding is still in accordance with the long tail theory, firstly discussed by Anderson in 2004.
The firms that have lower inventory costs will benefit the most of the long tail effect because they can increase the depth of their assortment more easily and to a wider extent then others. As a consequence these firms are better positioned to serve demand for niche products (that leads to the long tail effect).
A recent study argues that word of mouth can contribute to the explanation of the long tail phenomenon.
Harvas-Drane (2015) links recommendation systems with word of mouth, which led to some interesting findings. He argues that recommendation systems and word of mouth could benefit all consumers (because of the lower search costs), but they have a larger impact on those consumers that have less prevalent preferences. However, there are also benefits for the firms: the market share of niche products in the sales distribution will increase, leading to a longer tail. But also a reduced concentration of sales within the firm’s assortment is noted.
References:
Hervas-Drane, A. (2015). Recommended for you: The effect of word of mouth on sales concentration. International Journal of Research in Marketing .
Zhijie, L. (2014). An empirical investigation of user and system recommendations in e-commerce. Decision Support Systems (68), 111-124.
Before answering the question, it may seems interesting to see the evolution of the « tail » over time. As it is said in the article « Goodbye Pareto Principle, Hello Long Tail: The Effect of Search Costs on the Concentration of Product Sales » roughly 10 to 15 years ago, the Pareto Principle was prominent.
This principle is quite… Read more
Before answering the question, it may seems interesting to see the evolution of the « tail » over time. As it is said in the article « Goodbye Pareto Principle, Hello Long Tail: The Effect of Search Costs on the Concentration of Product Sales » roughly 10 to 15 years ago, the Pareto Principle was prominent.
This principle is quite simple : 20% of the products available on a market generate 80% of its sales. The distribution of the sales was therefore highly concentratred and a lot of niche products were never sold at all.
However, with the internet this trend may disappear, being replaced by what is called « the long tail ».
The authors identify two causes to this transformation of the market.
First, internet catalogues can offer much more choice than retail stores, as they do not need to have physical restrictions (stock size can be much bigger : eg. Amazon’s warehouses size).
Second, the customer can access much more information at lower costs and therefore be interested in niche products.
To answer the question and see if recommandation systems still allow to create a « long tail », we will first speak about personnal experience.
The first example we have is Spotify : it has been a few years now that we have subscribed and since we have discovered a lot of new songs. These songs are related to the kind of music that we listen, share and search, but is unkown and off the charts, there is a real feeling of « personnal assistant » looking for the songs we may like.
The second example is Netflix, that is available in Belgium since a few months. According to the article of the Huffingtonpost in the references, the firm is making changes in the market of entertainment by using a long tail strategy.
Indeed, the firm buy licenses for a lot of small shows that attract users and spur them to renew their subscription : these are the shows that create the « long tail » and that are suggested to the user according to his preferences. With the gains of the subscription, they pay for blockbusters as « House of Cards ».
From our personnal experience, the suggestions offered by Netflix are less « exclusive » than the ones offered by Spotify. The films, shows, series suggested are already known most of the time and although we also realize that there is some personnalization, we have the feeling that it is less advanced.
Maybe because of the size of the catalogue in Belgium for the moment, or because of a different algorithm.
Finally, it is quite difficult to find recent information (less than 2 years) about the subject, excepted from the paper of Oesterreicher-Singer and Sundararajan.
Nevertheless, considering the fact that the strategy of most services today on the web is a personnalized targeting and that most people (us included) want personnalized content, we can guess that recommandation algorithms developped by the firms will go this way and make the tail longer and longer over time.
I have notice that we can find a lot of things on the long tail phenomenon and the recommendation systems. The first paper I would like to mention has been written by Bin Gu, Qian Tang and Andrew B. Whinston and deals with the effect of online information on the long tail phenomenon. In this paper the authors mentions two… Read more
I have notice that we can find a lot of things on the long tail phenomenon and the recommendation systems. The first paper I would like to mention has been written by Bin Gu, Qian Tang and Andrew B. Whinston and deals with the effect of online information on the long tail phenomenon. In this paper the authors mentions two opposite beliefs concerning the effect of information on the formation of long tail. The first belief considers that the existence of online information available for customers encourages the formation of long tail. On the other hand, the second belief thinks that the customers ignore the available information and that they base their choices only on their personal beliefs.
The three authors have tried to determine which belief is the right one. Their study showed that positive comments improve the sale of popular product more than the sale of niche product. Consequently it means that the existence of information on the Internet has a negative impact on the formation of long tail.
I find this conclusion interesting because personally I always read comments of other customers before watching a film for example. Moreover I think those comments have more and more importance on the web. Indeed many people decide to make comments and therefore their influence should be increasing.
Huanxing Yang has developed in 2013 the second analysis I would like to mention. He has created a search model to explain the long tail effect. In his theory, the long tail effect is due to the decrease in the price of searching or an increase in the search targetability. What does the author means by the price of searching and search targetability? Today with the Internet, customers have easily access to a large number of products. They are note limited to the products available in the store they visit. Consequently the distribution of sales has become flatter and niche products have gain market share. So the long tail effect is encouraged by the decrease in the price of searching and the increase in the search targetability.
Finally I have noticed that several techniques have been developed in order to maximize the formation of long tail. Techniques such as adaptive clustering, hybrid music recommender system which combined usage and content data, ect.
References:
Jeyshirii, S., Thivakaran,T.K. (2014). Precise Recommendation System for the Long Tail Problem using Adaptive Custering Technique. International Journal of Innovation Research in Computer and Communication Engineering, 2(4).
Gu, B., Tang, Q., & Whinston, A. (2013). The influence of online word-of-mouth on long tail formation. Decision Support Systems, 52, 474-481.
Yang, H. (2013). Targeted search and the long tail effect. RAND Journal of Economics, 44(4), 733-756.
In order to articulate my point, I’m going to use the on demand music streaming service Spotify and its recommendation system.
Spotify is a rather new service that takes the tangent compared to its offline service counter parts like iTunes, Winamp and foobar2000. With the ever-increasing ease to have an internet access through WiFi and 3G/4G, Spotify has managed to… Read more
In order to articulate my point, I’m going to use the on demand music streaming service Spotify and its recommendation system.
Spotify is a rather new service that takes the tangent compared to its offline service counter parts like iTunes, Winamp and foobar2000. With the ever-increasing ease to have an internet access through WiFi and 3G/4G, Spotify has managed to make its service a leading force in the music software industry. Aside from being in the right place at the right time, Spotify has also made a name for itself thanks to its recommendation system that goes much deeper and further than any other in the genre.
Spotify has mostly relied on collaborative filtering which means that the users’ preferences come from historical data usage. While this system works and might be enough in most services (Amazon, imdb …), in the case of music, it has its flaws. Because it relies on usage data, the more popular and item is the more likely it will be recommended. Unpopular songs are therefore barely recommended which results in make the tail shorter.
Spotify have been developing what they call ‘Deep content-based music recommendation‘. Sander Dieleman, one of the authors or this method describes it as a way to “tackle the problem of predicting listening preferences from audio signals by training a regression model to predict the latent representations of songs that were obtained from a collaborative filtering model.” Of course we are not interested in the technical side of things but the fact that one of the leading It companies in the field of streaming service is looking to improve its recommendation system to lengthen the tail speaks volume.
The music streaming service market has hit a point where users need more and want to discover new things, sticking with a somewhat out-dated system like manual curation for Songza and Beats or manually tag attributes for Pandora or even audio content metadata for echonest doesn’t cut it anymore.
In conclusion, the actual technical method used for the recommendation system can lead to short or long tail. It seems the more mature an industry is, the more precise and well developed the recommendation systems are which leads to a longer tail.
Firstly we have to say that the recommendation system is a very useful tool to help the customer for finding the genre(s) and style of products they like. But it has to be used effectively. Indeed, like said in the article this concept of recommendation system is not new anymore and a majority of customers rely mainly of this kind… Read more
Firstly we have to say that the recommendation system is a very useful tool to help the customer for finding the genre(s) and style of products they like. But it has to be used effectively. Indeed, like said in the article this concept of recommendation system is not new anymore and a majority of customers rely mainly of this kind of system in order to find other product related to their taste.
The fact is that the utilisation of this system is so fast with internet now that it will be (for a lot of persons) a waste of time to do research of their own. We can point the example of ebay,alibaba or even amazon like it was presented in the text. The recommendation system in the site are just under the image of the command and then really easy and fast to find.
However eventough the recommandation systems have their avandtages we can easily point out bad consequences of those kind of processes.
we can explain those consequences in 2 main points.
The first one is what we call “collaborative filtering”. This represents the process of filtering information by using techniques involving collaboration among multiple agents. The main point is to help the customer to find popular films, musics, books or whatsoever related to their taste. The fact is that this system doesn’t take into account a lot of works made by newcomers or simply unadvertised works. Indeed the fact is that the customers can only see 5 or 6 other works in this kind of recommendation system, this increases then the inequality factor between those kind of works.
The second point is called “collective intelligence”: which are the cognitive abilities coming from the interactions in a community between the members. This knowledge of those members is then limited because their vision is highly restricted due to the fact that they share the same vision of the environment (world) and are then not fully conscious of the surroundings.
We can apply this to the case because a lot of recommendation system create this kind of community and then close the possibility of the members to expand their vision in the genre they like.
We can then conclude that only using recommendation system will favor only few works and then don’t give the chance to a lot of potentially good contributions. The only use of recommendation system will then create a shorter tail for the market with a huge difference between blockbusters and other works.
I do not believe recommendations make the long tail longer for every type of products. When it comes to books or movies, their tail is longer because of their creative/artistic aspect that sometimes lives forever. A good book remains good, even after the pick of sales, and the tail will, in my opinion, never end. I believe the question to… Read more
I do not believe recommendations make the long tail longer for every type of products. When it comes to books or movies, their tail is longer because of their creative/artistic aspect that sometimes lives forever. A good book remains good, even after the pick of sales, and the tail will, in my opinion, never end. I believe the question to ask in this case is if recommendation makes the tail ticker. The answer to this, in my opinion, is that it should. When I watch a movie, I like it, and I am recommended a similar one “You might also like …”, I most certainly purchase the recommended one, out of curiosity. A book or movie that was recommended to me and that I did not find extraordinary, I usually put the blame on the difference in taste, but never regret the decision to buy it.
But there are those products that become out of fashion/trends, and, in my opinion, no recommendation will make the tail longer for them. Let’s take for example mobiles phones: six years ago no recommendation system would inspire any type of consumer to buy a Nokia phone. Nokia missed the trend of touch-screen, went from being the world’s most respected mobile phones brand to being forgotten, and only reappeared slightly on the market with the Windows Phone. For this type of products, the recommendation system works only while they are in the pick period. A study on the influence of users’ review on Smart phones purchase suggested that people would not purchase a product that does not have reviews, but that the familiarity to the brand is more important that other customers’ reviews (Taskin, 2013).
Furthermore, I believe recommendation has to contain more arguments and explanations when the purchase decision is not easy (for important amounts of money). For me, the fact that the review is not anonymous gives it more credibility. What is more, the efficiency of a recommendation is also influenced by the previous experience with recommendation: if I bought something based on an online recommendation and I was not satisfied, I will have the tendency to not get influenced again by recommendations to avoid deception.
All in all, I believe recommendation has an impact on the tail, provided the tail was still positive.
Taskin, D., (2013). “Online users review: influences and influencers” (Master thesis). Louvain School of Management (Catholic University of Louvain), Louvain-la-Neuve.
According to Anderson, our economy has shifted from a small number of highly popular products (the head) towards a high number of niche products (the tail). Indeed, in traditional economies, popular products were favored in comparison to niche products because the physical room on the shelves was limited and expensive. With the internet revolution, it has become in the contrary… Read more
According to Anderson, our economy has shifted from a small number of highly popular products (the head) towards a high number of niche products (the tail). Indeed, in traditional economies, popular products were favored in comparison to niche products because the physical room on the shelves was limited and expensive. With the internet revolution, it has become in the contrary easier to build a business on the long tail given that inventories are not limited and that niche products are easier to find.
In his book, Anderson explains that 2 conditions lead to successful long tail business: making everything available and helping the customers to find the niche product. The second condition is fulfilled with the recommendation tools develop by players in long tail business. Recommendation systems design aims at making the consumer select the most important item for him.
What is at stake in this post is the effect of recommendation systems on sales. I will consider the 2 issues that are the most present in the literature: the impact on sales volume and sales diversity. Diversity is directly linked to the long tail phenomenon. Indeed, recommendation systems can have two distinct impacts:
Helping consumers discover new products that fit better with their preferences, hence increasing the diversity;
Reinforcing the popularity of already famous products, hence leading to more concentrate sales and thus less diversity.
It had been proved that recommendation systems have a significant impact on sales volume but the impact on the diversity of purchase seems to be more controversial.
This post was originally written in 2012 and I would like to update it with a paper written in the context of the 35th international conference on information systems in 2014 by Lee and Hosanagar.
In this paper, the authors analyze the impact of different recommender algorithm on both sales and diversity since, according to the authors, different algorithms have different impact on these two variables. The debate in the literature on the effect on diversity by recommenders systems is due to these differences. Recommendation systems are classified by Schafer, Konstan and Riedi in 6 types: the collaborative filtering (consumers who bought this item also bought that item), the content-based (according to your consumption history), the hybrid (combining the first two types), the social-network (what your friends bought), the popularity recommender (the most purchased items) and the recently-viewed (the most viewed items). Lee and Hosanagar analyze the impact of different recommender systems by measuring the diversity at the item and genre level with the use of the Gini coefficient mentioned in this post. The Gini coefficient has a range of values between 0 and 1 and measures the level of concentration diversity. A coefficient of 0 means that all products have the same amount of sales, while a coefficient of 1 means that a number of popular products account for most of the sales.
As explained before, the kind of algorithm used has an influence on sales volume and diversity. I am going to compare two of them: the collaborative filtering and the recently-viewed algorithm. One of the results of the study is that a collaborative filtering algorithm increases the sales volume and the individual diversity but decreases the overall aggregate diversity. Indeed, it seems that individuals discover a higher diversity of products through links provided to other items that can match with their preferences. However, the algorithm introduces a popularity bias since consumers move towards the popular items rather than towards niche products, which introduces a concentration in the diversity at the aggregate level, and so making the tail shorter. As far as the recently-viewed algorithm is concerned, it decreases the aggregate diversity and doesn’t influence the individual diversity. In fact, this algorithm, by showing you the recently viewed items, decreases the views on other less popular items and so decreases the diversity.
To conclude, in this comment, I have analyzed some issues related to niche and popular products and the influence of recommendations on them. However, one further investigation could be to analyze to what extent the previous analysis applies to products that are classified as neither popular nor niche item.
References:
Anderson, C. (2006) The Long Tail: Why the Future of Business is Selling Less of More. Hyperion.
Schafer, J. B., Konstan, J., & Riedi. (1999). Recommender systems in e-commerce. Proceedings of the 1st ACM conference on Electronic commerce, 158–166.
Lee, D., & Hosanagar, K., (2014) Impact of Recommender Systems on sales volume and diversity, 35th international conference on Information Systems.
Shanthi, K., & Kalimuthu, M., (2013) Improving diversity using contend based approach in recommender systems, Research Journal of Computer Systems Eengineering, 4.
Due to the exponential growth of the online market, the competition entered in this field. The owners of the sites need to attract the customers by providing attractive facilities. One of them is the Recommender Engines that attempt to recommend books, movies or articles based on our past actions. For example, Netflix offered a $1,000,000 prize to researchers who could… Read more
Due to the exponential growth of the online market, the competition entered in this field. The owners of the sites need to attract the customers by providing attractive facilities. One of them is the Recommender Engines that attempt to recommend books, movies or articles based on our past actions. For example, Netflix offered a $1,000,000 prize to researchers who could improve the quality of their recommendations.
Gaurav Arora, Ashish Kumar, Gitanjali Sanjay Devre and Amit Ghumare analyse the recommendations system for movies to users. The existing system is based on individual users’ rating that can be useless for users who have different tastes. The new system compute the similarities between different users and the recommendations are based on these rating given by users of similar tastes. It provides more precise recommendations.
The system proposed by the authors it the batter system than any other existing system, according to them. “This system has added the positive features of existing systems and has overcome the drawbacks of existing systems. The system uses all the existing algorithms i.e. content based, context based and collaborative based algorithms. All these algorithms are combined to give more precise result.”(Page 768). There are four different modules developed:
1. Admin: this system allow to add movies in a database, view movies and update it;
2. Recommendation Engine: compute the similarities between the different users. It will recommend movie to a user;
3. Movie Web Service: this service allows user to rate movies, comments them and show the movie recommendations to the users;
4. Android user: the user can rate, comment and see movies.
This new system of recommendations is a hybrid one that improves the performance by overcoming the disadvantages of traditional recommendation system.
Based on: Gaurav, A., Kumar, A., Devre, G. S. and Ghumare, A. (2014). Movie recommendation system based on users’ similarity. International Journal of Computer Science and Mobile Computing, 3(4), 765-770.
Regarding the purchase, some surveys have been conducted. Here is a summary of the different results:
• According to Metail, 56% of consumers say they would be more inclined to use a retailer if it offered a personalised experience;
• An Infosys study on consumer behavior showed that 59% of shoppers who have experienced personalization in their shopping agree that it affects their buying behavior;
• A study launched by the eTailing Group show that recommendations system could improve purchases: 77% of online customers admit to have made additional purchases based on personalized product recommendations;
• 74% of online shoppers get frustrated with sites that shown them content that has nothing to do with their preferences or past buying behavior (study of JanRain).
My comment will be structured in three parts. I will first give my opinion about long tail theory, then about the recommendations systems and I will finish with the relation between the two previous.
From my opinion, the long tail theory doesn't apply to every sectors or at least it doesn't apply with the same intensity. In some industries, it's true… Read more
My comment will be structured in three parts. I will first give my opinion about long tail theory, then about the recommendations systems and I will finish with the relation between the two previous.
From my opinion, the long tail theory doesn’t apply to every sectors or at least it doesn’t apply with the same intensity. In some industries, it’s true or more true that a large part of selling products distribution came from a large number of products each sold in low quantity. As Chris Anderson show it with the examples of Amazon and Netflix. I think that it’s also apply in the book sector more than in the music or film sectors. I see two distinct reasons. Firstly, production studios as Warner or Universal exert a massive brain-washing and they promote a lot a few products. They have a larger impact on people than publishing houses. Secondly, I guess people reading books regularly have more critical and like to make their own judgement. More than that, I think readers choose more wisely their books because it asks them an effort and it’s a real engagement where watching a movie take only 1 or 2 hours. For these two reasons I think that in the book sector the tail is longer, I mean that a larger part of the books sold doesn’t come from best sellers (in comparison with the movie sector).
Concerning the recommendations systems, I think they are useful and more present than before. Nowadays, people are more connected (by trivial relation sadly) and they are asking for the opinion of others. As an example, in the app-store or on Android, the apps are ranked by their quotation and the top ranked, the more recommended are the first you can see.
Now the question is to know if such recommendations systems impedes the long tail phenomenon or not… Taking the example of Netflix, I think they do. Indeed, I am more likely to recommend a movie or a series which is not very famous than the new blockbuster. For example, I would find it ridiculous to promote the 12 grammy’s series Breaking Bad, because I suppose every body knows it. I am more likely to recommend a new series for example. So from this supposition that recommendations systems promote unknown products, it surely impedes the long tail theory.
But can I say that it’s always true, probably not. It depends a lot of the type of recommendation system. On downloading websites, here is a ranking of the most downloaded products. If we accept that as an indirect recommendation system, we have to say that in some cases they reduces the long tail effect because it promotes the “best-sellers”.
In conclusion I would say that the long tail effect mainly depends of the type of products and of the way they arte purchased (on-line vs. off-line,…).
Nowadays, recommendation systems have an undeniable importance on Internet as they can be seen nearly everywhere: social networks where they help you find people you might know or pages on subjects that might interest you, online buying platforms ("you have bought that product, you could like this other product"), or simply on Youtube where you are proposed related videos and… Read more
Nowadays, recommendation systems have an undeniable importance on Internet as they can be seen nearly everywhere: social networks where they help you find people you might know or pages on subjects that might interest you, online buying platforms (“you have bought that product, you could like this other product”), or simply on Youtube where you are proposed related videos and online news sites (where you can’t look at related articles). This mechanism can be very helpful to exapnd your personal tastes (Youtube, Spotify,…) or discover hings that you maybe didn’t even know existed, but I don’t think it has only positive aspects…
From my point of view, there are two situations and my opinion on recommendation systems will vary considerably depending on the actual situation.
The first situation occurs when there isn’t money “in the balance”. For this situation, we have examples like Facebook or Youtube. In this case, I enjoy the presence of recommendation systems. I don’t say all the propositions are interesting (because actually a big proportion of them isn’t…) but it enables me to discover new things related to things I already like. And because money isn’t involved, I don’t really care about “trying” what is suggested to me.
On the other hand, we have all the recommendations about things you might be interested to buy. For this category, I feel much less at ease with recommendation systems. The main reason for this is that I don’t really trust online recommendations whether they are made by consumers or automatically because I feel like I’m being oriented in the direction the site wants (because a particular product doesn’t have the expected success or any other reason). As far as money is concerned, I strongly prefer to rely on the experience of people I know (and trust) or on my own research on the subject rather than on any recommendation with an online source.
The article is based on the theory of long tail. The theory was developed by Chris Anderson and describes the fact that the internet sites sell a high proportion of niche products in small quantities. This theory is often applied in e-commerce because the logistic costs are reduced and the centralization of products allows to manage products in small quantities.
Before,… Read more
The article is based on the theory of long tail. The theory was developed by Chris Anderson and describes the fact that the internet sites sell a high proportion of niche products in small quantities. This theory is often applied in e-commerce because the logistic costs are reduced and the centralization of products allows to manage products in small quantities.
Before, the economists referred to another law : the law of 20-80. The law of 20-80 led to place priority on customers that represent 20% of the total customers who contribute to over 80% of sales. With time this idea changed. Now the law of the long train is in place. With this law, the idea consists of paying attention to customers who represent 80% of the total mass and 20% of sales. Now, it is possible thanks to internet.
Instead of offering only the 20% of references that make 80% of sales, it is possible to be profitable by offering 80% of references that make traditionally 20% of sales. It is the revolution of internet, we can approach the niche markets. Firms sell small amounts of products but when you add the sales, they sell a lot. Amazon, the example of the article, makes 57% of sales with unknown references.
The reason being that the uncasual users want more than popular contents. The role of recommendations is to satisfy and retain these customers. For that, the site promotes niche contents of which the profit margin is more important than the popular ones and allow to diversify the offer.
After a few years, the results are not convincing. The long train effect is small and does not have any impact on the diversity. Today, we are witnessing the opposite phenomenon, some writers, musicians buy up all of demand. Now, we pay attention to « the superstar effect ».
According to Marcello Vena, responsible of numeric department RCS Mediagroup (books), the concentration of numeric market is the crux of the problem. The concentration of retail digital titles around a few global players makes the promotion of editorial diversity impossible. The monopoly is the problem. But Marcello Vena also says that the recommendations motors are truncated with their recommendations. To sell more they offer you another title of the same author than the original title of other books that you do not know. But it is the same problem with the social recommendations because the motor offers you the most recommendations shared and not recommendations with small views that could be an interesting discover. There is also a risk of media bias. Some recommendations systems could be associated with writers or musicians and generate a bias like we studied last week.
Nevertheless, I think that the recommendations are still important for e-commerce. For me, they replace the sellers in the physical shops. Indeed, in a physical store, the sellers help the customers to make their choices, but on the internet, there is no seller. Thus, the recommendation is used to help client for e-commerce.
So the recommendation system seems to act like the word of mouth. Indeed, we can consider the human beings like the best tool to communicate thanks to the interactivity and personal opinions. Before, the sites tried to have comments from authors or specialists in the field. But now, the sites try to attract recommendations from users and create a sort of community. This community votes and gives opinions. A book can be appreciated by the experts in the field and not be found interesting by the users. That is why, with time, the recommendations of users have emerged and become essential.
With the appearance of social networks such as Facebook, Twitter or GetBlue, the sense of community is reinforced. GetBlue allows users to indicate where they are but also recommendations for cultural goods (books, music,…). GetBlue gives notifications on what your friends do, what they listen or read and you can follow the activities of users with an option « friends to follow ». The friends give recommendations and the system allows to select its own sources of recommendations on the public profile of publications or tastes.
Finally, I think that everyone is sensitive to recommendations because these respond to our expectations. We want to be sure that the purchase that we will do is a good deal, we want to reassure ourselves that we made the right choice.
I did not find any good readings proving or not what was only assumptions at the moment the article was written, but this comment will show some new ways marketers found to make us buy what we did not know whe needed.
Indeed, the article only spoke about « automatized » recommendations systems (by that I mean not taken care by… Read more
I did not find any good readings proving or not what was only assumptions at the moment the article was written, but this comment will show some new ways marketers found to make us buy what we did not know whe needed.
Indeed, the article only spoke about « automatized » recommendations systems (by that I mean not taken care by an employee of the company website you are visiting), that directly appears while checking the page of a product. But same year as the publication on the blog was published, another one was posted by « Fortune » (link at the end), which stated that « Amazon also doles out recommendations to users via email. […] the company provides some staffers with numerous software tools to target customers based on purchasing and browsing behavior. But the actual targeting is done by the employees and not by machine. If an employee is tasked with promoting a movie to purchase like say, Captain America, they make sure customers who have viewed other comic book action films receive an email encouraging them to check out Captain America in the future. » I do not have information to assess if that sytem is still used nowadays, but I found it quite interesting to see that it was not only « computerized ».
Also, with the well known success story smartphone had, it seemed pretty obvious that it will have a rôle to play in the recommendation systems. In 2013, a project was being studied and tried, consisting on the implementation of an eye tracking system directly implemented in ordinary glasses, and connected to a smartphone in order to play the same role as an ordinary recommendation system, and also help the consummer in his decision process. According to the article i also linked in the end of my comment « instead of learning from click behaviour and past customer ratings, as it is the case in the e-commerce setting, the mobile RA (recommendation agent) learns from eye movements by participating online in every day decision processes. »
In a first pilot study they did with five randomly chosen participants in a supermarket, they tried to get a first impression of the user’s behavior in a situation where he would need such technology, and his acceptance of the technology. First results show long eye cascades and short fixations on many products in situations where users are uncertain and in need for support. They also found a surprising acceptance of the technology itself throughout all ages (23 – 61 years), but at the same time, consumers expressed serious fear of being manipulated by such a technology. For that reason, they strongly prefer the information to be provided by trusted third party.
Since the article do not give a name to that project, i wasn’t able to see the actual state of the project, but I think that some Google Glasses features may have been inspired by that kind of project.
In conclusion, recommendation systems are getting more and more sophisticated and use new means in order to assess which product would fit each customer, so I do not really doubt that, if it is not already the case, marketers and developers will soon find a way to sell even the niche products to the customers that would be interested by them, and who would never had been aware of their existence without the RS.
I did not find any good readings proving or not what was only assumptions at the moment the article was written, but this comment will show some new ways marketers found to make us buy what we did not know whe needed.
Indeed, the article only spoke about « automatized » recommendations systems (by that I mean not taken care by an employee… Read more
I did not find any good readings proving or not what was only assumptions at the moment the article was written, but this comment will show some new ways marketers found to make us buy what we did not know whe needed.
Indeed, the article only spoke about « automatized » recommendations systems (by that I mean not taken care by an employee of the company website you are visiting), that directly appears while checking the page of a product. But same year as the publication on the blog was published, another one was posted by « Fortune » (link at the end), which stated that « Amazon also doles out recommendations to users via email. […] the company provides some staffers with numerous software tools to target customers based on purchasing and browsing behavior. But the actual targeting is done by the employees and not by machine. If an employee is tasked with promoting a movie to purchase like say, Captain America, they make sure customers who have viewed other comic book action films receive an email encouraging them to check out Captain America in the future. » I do not have information to assess if that sytem is still used nowadays, but I found it quite interesting to see that it was not only « computerized ».
Also, with the well known success story smartphone had, it seemed pretty obvious that it will have a rôle to play in the recommendation systems. In 2013, a project was being studied and tried, consisting on the implementation of an eye tracking system directly implemented in ordinary glasses, and connected to a smartphone in order to play the same role as an ordinary recommendation system, and also help the consummer in his decision process. According to the article i also linked in the end of my comment « instead of learning from click behaviour and past customer ratings, as it is the case in the e-commerce setting, the mobile RA (recommendation agent) learns from eye movements by participating online in every day decision processes. »
In a first pilot study they did with five randomly chosen participants in a supermarket, they tried to get a first impression of the user’s behavior in a situation where he would need such technology, and his acceptance of the technology. First results show long eye cascades and short fixations on many products in situations where users are uncertain and in need for support. They also found a surprising acceptance of the technology itself throughout all ages (23 – 61 years), but at the same time, consumers expressed serious fear of being manipulated by such a technology. For that reason, they strongly prefer the information to be provided by trusted third party.
Since the article do not give a name to that project, i wasn’t able to see the actual state of the project, but I think that some Google Glasses features may have been inspired by that kind of project.
In conclusion, recommendation systems are getting more and more sophisticated and use new means in order to assess which product would fit each customer, so I do not really doubt that, if it is not already the case, marketers and developers will soon find a way to sell even the niche products to the customers that would be interested by them, and who would never had been aware of their existence without the RS.
A bit scary… We’ll have the opportunity to discuss the issue of privacy in a coming lecture.
Joseph Abi-Khalil
A recommender system learns from a customer and recommends products that she will find most valuable from among the available products. In this comment, I took the Spotify case to show how the recommendation system has evolved and that it could be based on other than historical data and also try to show if this new technique would make the… Read more
A recommender system learns from a customer and recommends products that she will find most valuable from among the available products. In this comment, I took the Spotify case to show how the recommendation system has evolved and that it could be based on other than historical data and also try to show if this new technique would make the tail longer or not.
Spotify is well known to rely on collaborative filtering approaches I.e. the process of filtering for information or patterns using techniques involving collaboration among multiple agents viewpoints, data sources, etc. This technique determines the user preference from historical usage data. Note that the collaborative filtering techniques do not use any of information concerning the items that are being recommended. However given this last characteristic, popular items will be easier to recommend than unpopular items since there is more usage data available for them making recommendation systems predictable. Moreover, the heterogeneity of content with similar usage patterns i.e. having an original song and a remix song from the same artist and having the recommendation system only taking the original song narrowing the choice of the system. However, the most important problem with the collaborative filtering technique is given that new and unpopular songs don´t have any usage data to be analyzed they will not be recommended this Is called the cold-start problem.
Thus, given these different difficulties, Spotify proposed another approach based on the content information itself. The company is trying to integrate other sources of information into their recommendation scope to tackle the problems explained previously. The other source of information could be: tags, artists, lyrics or reviews from the web and the audio signal itself. The audio signal could be the genre of the music or the instruments used. Engineers at Spotify wanted to develop recommendation systems out of audio signals. In order to do so, they made a regression model to predict latent representations of songs that were obtained from a collaborative filtering model to actually predict the representation of songs that had no historical usage data. We will not cover the details of the regression in this article but they could be found in the ´´References´´section. The result of this technique, was a built up of different playlists based on the instruments used in songs, it filtered also songs that had harmonic singer voices. This new algorithm made the platform recommend new and unpopular music which wasn´t the case for the collaborative filtering technique, making Spotify able to recommend the right music to the right public.
In conclusion, we can clearly see that these new techniques would make the long tail even taller by having new kind of products recommended to the public and by lowering the probability of listening to the same songs over and over again. This technique could also be used to other products like books but of course the variables taken into consideration would be different.
Very interesting: Where economics and computer science meet.
Valencia Katanga
In the years following Anderson theory of the long tail the literature on the subject seemed to agree that the success of retailers such as Amazon.com and Netflix is largely attributed to the “ long tail” phenomenon. Indeed niche products, that are not available at limited-inventory competitors, generated a significant fraction of total revenue in aggregate. However most of… Read more
In the years following Anderson theory of the long tail the literature on the subject seemed to agree that the success of retailers such as Amazon.com and Netflix is largely attributed to the “ long tail” phenomenon. Indeed niche products, that are not available at limited-inventory competitors, generated a significant fraction of total revenue in aggregate. However most of existing recommender systems were not able to recommend tail products due to the data sparsity issue. Based on that realization interested parties then focused on developing effective algorithms to recommend long tail items.
Furthermore, in his 2010 book “ Music Recommandation and Discovery”, Oscar Celma assessed that Anderson’s success theory may have been a failure but that his core idea still is an interesting way to explain the web-provoked changes, in terms of the availability of all kind of products (from hits to niches). Later he states that a recommender should focus on promoting the tail of the curve by providing relevant, personalised novel recommendations to its users. It would imply smoothly interconnecting the head and mid regions with the tail, so the recommendations can drive interest from one to the other.
In 2014 more and more tend to think that this long tail theory might have been an utopia. Daniel Kaplan said “The long tail effect is so small that it can cause no change in the market and has no impact on cultural diversity”. Some even develop contradictory theories for instance the superstar effect where blockbusters sell more than ever. Thierry Crouzet went as far as saying : ‘We are witnessing the opposite phenomenon. The curve has certainly lengthened, but it has widened enormously. It has some titles that sell a lot, and all others who are fighting for crumbs.”
In her book “ Blockbusters” Anita Elberse also shows that even companies supporters of the long tail have changed their tune. On one hand, Amazon seeks above all to subcontract to other players, and is looking with Kindle to set foot in the editing and production. On the other hand, now that Netflix business model is migrating towards online video ( Netflix used to primarily send DVDs at home at the time of Anderson’s interview), the company is obliged to secure the license of his films where previously it only had to buy DVDs. However, these licenses become increasingly expensive. Netflix evolved into a studio and invests hundreds of millions dollars in the production and promotion of its own shows like “House of Cards” or “Orange is the new black” buying into the blockbusters theory. Youtube slowly seems to follow the same path.
So my conclusion would be that even if the theory of the long tail is valid it doesn’t pass the test of reality. They are economic factors that make its realisation impossible. Another thing is that we are constantly faced with too much information, too much choices on the web and it can be a source of anxiety. So bestsellerization and blockbusters are what work right now as they also answer the need for easy access to novelty.
The idea of sharing information between each other is not something quite new. Take for instance the caveman; they already used information from others. One of them was hungry and saw a nice red fruit. His first thought is that he wants to eat it but it might be poison. As you are not as hungry as your fellow, you… Read more
The idea of sharing information between each other is not something quite new. Take for instance the caveman; they already used information from others. One of them was hungry and saw a nice red fruit. His first thought is that he wants to eat it but it might be poison. As you are not as hungry as your fellow, you will let him taste first and see if it is safe. Indeed if he survive you will get the information that the fruit is eatable and you will start to do the same. Thus, you learned the lesson without having to eat the fruit yourself. Moreover, critics of movies or books have existed for a long time. Many of us like to have others’ opinion to have an idea of what will be good or bad.
Nowadays, technology has evolved and it allows us to share and communicate our opinion, our preferences and our tastes with many more people. In recent years, recommender systems appeared in a large panel of application (Google, Amazon, YouTube, etc). E-commerce applications have found this system very useful in order to deal with the information overload problem. However, generating recommendations needs to be improved. Indeed, it suffers from data sparsity and cold start. The first issue concerns the difficulty to collect enough reliable data as the active users rates a small portion of items. Plus, the cold start refers to the difficulty of finding the right information to please the cold users as they rate a small number of items. Few solutions are developed in the link below.
Relying on personalized information, many Internet users are making important decisions. Most of the time those people are not even aware that the information to them. Nevertheless, it is still hard to find good update on the positive and negative effect of recommendation system. Knowing these consequence will help to minimize the negative ones and maximize the positive ones.
Comments for Do recommendation systems make the ‘tail’ longer or shorter?
Read more
The long tail theory states that there is a shift in today’s culture and economy from a relatively small number of mainstream products called “hits” towards a huge number of niches. The e-commerce has the tremendous advantage of the so called infinite shelf space giving less popular products the possibility to arise and attract consumers as distribution costs especially shelf space costs are disappearing. Niche products have the ability to better target interests and thus, are said to represent a considering competition for the hits. (1)
In the context of the growing digital market, consumers are facing an unlimited range of products which why is deciding on one product often represents an overwhelming task for the consumer. This is where recommendation systems have come into existence. Indeed, recommendation systems, often based upon past purchases, ease the product search for the consumer as they help him reducing its choice. Moreover, they reinforce the customer’s engagement towards the brand or the firm by offering personalized product information and suggestions. (2)
In the following part, we will discuss if recommendation systems lead to an increase in sales in the long tail and thus, support the theory of the long tail, or not.
To start with, in the context of the digital area, we are moving away from a world of scarcity to a world of abundance. As stated by Anderson’s theory of the the long tail, the digital market offers unlimited choice for the consumer as there is space for everyone. Thus, as everyone’s taste departs at some point from the mainstream, the online retail and distribution offers the consumer the possibility to discover new products, niches, related to his specific interests. Besides this, Anderson claims that there is a strong demand for these niche products. (3).
However, recent studies have observed that recommendation systems do not promote niche products. Rather than boosting sales diversity, recommendation systems lead to an enhancement of the best selling products. In fact, recommendation system do push consumers towards new products, but they often push different consumers towards the same new product. This results in a richer-get-richer effect for popular products and unpopular products often remain undiscovered. (4)
The previous argument is affirmed by Marcelllo Vena who states that the concentration of the digital market hinders diversity. Indeed, the market share of the global players is continuously growing leaving the small and independent retailers with only a negligible part. Hence, the more a market is concentrated, the more it tends to promote best selling products. Recommendation system seem to follow this pattern as they often suggest items that are the most shared or the most sold favouring sales rather than diversity. Although Vena’s study is mainly focused on the digital book market, we can easily understand the concept behind and apply it to other sectors. (5)
Following these studies, recommendation systems shorten rather than lengthen the long tail.
To conclude, it seems clear that recommendation systems contribute to an increase in the total sales. However, opinions differ regarding whether recommendation system enhance sales diversity or strengthen the purchase of best selling products. Therefore, the impact of recommendation systems on the long tail can be evaluated as limited.
As far as I am concerned, Anderson’s theory of the long tail should be considered with precaution. Indeed, being a consumer in the digital market myself and using reviews for my purchases, I believe that recommendations have made me discovered items that otherwise I would not have found, but I have also observed that the same items are proposed again and again which confirms the richer-get-richer effect.
References:
Show less(1) http://www.longtail.com/about.html
(2) http://www.cs.umd.edu/~samir/498/schafer01ecommerce.pdf
(3) http://www.wired.com/2004/10/tail/
(4) http://vldb.org/pvldb/vol5/p896_hongzhiyin_vldb2012.pdf
(5) http://www.digitalbookworld.com/2014/the-lost-tail-the-myth-of-book-publishings-long-tail/
Read more
Along with the digital evolution, new phenomenons have arisen. This is the case for the so called ‘Long tail theory’ by Chris Anderson. This theory focuses on the strategy of selling a broad spectrum of products, each in small quantities.
It is commonly known that shops have a limited offer and prefer thus selling best sellers or hits as they face the issue of having a limited stock. Nevertheless, digital markets make it nowadays possible for online retailers to offer unlimited choice as their stock is virtual. Therefore, C. Anderson states that the easier the disappearance of physical shelf place is, the higher profits for unpopular goods will be. As a result, niche products’ profits can surpass mainstream products’ profits by satisfying narrow interests. In other words, everything is available on online selling platforms and unpopular products can remain in the tail forever with a chance to become popular one day. (2)
Bearing in mind online retailers have unlimited offer, they need to assist the customer in its researches by means enabling him to discover and explore that extra available data. This is the reason why they have set up recommender systems putting forward suggestions usually based on predictions of customers’ interests, past history of purchases, product searches or products’ rating.
The question we should ask ourselves is if recommendation systems rest on the fact that the success of business is to sell less in quantity but more in diversity ?
From a customer’s point of view, I do not have any idea of how large the available offer of the online retailer is. Therefore, thanks to the recommendation system, I will be suggested movies, songs or books I had never heard about before. Indeed, as a human being I do not have the knowledge of the broad offer of the sector. Up to this point, I would tend to state that recommendation systems insists on diversification and makes thus the tail longer.
However, we can notice that those suggestions are backed up by ratings of customers or number of views. Consequently, although I was not aware of the products’ existence, the product was already more or less popular on the platform. Therefore, I see in the recommendation system a kind of vicious spiral leading to the fact that we finally end up watching the same movies, listening to the same music or reading the same book. Indeed, we contribute to the fame of the same products being influenced by the suggestions based on certain factors.
Studies have observed that our economies are not as diverse as C. Anderson would affirm. But, what are the factors supporting that statement ?
According to Marcello Vena, the digital market concentration is to blame for the lack of diversity of sales in the book sector; big worldwide actors are increasing their market share becoming numeric monopolies and making it difficult for independent retailers. Therefore, as the market share for monopolies increase, the one of independent retailers decrease.(4)
Besides this, recommender systems mislead the customer by favouring the chances of future purchases instead of the discovery. (5) This phenomenon is understandable as retailers’ first aim is to boost their sales. Consequently, they use the recommendation tool to seek their interest.
For instance, in the book industry, e-commerce platform will tend to suggest another book of an author the customer had already bought from beforehand. In fact, the customer is more likely to like it and this will undeniably boost the volume of sales.
In the music industry, it has been observed that only 3 millions out of the 13 millions available-for-sale titles are being purchased. Besides this, 3% of the sold title represent 80% of the revenue. (6)
Moreover, another striking fact is that recommendations are usually based on social facts and so, it is always the better rated or most liked product that will be recommended. (5) Consequently, popular products being are first suggested to the customer. A s a result, it is difficult to state that recommendation tools lead to a higher diversity.
To cut a long story short, I would state that the diversity of the offer does not absolutely lead to the diversity of purchased products. In fact, recommendation systems bring forward products based on algorithms leading to the reinforcement and enhancement of hits’ sales. Therefore, I believe the tail becomes shorter due to recommendation systems. However, if e-commerce platform’s first aim was not driven by profit, the tail could become longer.
Sources
(1) https://hbr.org/2008/06/debating-the-long-tail
Show less(2) http://www.longtail.com/about.html
(3) http://infolab.stanford.edu/~ullman/mmds/ch9.pdf
(4) http://www.digitalbookworld.com/2014/the-lost-tail-the-myth-of-book-publishings-long-tail/
(5) http://lafeuille.blog.lemonde.fr/2014/06/24/pourquoi-la-longue-traine-ne-marche-pas/
(6) http://www.internetactu.net/2009/01/22/que-faire-de-la-longue-traine/
Read more
It is no big news that we’ve entered a new economical era. We’re two feet well within the digital era and this has a ton of economical consequences. This paper focuses on recommendation systems. Those may look like a simple add-on of websites and online services but they have actually deep consequences.
One aspect to keep in mind is the current change of our culture. Our hit-driven culture is slowly but surely shifting toward a niche-driven culture, where the trend is to like what’s not trendy, as paradoxical as it may sound. This is of course in direct link with the Long-Tail theory. Whereas, earlier, people had access only to certain amount of big hits, the numerous online platforms gives them access to a much bigger market, allowing them to purchase more of the less popular items/articles/hits/… . This new accessibility combined with the change of culture have for a consequence that people start to buy more of the previously unknown items (the niches) and less of the big hits, making in this sense the tail longer. As studies show that recommendation systems increase the sales of the niches, it would seem like those systems are working in this direction as well. But a much more difficult question to answer is have they created or pioneered this trend ? If I was to give my opinion, I would say no. I don’t see such systems being able to influence human behaviors that much. I’d rather say that recommendation systems were set up in order to answer to this change in our culture, but that’s just my opinion.
But we can also find arguments to say that recommendation systems don’t make the tail longer but rather smaller. The current generation is composed of three big families : content-based, collaborative and hybrid approaches. Those systems still have a few limitations. I would name for example the problem of “over-specialization” for content-based recommendation. This means that systems can’t offer anything else than articles similar to those already bought/rated, preventing therefore the diversification of the user. In this way, we could say that recommendation systems are actually making the tail shorter, as they confine users in the market they already know and prevent them from going to something new (and therefore making the tail longer). As far as collaborative systems are concerned, we can name the “new item” problem. New articles are added regularly on most platforms. Collaborative systems base their recommendations on users’ ratings. Therefore, a new item won’t appear in the recommendation until a certain (often large) amount of users have rated it. In this way, those systems prevent new items from being known by only recommending the already popular items. This could also be seen as a “tail-shortening” characteristic. Of course, those problems are being addressed by the various platforms and new improvements such as multi-criteria ratings, incorporation of contextual information, etc. are being set up.
To conclude my comment I would like to give a little thought to what’s going to happen next. According to the Harvard Business Review, the tail does become longer and longer but what’s interesting is that it’s also becoming flatter. This means that even though the array of products being sold is increasing, the demand for those less popular items remains low. They also predict that there will be less blockbusters but that those will be bigger than nowadays. This could be kind of contradictory to trend of our culture shifting from the hit-driven to the niche-driven. I’m therefore wondering what the tail will look like 10 years from now ? Will we see the niche keep on growing, transforming the tail into a bulker one ? Or will the blockbusters capture an even bigger part of the market, leaving a tail extremely long but flat ?
Sources :
http://infolab.stanford.edu/~ullman/mmds/ch9.pdf
Show lesshttp://homepages.dcc.ufmg.br/~nivio/cursos/ri13/sources/recommender-systems-survey-2005.pdf
https://hbr.org/2008/07/should-you-invest-in-the-long-tail
http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.421.1833&rep=rep1&type=pdf
Read more
I personally have some interrogations about recommendations and the long tail theory.
Firstly, the big theory is that recommendations are made about previous choices of the consumers. But how to be sure that big firms don’t pay the website to be in the recommendations? We see this kind of thing for the Search Engine Advertising, which are “advertisements shown on search engines pages when specific keywords are typed” (1). The process explained in the article “Do recommendation systems make the ‘tail’ longer or shorter? “ is based on this “free and magic” peace of information but there is written nowhere that this information is truthful. And thus the tail will not be taller because the biggest companies will pay to have their mainstream products in the recommendations.
Secondly, if we agree that these recommendations are really previous choices, I am not sure that it will really enlarge the tail because some choices are logical. Indeed, if I am buying the first “Game of thrones” book, I will probably look after the second one and so the choices for these kinds of saga are distorted. The same reasoning is valid for the people who buy book (or Dvd and so on) from the same author or actor. And the principle to “enlarge the tail” is here not meet because there is no surprised (2).
Thridly, some companies have difficulties to have a database large enough and thus they have to buy one (so it is not a base on previous choices) or they ask consumers to rank some products on a “top 10”(2). But I think it will not enlarge the base. Let’s take an example and before buy a movie online you have to make “a top 10 of your favorite movies”. You will do it quickly because you don’t have time to waste and you will write the first 10 movies you have in minds which are most of the time really mainstream and that have connections between them. And there is a difference between products I like or book I really bought. So the database will not be fair and the tail will not be very taller.
In conclusion, there are some interrogations but if the data base is true, based on previous choices and not distorted, we can say that the tail is enlarged.
(1) http://onlinemarketingin60minutes.com/traffic-sea-search-engine-advertising/
Show less(2) http://www.forbes.com/sites/lutzfinger/2014/09/02/recommendation-engines-the-reason-why-we-love-big-data/#4a00d554218e
Read more
Nowadays with the growth of online retailers with less production costs, we have the possibility to buy a wider range of products. The traditional retail was focused on hits and blockbusters because of the limited storage space.
The theory of Long Tail is the strategy of selling a large amount of different small quantities of unique products (the tail of the demand curve) additionally of selling popular products (the head). (1) & (2)
Relating to the online retailers, lots of recommendation systems have emerged. Filters could help to provide products who better reflect the customer’s needs. If the customers could easily find new products they like using recommendation system, this could lead to an increase of sales in the tail of the demand curve.
But what is exactly the effect of these filters on the general shape of the curve?
Some studies such as these of Bart Van Looy and Annelies Geerts from the KU Leuven argue that the curve becomes longer but flatter. (3) The blockbusters products are still selling but in smaller quantities than before, they are less popular.
If we are shifting from a retailing of hits to niche products, does it mean that the total demand would be bigger?
In my opinion, the customer’s satisfaction increase with filters and they will therefore be more inclined to buy more items. Besides, knowing some hits will still remain, they could lead to an increase of the sales of similar unknown products with the recommandations like it was the case with the Simpson’s book “Touching the Void”. (4)
Moreover, it has been demonstrated that consumers in the movie sector who rent obscure movies are more likely to be heaviest customers than those who focus on blockbusters. (5) The study shows that they rented 50 movies against 20 for those chosen hits products in the six-month period under review.
In conclusion, presenting a large choice of products in addition to performant recommendation systems will lead to more sales and more satisfaction from the customers.
(1) http://www.longtail.com/about.html
Show less(2) https://en.wikipedia.org/wiki/Long_tail#Chris_Anderson_and_Clay_Shirky
(3) http://www.plan-c.eu/imadedocs/4_C_IncentimLongTail.pdf
(4) http://www.wired.com/2004/10/tail/
(5)https://hbr.org/2008/07/should-you-invest-in-the-long-tail/ar/1
Read more
There is no doubt that recommendation systems can and have had very successful results for companies like amazon and Netflix, there ability to increase the size of and profit from long tail has been impressive. For Netflix the recommendation system serves to increase switching costs for there service, and Amazon is able to increase their long tail by recommending similar or linked products.
Recommendation systems are a natural result of the world that we now live in. the sheer quantity of information, media, products and services that are available online is staggering. So naturally people developed ways to reduce searching costs from the internet to bring the exact content you are searching for. These systems are becoming more and more relevant and being applied to more and more products.
Despite all the positive effects recommendation systems can have I thought it would be interesting to look at some potential negative effects from recommendation systems. Specifically, what social medias effect has been on the news and its role in creating Eco chambers (or information bubbles). Facebook news allows people to customize the news the see and while opposing views can still get through. However, people create there own Eco chambers based on the things they like and click on. A study from Facebook also found that “liberals tend to be connected to fewer friends who share conservative content than conservatives who tend to be linked to more friends who share liberal content”. And with this and confirmation bias these Eco chambers have done a lot to cause divided political views. While sites like Facebook are not specifically for news a PEW study found that 61% of millennials obtain their news from social media (Facebook and twitter). Obviously it is not Facebook’s intent to create these Eco chambers however it is a side effect of their services. It is important to note that there are many other problems with the news including sensationalism, lack of trust for media, the effects of citizen journalism, accuracy of different news sources, etc.
This effect has not gone unnoticed. The director of product at Facebook Mike Hudack had a rant on the state of the media recently with the quote “It’s hard to tell who’s to blame. But someone should fix this shit.”. While the effects of the long tail have been observed for many recommendation systems, social medias effect has been far more polarizing.
Overall I believe that effective recommendation systems for things that we want are important to sift through the vast amount of content that comes with a digital world. However, when we rely on a system for something we need like the news its easy to see how only seeing the news we want could cause some problems.
http://www.journalism.org/2015/06/01/millennials-political-news/
Show lesshttps://www.facebook.com/mhudack/posts/10152148792566194
http://www.pnas.org/content/113/3/554.full
http://www.vox.com/2016/4/12/11406334/social-media-echo-chamber
http://cn.cnstudiodev.com/uploads/document_attachment/attachment/681/science_facebook_filter_bubble_may2015.pdf
https://www.ted.com/talks/eli_pariser_beware_online_filter_bubbles?language=en
I suppose you mean ‘echo chambers’. That’s an interesting perspective, thanks.
Read more
Nowadays, it becomes obvious that a lot of sales platforms use the recommendation system which is certainly in their interest from an economical point of view. Nevertheless, it is easier for big platforms as Amazon and Netflix to deliver niche product thanks to their “unlimited inventory” as opposed to “limited-inventory” competitors. Amazon and Netflix benefit significantly from the long tail because those niche products, influenced by the recommendation system, that are present in the tail represent an important share of their sales.
Now, to answer the question whether the recommendation systems make the tail longer or shorter I would say that they make it shorter or unchanged.
Why?
For instance, if you’re looking for a specific music genre, even though the songs are unlimited in this genre, recommendation systems will make some niche songs more popular and put them into the head of the tail. Following this logic, the head of the tail will become longer or you may also say the tail is becoming broader and shorter. In this situation, the recommendation systems will make the tail shorter.
Following an other logic, recommendation systems might have the effect of “popularizing” niche songs while in the mean time, previously “popular” songs might join the end of the tail. In this situation, there would be a replacement effect. “End of the tail” songs would replace “head of the tail” songs which would make the situation unchanged.
From my point of view, those situations would arise because there is still a limit to recommendation systems. What I mean is that there’s some stability point where the tail cannot get any longer. A song has a life cycle. Most of the popular songs stay popular for several years, then end in the tail and finally “completely disappear”. It’s the same with clothing, movies and other products. In general, I think the tail has a tendency to keep its length even if from time to time, some “shocks” might modify the length.
REFERENCES:
Show lesshttp://vldb.org/pvldb/vol5/p896_hongzhiyin_vldb2012.pdf
http://www.longtail.com/about.html
Read more
As an introduction, I will remind the long tail theory.
In the chart (1), we can clearly see what is the long tail: Seen by Chris Anderson, the long tail includes the products that are less popular than the head. The fact is that the popular products are sold in large quantities where the niche products are sold in low quantities but consist of a large variety of products. The long tail theory says that the (niche) products in the long tail outnumber those ones in the head. (1)
Indeed, the Long Tail theory asserts that “the products with a low demand, or a low sales volume can collectively make up a market share that rivals or exceeds the relatively few current bestsellers and blockbusters, but only if the store or distribution channel is large enough.”. (2)
It seems to me that this long tail can be a potential market share that Internet made available with the fall of the costs of production and distribution.
As explained in the comment, the long tail is becoming thicker (and longer) as the niche products are doing better in the digital age and become more popular. Hence, those niche products are interesting and big retailers understood it.
A little example may illustrate the long tail theory: An Amazon employee said that “We sold more books today that didn’t sell at all yesterday than we sold today of all the books that did sell yesterday.” (3)
After this remind, we can ask: How the recommendation systems impact the long tail and the niche products?
There exist more and more recommendation systems. “The top more popular in this category” but also “People who bought this also bought…”. They appear on the big retailer website and on the mobile platform such as Google Play and Apple. That can be a problem because the recommendation systems are based on the sales and rating. Hence, those systems will redirect the customer to popular products again and again. As said in this article: “Because common recommendation systems are based on sales and ratings — for example, people who bought this also bought this — they’re unable to surface truly novel items that have not been discovered by many other people.”. (4)
I think that as long as customers will use the recommendation systems based on sales and ratings, they won’t likely discover niche products, as explained in this article: “This tends to create a “rich gets richer” effect for popular items, and it might also prevent consumers from finding better product matches because of this bias for items that have been purchased by others or that have been rated well by others.”.
However, a little further in the article, they made two conclusions: the first one is that however, the recommendation systems were supposed to help the niche products to obtain opportunities, recommendations systems doesn’t necessarily help to discover niche products
The other conclusion is that when a product with low rating is recommended, people may give the benefit of the doubt and give them more attention. (4)
References:
Show less(1) http://www.longtail.com/about.html
(2) http://www.investopedia.com/terms/l/long-tail.asp
(3) http://longtail.typepad.com/the_long_tail/2005/01/definitions_fin.html : Third comment
(4) http://knowledge.wharton.upenn.edu/article/recommended-for-you-how-well-does-personalized-marketing-work/ (December 2015)
Read more
At first glance, it may seem clear that recommendation systems foster niche markets. Indeed, we may think that it opens the gates between people and less known products, that it increases the chances of SMEs etc… It is partly true. Undeniably, thanks to the recommendation systems, people discover new items that they did not know before, they have access to more products and they can directly find products that perfectly match their tastes. For example, here is how the Google News’s recommendation system works: «The recommendation system builds profiles of user’s news interests based on user’s click behavior on the website. This system helps users to find new articles that are interesting to read” (1)
Show lessHowever, even if recommendation systems may help niche markets to be known, it is less obvious that those systems make the ‘tail’ longer. Effectively, it diversifies the recommendations for individuals, but if we aggregate the recommendations, the recommendation systems often drive people to the same items and thus there is no real diversity. This happens because “most of existing recommender systems struggle to recommend tail products because of the data sparsity issue”. Thus those systems only recommend popular items which creates a “rich gets richer” effect. (2) Users don’t really discover truly novel items that has not been searched or bought by many other people because common recommendation systems are based on sales and ratings. (3) To return to the example of online news reading, we have to be cautious because algorithms and big data used by recommendation systems often have bias. As a result we may miss the big picture of the news. (4)
To conclude, I think that recommendation systems are sure a great opportunity for niche markets but we must continue to improve their process, remove biases from algorithms and big data they use in order to really make the ‘tail’ longer.
(1)Liu, J., Pedersen, E., Dolan, P. (2010). 2010 International Conference on Intelligent User Interfaces.
(2)Yin, H., Cui, B., Li, J., Yao, J., Chen, C. (2012). Challenging the Long Tail Recommendation. Department of Computer Science & Key Lab of High Confidence Software Technologies (Ministry of Education), Peking University.
(3)Hosanagar, K. (2015). ‘Recommended for You’: How Well Does Personnalized Marketing Work? http://knowledge.wharton.upenn.edu/article/recommended-for-you-how-well-does-personalized-marketing-work/
(4)ibid.
Read more
Let’s start this case analysis by trying to make a clear definition of the longer tail “effect”: If all the goods from a market were normally distributed, the tail of the distribution will represent a niche market (sells unique goods in small quantities) when the center of the distribution will represent the main products (blockbuster, best-seller, …). In other words, the well-known products have a more powerful sales-force than the unknown-one: logic. But the interesting thing is that with the internet revolution and indirectly, the e-commerce, companies are not physically constrained anymore: they can put in plain view all the products they want. It doesn’t sound that significant, but back in the time, a bookseller for example, has no interest to a book which has not any guarantee of success in his store. Indeed, it is most likely that a very small amount people will buy it. Especially when the last book of a famous writer just came out. As he is physically constrained (limited capacity), he is more or less obliged to only propose mainstream books. This small example allows us to realize some important effects.
In today digital world, anyone can put anything anywhere on the internet and this includes big companies like Netflix and Amazon. They transformed this new power in a marketing tool which allows them to adapt the content of a website page according to the user’s interest (recommendation system). Therefore, nowadays, they are able to offer us a larger ranged of products, including the less known one which nobody is aware of. Economically speaking, on one side, consumers will be satisfied as they discover new products which are potentially going to interest them, and on the other side, selling these products is going to be more profitable for the niche market producers. Both consumer and producer welfare is enhanced. To me, it is definitely something crucial and important. This effect is lowering the normal distribution and rebalance the situation.
Furthermore, the recommendation system (personalized marketing) has evolved with an extension to the initial concept: the online reviews. The figures speak for themselves: 90% of consumers read online review before visiting a business, 88% of consumers trust online reviews as much as personal recommendations. Here again, we can see the strength of a personal recommendation system as a personalized marketing tool.
To conclude this discussion about the recommendation system, we can see that internet and the globalisation are making the long tail taller, increasing the numbers of available products on the market. Personally I think that this will progress again and again, and that marketers will never stop to innovate, in a good way I hope. This is a very interesting subject which is complex and has more impact than just a variation in quantity/price. I mean, we can question ourselves about the respect of privacy of companies using this tracking techniques. Most people using internet don’t even know that their data are collected. Anyway, it has a significant positive impact on consumers/supplier. In my opinion, I think this is a step forward for a better balanced economy where all economic players have more or less the same success chance. Ultimately, this further strengthens the competition instead that this time, it’s not just a harder fight between the main actors of the market…
References:
Su, N., Levina, N., & Ross, J. W. (2016). The long-tail strategy of IT outsourcing. MIT Sloan Management Review, 57(2), 81-89.
http://knowledge.wharton.upenn.edu/article/recommended-for-you-how-well-does-personalized-marketing-work/
Show lesshttp://www.business2community.com/infographics/impact-online-reviews-customers-buying-decisions-infographic-01280945#8IAPzyX6JW6rg88v.97
http://www.cs.umd.edu/~samir/498/Amazon-Recommendations.pdf
http://www.marketing-schools.org/types-of-marketing/long-tail-marketing.htmlV
Read more
Firstly, I will start by giving you some explanations about the Long Tail phenomenon and illustrate them through the example of Amazon. Then, I will talk to you about the recommendation systems and their link with the Long Tail theory.
Anderson’s Long Tail theory is illustrated by a decreasing curve on a graphic: popular brands/services/products on the vertical axis (called the “Head”) are distributed through traditional retailers. On the horizontal axis (called the Long Tail), an infinite array of smaller and niche brands are spread and distributed through the Internet. By using the example of Amazon, Anderson explains that more and more firms have understood that Pareto Principle – which states that 80% of the effects come from 20% of the efforts – is not always the best choice. Effectively, if everyone focused only on the 20% of the market representing 80% of the benefits, competition would increase and profit per market share become limited. On the contrary, focusing on the remaining 80% of the market which are neglected could be very lucrative if you can entirely incorporate it. It is sometimes more beneficial to have 20% of the market that to share 80% with the rest of the world.
This was what Amazon understood by offering products called “normal” (represented by the 80% of earnings) and on which they met strong competition, as well as products called “niche” (very specific and difficult products to find, representing 20% of profit from the market). Some researchers have confirmed that a huge quantity of sales come from unpopular books that were not available in brick-and-mortar stores. This leads us to believe that if Amazon only focused on generic products, it would not be as successful as it is now, simply because the competition between these products is too tough to differentiate efficiently. Therefore, we observe that consumers’ primary value of the Internet “comes from releasing new sources of value by providing access to products in the long tail”.
Actually, to determine whether a sales distribution has a long tail or not is the cost of inventory storage and distribution. Before Internet, it was difficult to reach niche markets: since storage and distribution costs are high, only the bestseller products were sold and competed for physical shelf space in physical retailers such as Walmart or Target. But nowadays, with the emergence of the e-commerce, it becomes economically viable to sell unpopular products. Effectively, by centralizing their warehouses, online providers such as Amazon or Netflix see their storage costs significantly decrease. Moreover, whether it is a popular or an unpopular movie, distribution costs remain the same. That way, it allows e-companies to stock a wider range of movies compared to traditional physical retailers. And effects are felt: Netflix for instance has observed that in aggregate, unpopular movies are rented more than the popular ones.
According to MIT Sloan Management Review article titled “From Niches to Riches: Anatomy of the Long Tail”, tools such as recommendation systems are allowing customers to find products outside their geographic area without being confined only to local markets. No perimeter on market demands exists anymore since Internet provides an unlimited selection of products. On the supply side, the long tail has possible effects for companies’ culture and politics. As a matter of fact, it gives them opportunities to introduce products in the niche category and encourage the diversification and innovation of new products.
Economists in an article from “The Economist” (2015) stated that two benefits of e-business were provided to consumers: prices would decrease and become uniform and the selection available to consumers would increase. But the latter statement seems not to be correct. For instance, a rare book is more valued by consumers who have troubles to find it in physical stores. “By posting its inventory online, all those readers scattered around the world are now added to the potential demand for the store’s copy. Higher demand translates into higher prices which clearly makes the bookseller better off. But so is the reader since without the Internet he would not have found the book.” It seems obvious that some customers will be willing to pay a lot if they finally find on the Internet a rare product they have researched for a long time. “Higher prices are therefore a sign that buyers are being better matched to books they want. […] On the contrary, benefits are less likely to hold for easy-to-find, commoditised products; online prices of popular, usually in-print, books were less dispersed and closer to offline prices.” Nevertheless, the success of this Long Tail approach shows that you need both ends of the curve. Effectively, bestsellers still matter in attracting consumers in the first place.
My opinion about recommendation systems (as many other tools such as search engines, sampling tools etc.) is that they play an important role in the growth of the niche markets. We can quote an extract coming from The theory of the Long Tail by Chris Anderson which illustrate quite well those systems: “For instance, the front screen of Rhapsody features Britney Spears, unsurprisingly. Next to the listings of her work is a box of “similar artists.” Among them is Pink. If you click on that and are pleased with what you hear, you may do the same for Pink’s similar artists, which include No Doubt. And on No Doubt’s page, the list includes a few “followers” and “influencers”, the last of which includes the Selecter, a 1980s ska band from Coventry, England. In three clicks, Rhapsody may have enticed a Britney Spears fan to try an album that can hardly be found in a record store”. Amazon and Netflix also do the same thing through other recommendation techniques (“Customers who bought this also bought …”). But with all the same goal: drive demand down the Long Tail.
Recommendation systems could be therefore very efficient for small businesses with a small budget that try to innovate, promote and launch new things.
SOURCES:
1. http://www.economist.com/news/finance-and-economics/21638142-consumers-reap-benefits-e-commerce-surprising-ways-hidden-long
Show less2. http://www.economist.com/node/12762429
3. http://www.julienrio.com/marketing/french/theorie-longue-traine
4. http://www.longtail.com/the_long_tail/
5. http://www.therobinreport.com/the-long-tail-theory/
6. https://en.wikipedia.org/wiki/Long_tail-Goodbye_Pareto_principle.2C_welcome_the_new_distribution
Read more
Introduction
Recommendation systems « identify recommendations autonomously for individual users based on past purchases and searches, and on other users’ behavior ». (1) These systems have become more and more common in recent years and I often face with these one while I’m visiting a panel of different applications such like music on Spotify, books on Amazon or movies on Netflix.
Toward the realization of this comment, I have focused myself on the music application called Spotify I’m using everyday in order to understand the evolution of the recommendation system and how this technique of recommendations could make the tail longer or shorter.
Collaborative filtering system
Spotify recommender system was mainly based on collaborative filtering techniques. These approaches are defined as follows « collaborative filtering is the process of filtering for information or patterns using techniques involving collaboration among multiple agents, viewpoints, data sources, etc. » (2) In other words, « the idea of collaborative filtering is to determine the users’ preferences from historical usage data » (3). For example if two users listen to largely the same set of songs, their tastes are probably similar.
Limits of the system
But this system seems to have certain limits. In fact, two main issues called « content-agnostic » problem and the « cold-start » problem were encountered by Spotify. The first one is met because the technique of collaborative filtering don’t use any kind of information about the items that are being recommended. Given this, « recommendation systems become predictable and boring in the sense that popular items will be much easier to recommend than unpopular items, as there is more usage data available for them ». (4) The second one, called « cold-start » problem, is encountered when songs are new or unpopular. In fact, in this case, there is no data to analyze and then the collaborative filtering system breaks down : recommendations don’t work for these songs. A last problem from the collaborative filtering system, which can be mentioned here, is the heterogeneity of content with similar usage patterns. In fact, when I’m listening a playlist or the entire album from an artist, these one may contain intro tracks, cover songs and remixes. The collaborative filtering algorithms are only coded to take the original song and won’t take all other forms of songs to draw recommendations.
The solution called « Deep Learning »
In order to solve these different issues, Spotify has had to take different actions to incorporate other sources of information into its recommendation system. Spotify found another approach called « the content-based recommendation ». In this system, the firm is taking other sources of information as tags, lyrics, text mined from the web (reviews, interviews,…) or the audio signal itself for example.
It’s precisely this last source of information (the audio signal) that Spotify began to analyze in order to predict what songs users will enjoy. The algorithm created by Spotify is able now to learn the specific content of the songs following different factors as specific pitches, chords, sounds, and others. With this, filters are set up to cluster songs in different categories following these last factors. For instance, if a new song comes on the platform or if this song comes from an artist that’s not popular, it would be hard to recommend that to users if Spotify is using more traditional methods because no one else is listening this song. But with this new method, the algorithm is relying on the acoustics of the song (for example), it can make better recommendations (lesser-known artists i.e.) for users.
Conclusion
As a conclusion, people in the past were discovering products which were already rather popular in the whole population. However, Spotify managed to overcome and resolve this problem. In fact, the firm created a new algorithm which allows them to recommend new and unpopular musics what wasn’t possible with collaborative filtering systems.
In general and based on the Spotify case, we can see that firms are in the process of finding systems which are able to recommend new products or unknown products to customers. In the past, one would have thought that recommendations systems were making the tail shorter. But in reality and thanks to new technologies, we see that the tail is becoming more and more longer…
References
(1) http://www.ibm.com/developerworks/library/os-recommender1/
Show less(2) https://en.wikipedia.org/wiki/Collaborative_filtering
(3) (4) http://benanne.github.io/2014/08/05/spotify-cnns.html
– http://www.wired.com/2004/10/tail/
– http://www.webopedia.com/TERM/R/recommender_systems.html
– https://www.washingtonpost.com/news/the-switch/wp/2016/03/11/spotify-might-be-secretly-stealing-your-taste-in-music-and-pitching-it-to-other-people/
– https://gigaom.com/2014/08/05/how-spotify-is-working-on-deep-learning-to-improve-playlists/
– http://thefuturesagency.com/2015/08/01/spotify-using-deep-learning-to-create-the-ultimate-personalised-playlist/
– http://observer.com/2015/05/spotifys-head-of-deep-learning-reveals-how-ai-is-changing-the-music-industry/
– https://www.macstories.net/linked/recommending-music-on-spotify-with-deep-learning/
– http://qz.com/571007/the-magic-that-makes-spotifys-discover-weekly-playlists-so-damn-good/
– http://fr.slideshare.net/MrChrisJohnson/interactive-recommender-systems-with-netflix-and-spotify/94-Step_4Radio_thumbs_feed_back
– http://dataconomy.com/spotify-used-deep-learning-recommend-songs-youll-love-never-knew-existed/
Thanks! Your comment should contribute to a ‘deep learning’ of the topic.
Read more
According to my research and in order to submit a comment relevant as possible, I decided to divide the content of my analysis into 3 parts, as follows:
1. Definition of the concept of “recommendation system” and areas of application:
Commonly used on e-commerce, social media and content-based websites, this information filtering technology analyses available data about the registered user’s profile in order to provide him suggestions in which he might be interested.
Initially, this recommender system was used by Amazon, over more than 10 years ago, when the core business of the company was essentially based on online bookstore. Their idea was revolutionary and intended to suggest books to customers, using the gathered data about their previous activity. But since then, even if this technology remains mainly used by online information goods retailers (books for Amazon, songs for Spotify, movies for Netflix, ect.), its scope has extended to diversified fields.
Indeed with his “jobs you might be interested in” section, “Linkedin” applies the available data about its members, such as background, locations, skills to adapted algorithms, in order to come up with suitable job suggestions. Furthermore, this technology even extends to the medical field, with for instance, the “knowledge-based dietary nutrition recommendations”.
2. From an analytical point of view:
Based on cutting edge algorithms, those recommender systems use a number of different technologies, which can be classified into 2 broad groups:
– Content-based systems also referred to as cognitive filtering, recommends items based on a comparison between the content of the items and a user profile. For instance, if a Netflix user has watched many cowboy movies, then recommend a movie classified in the database as having the “cowboy” genre
– Collaborative filtering systems recommend items based on similarity measures between users and/or items.
In fact, from a conceptual standpoint, those algorithms intend to assess and the unknown and not available rating value (representing the user’s preference for the item) with respect to a specific item the consumer hasn’t tried/doesn’t possess yet. This way, starting from the user’s profile or from similarities, a high predicted value would mean that the user could be really interested in the product in question.
3. Economic impact:
In order to introduce this part of my analysis, I wanted to highlight the fact that those recommendation systems were clearly profitable for online information goods retailers. The “Netflix challenge” which occurred a couple of years ago is a good example of this “profitability”. Indeed, in view of the great opportunity it represented, the company offered a prize of $1,000,000 to the first person or team implementing an algorithm beating their own recommendation one of 10%.
This above-mentioned opportunity could be explained by the “Long Tail theory”.
Initially encouraged by the emergence of the e-commerce, this theory is based on the fact that traditional physical delivery systems are limited by the resources they have at hand. For instance, physical bookstores, restricted by their shelf space, can show only a small fraction of all the choices that exists, while Amazon proposes millions of them. Therefore, it allow such online information goods retailers to sell small quantities of special goods, located in niche markets and that would not be available in the shelves of traditional physical retailers. Those aggregated quantities represent the so-called “Long Tail” on the graph on which items are ordered on the horizontal axis, according to their popularity (most popular items on the left side and niches market on the right side of the decreasing curve). This way physical retailers offer only the most popular items while on-line institutions can provide the entire range of items (the tail as well as the popular items).
Considering the size of the “Long Tail”, composed by the aggregated niche markets, and the size of market dedicated to popular products, a remaining question concerns the generated impact of such a system on sales and profit. Does it simply increase the benefits through adding a share of special products to the amount of popular items sold, or does it reallocate demands from those popular items to the niche markets?
In my opinion, even if a share of the items previously bought will switch from the popular side to more specific products, this technology still increases the number of product sold. This way it remains profitable for online retailers through increasing the accessibility to specific items and therefore through inciting users to buy more.
Sources:
http://link.springer.com/article/10.1007/s10799-015-0218-4
Show lesshttp://whatis.techtarget.com/definition/recommendation-engine
http://infolab.stanford.edu/~ullman/mmds/ch9.pdf
http://www.inet.econ.cam.ac.uk/working-paper-pdfs/wp1520.pdf
http://www.economist.com/blogs/freeexchange/2013/03/utilities
Read more
Personally when I go on the internet looking for clothes, shoes, books, any item I need, I appreciate the recommendation system of some websites. It may lead you to discover new products, to which you would never have thought, related to the product you are looking for. For example when you look for clothes, it can give you some ideas of others clothes you can wear with the piece of cloth you want to buy. And this make sure customers’ search costs will decrease as people will gain time while looking for the right piece they want. (1)
Recommendation leads to increase the purchases of niche products and so supports the theory of the long tail but recommendation systems cannot be only based on previous sales otherwise it will only highlight popular sales as limited sales won’t appear in the recommendations. And it will lead to a “rich gets richer” effect. (2)
Customers pay also attention to reviews of previous buyers. It helps increase the conversion rate of users. It has been shown that 63% of customers are more likely to make a purchase if there are reviews. (3) This will thus boost the sales and make the tail longer.
But this system remains really subjective as, like for reputation systems, it is based on other customers’ previous sales or on the type of product you already looked at. If you liked one product, you won’t especially like another product recommended by the website just because someone else also looked at it or bought it.
Although I find recommendation systems a great adviser, I don’t like remarketing which is based on your personal website visits or sales. I can be annoying being remarketed as, as soon as you leave a specific website, you will get an ad for the product you just looked for on every website you will visit afterwards. It is a great marketing tool as it was already shown that it can increase sales, but for the customer it is disturbing. Therefore, it needs to be controlled with publication rates. The brand can’t post an ad on every website you visit every day. It needs to be spread over time. Otherwise the customer will get angry about it.
(1)http://misrc.umn.edu/wise/2014_Papers/Strategic%20Product%20Recommendation%20WISE%202014.pdf
Show less(2)http://knowledge.wharton.upenn.edu/article/recommended-for-you-how-well-does-personalized-marketing-work/
(3)https://econsultancy.com/blog/9366-ecommerce-consumer-reviews-why-you-need-them-and-how-to-use-them/
Read more
Into the world of big data, recommendations systems are without disputes huge useful tools for the e-commerce in every industries, whatsoever in goods industry, like for music or clothes but also in services industry. These systems are so many effects on the customers and the e-tailers.
At the point where the customers prefer web sites with recommendations systems to make their shopping (1). Hosanagar K. and Lee D. (2014) say that with the long tail phenomenon the web site offer a large range of items and the recommendation systems help the customers by reducing the choice, with items based on their preferences. By using recommendation systems, the customers can be sure that their search costs will be reducing, they will find faster products that they want and like. This is on line with the personalization of the web and the need for speed of the society (3).
Moreover, by allowing to discover items towards which we didn’t go usually and by using filters more personalized, the recommendation systems allow to developed the consumption because people buy more than without these systems. They allow also to aim more the consumer segments. In fact, with products proposed and recommended, it attracts more customers with needs and cultures different (2).
For the e-tailers (the biggest who use it are Amazon and Netflix), recommenders are useful in the management of huge inventory present in the supply-side. Popular items ask an important part of the space of stocking to meet demand and the less popular items are in competition for the limited space remaining (2).
There are also others positive impacts: on the sales volumes and the revenue. Hosanagar K. and Lee D. (2014) say that “according a recent study, the recommendation engines were reported to increase revenue up to 300 percent, conversion rates up to 150 percent, and consumers’ average order value up to 50 percent” (1).
In addition to have impacts on the sales of the company, recommenders help also to provide a better image of the brand, offer a feeling of trust at the customers (3). This enhances also the online experience provides to customers.
Sources:
Show less(1) Hosanagar K. and Lee D. (2014). Impact of recommender systems on sales volume and diversity, Thirty Fifth International Conference on Information Systems. On line: http://aisel.aisnet.org/cgi/viewcontent.cgi?article=1349&context=icis2014
(2) http://www.marketing-schools.org/types-of-marketing/long-tail-marketing.html#link2
(3) http://ijcai13.org/files/tutorial_slides/td3.pdf
Read more
The long tail theory states that the volume of sales of few bestselling products are comparable to the volume of the many products left sold in small quantities (often niche products). The tail symbolize then these leftover products sold and available in reduced quantities.
Due to the recommendation system in the digital world, consumers are easily set in relation with very specialized products which fulfill their needs with more precision. The key question is thus to know if this system makes the tail longer or shorter than before.
My first thought is that the recommendation system makes this tail shorter. On internet, there is (almost) no capacity restrictions anymore. Therefore, much more products are available and this will make the tail eventually longer. But on the other hand, due the more global availability of the goods, the tail is also thicker as more people will buy them. Globally, the tail will become stronger than before, with more volume traded through internet.
With the recommendation systems, people will have more choice than before, according to their needs. They will thus be more likely to buy from this recommendations. This creates a certain loop, as the recommendation system becomes “smarter” by analyzing the buyers’ behaviors.
The recommendation system will present to the next customers more accurate data fulfilling their needs and the loop will grow bigger and bigger. Eventually, what was categorized as “niche products” will become more popular and become bestseller with the others “older” bestselling products.
This will make the tail smaller and thicker, and the head of the sales, the bestsellers, will also become thicker and a bit longer, according to me. The wholes sales will then become more even, with less volume differentiation between the head and the tail.
References:
ScienceDirect’s Article Recommender gets smarter and casts a wider net, https://www.elsevier.com/connect/sciencedirects-article-recommender-gets-smarter-and-casts-a-wider-net
Show lessDigging a Smarter Crowd, https://www.technologyreview.com/s/410424/digging-a-smarter-crowd/
Long Tail blog, http://www.longtail.com
Read more
On the whole, I believe it is important to consider the question of whether internet recommendation systems is making the media content tails longer or shorter from multiple angles and under the lens of multiple different media types. On the whole I believe it is undeniable that complex algorithm based recommendation systems certainly do expose users to different media sources they might not otherwise be aware of, however, the design of these systems can vary greatly in their effectiveness in lengthening out the tail.
Within music, it seems that the most relevant music delivery services (Music Streaming) are noticing the consumer demand for lengthening the tails of the content they can offer and are reorganizing their recommendation systems to accommodate this. The most prominent firm utilizing this is Spotify. This service, dubbed “Fresh Finds” seeks to complement Spotify’s current popularity based recommendation system (Shortens the tails), discover weekly. The key difference with Fresh Finds intends to populate its suggested playlists by focusing on lesser known artists by scouring 50,000 blogs, critics and other taste influences. This should have the effect of greatly widening the curve for many artists as the selection is going to be based off of little known artists that don’t have the large marketing budgets that a major record label could provide. This also implies that Spotify will need to work with the artists that this newly designed discovery system uncovers to upload their work to the platform, allowing the work to be consumed by new, much larger audiences and indeed monetized. [1]
With that being said, there is no credible evidence to suggest that the proliferation of online music content delivery systems has not helped out the lesser known or wider tail arists, in fact, quite the opposite. According to The Economist, Independent Record Labels (“Indie”)’s share of the Bilboard top 200 albums has grown to 35% in 2010 from 13% in 2001. This lends credence to the popularly romanticized ideal of the internet democratizing the creative media space.[2]
Another key content viewing platform, Youtube, has been at the center of a lot of controversy over whether the algorithms that determine which videos get selected to be “Reccomended” or on the front page is unfair to the smaller or lesser known content creators. The current system uses a time spent viewing weighted approach to identifying the best, most relevant content. [3] this problem has led to popular, early adopter youtube content creator Philip Defranco to state that he does not believe it would be possible for him to recreate the sucess of his channels in the current environment as it is much, much harder for great content to get noticed under the current system.
1) https://www.washingtonpost.com/news/the-switch/wp/2016/03/11/spotify-might-be-secretly-stealing-your-taste-in-music-and-pitching-it-to-other-people/
Show less2)http://www.economist.com/news/finance-and-economics/21638142-consumers-reap-benefits-e-commerce-surprising-ways-hidden-long
3)http://www.webpronews.com/youtube-recommends-videos-based-on-engagement-not-just-clicks-2012-03/
Read more
To know if the Anderson’s theory of the long tail is well reflecting the facts, I would suggest we have an “a posteriori” look at the reality of business today.
The theory implies that the consumers, thanks to the internet and the recommendation systems it offers, will find a new way of consuming: retailers should now offer a larger number of less mainstream products (and focus more on the tail of the demand curve, which should become longer and thicker) where they used to offer a few “blockbuster” products (they used to focus on the left/head part of the demand function). However, is it really the case today ?
Of course, the recommendation systems could give the chance to smaller production as the internet provides indeed an unlimited space on the “shelves” to offer less mainstream products. Let’s take the example of Amazon VS library : for a traditional library, their total number of shelves can be seen as a rent cost so, to maximize their profits, they have to offer as many best-sellers as they can to sell as much as possible. On the other side, Amazon does not have this problem of room size and they can offer thousands of books, CD’s,… that a traditional library could never hold in one room.
Besides, the rising of the internet and recommendation systems can also give the chance to good products to be known even if they have not enough budget to make an advertising campaign as the internet can be seen as a free showcase.
These are the arguments provided by the long tail theory advocates and by companies (like Netflix or Amazon) coming from this “long tail era”.
However, I personally share the opinion of Harvard Professor Anita Elberse, which is very well explained in the articles [1][2] linked below. I think that many benefits of the recommendation systems are parts of an utopia.
As showed in these articles, we are indeed more than ever in a blockbusters era where taking risks by investing a lot of money in one project is often the best way to success and where being wise by spreading the investment is counter-intuitively synonym of risk taking. As an example of this blockbusters’ supremacy, the articles compare two different strategies of these last years : Warner Bros VS NBC (TV channel).
The Warner Bros philosophy is a legacy of the strategy which was first used by the studio producing “Jaws” in 1976 : money spending by millions in advertising and trying to have as many movie theater partners as possible at the beginning to avoid word-of-mouth, so, “take the money and run”, even if the movie is bad. This strategy is obviously successful as we can see with the huge success of some of the giant productions of these last years (Harry Potter saga, Batman, Ocean Eleven,…).
On the other side, we have the NBC vision. A few years ago, they decided to use the opposite strategy by avoiding risks and produce series with not mainstream actors, trying to spend less money on ONE production to be able to produce other series and multiply chances of success. Despite a few successful productions, the decline of audience of the channel can be seen as a proof that this strategy was not the winning one.
A few other examples are given by A. Elberse such as having stars in a sport team (to sell more tickets, whatever the result of the match) or the Apple products (which can be seen as stars or blockbusters).
So, today we can say that we are still in a blockbusters’ dictatorship era. It is obviously the “winner takes all” marketing strategy which is working and companies that used to be heirs of the long tail philosophy (and used to be very proud of that status!) such as Netflix have now become producers to follow the laws of business. They have to promote their series (e.g. House of cars) with mainstream actors (Kevin Spacey) and by using their recommendation systems to present their own productions.
Finally, to conclude, I would like to take another example, coming from my personal experience. As a user of YouTube, I have never seen an independent video with only a few hundred views in the “Recommended for you” section. The algorithm clearly focuses on VEVO channels (which are partially YouTube/Google produced channels [3]) and mainstream videos with already millions of views.
As a consumer, I think that these mechanisms are a brake to inventiveness and diversity as they force studios and artists to be part of a recurrent successful model in the name of business.
REFERENCES :
Show less[1] http://www.slate.fr/tribune/84585/longue-traine-blockbusters
[2] http://harvardmagazine.com/2014/01/the-way-of-the-blockbuster
[3] https://fr.wikipedia.org/wiki/Vevo
Read more
The long tail theory focuses on the market of niche products: low volume sales of a large number of product . Thanks to the internet channel, the theory argues that it is possible to target this market more efficiently, so as to compete with the current sales of few blockbuster and even to sell more of these niche products (Anderson 2015).
The “hit-driven business model” has been challenged by the increased competition between firms. Indeed, the internet offers the possibility for online stores to offer such a large range of products that it becomes nearly impossible for consumers to chose among them, and more importantly to know about every product. Moreover, the rise of internet has changes the consumer behavior: consumers want a dialogue rather than just an overload of information and want to build a trust-relationship with firms. “Engagement” is key for e-marketing. Firms have then to develop strategies to get the consumer’s attention.
In the growing e-commerce market, one of these strategies is the online recommendation system. “We are leaving the age of information and entering the age of recommendation” said Chris Anderson. Recommendation systems have not only a direct impact on sales (by providing a review and so an incentive to buy or not), but more importantly an indirect impact on it, thanks to an increase un consumer satisfaction. The system responds to what the current consumer wants, by increasing brand loyalty and consumer engagement trough personalized offers which build this trust-relationship mentioned before. A personalized service is possible because people that are leaving reviews often use key words that other consumer will use too when searching for the same kind of product.
A recent study (2015) proves that in the pre-purchase phase, 90% of consumers read reviews,
88% of consumers trust online reviews as much as personal recommendations, 72% say that positive reviews make them trust the business more, 31% spend a larger amount on good reviewed sites and 86% will hesitate to purchase from bad reviewed businesses.
While recommendations clearly are important to consumers, a study of the University of Wharton (December 2015) shows empirically that “because common recommendation systems are based on sales and ratings — for example, people who bought this also bought this — they’re unable to surface truly novel items that have not been discovered by many other people. This tends to create a “rich gets richer” effect for popular items, and it might also prevent consumers from finding better product matches because of this bias for items that have been purchased by others or that have been rated well by others. ». This is because even if recommendations let us find new products, it does with all consumers so that the “new” products are the same for everyone. So the total effect isn’t an increase in diversity of purchases, but an expansion of best-seller purchases. So, recommendations partly “benefit the long tail” but don’t permit us to discover real novel niche products. This analysis confirms the argument of Belleflamme and Peitz.
Another study (University of Cambridge, November 2015) « estimates the effect of product saliency, affected via recommendation sets, on user choice in online markets. We find a sharp and robust 6% increase in the sales of a product when it is recommended by a highly popular new product. This effect is however short-lived, lasting for approximately only four days. On average the daily increase in sales attributable to saliency is around 5%. We also find that products recommended in smaller sets experience larger effects of saliency as they have to compete less for user attention. » « We show that such recommendations can temporarily affect the sales and appeal for products by shaping and expanding consumers’ consideration sets. »
Finally, a third study (2016) goes further and argues that recommendation systems (word of mouth) and the concentration of sales depend on consumer preferences. « When consumers with more prevalent preferences are the main beneficiaries of these improvements, the concentration of sales increases, generating a “superstar” effect. This increases the performance of bestselling products. When consumers with less prevalent preferences, such as niche items, are the main beneficiaries, the concentration of sales is reduced, generating a long tail effect. »
These three studies highlight that various aspects have to be taken into account to formulate a complete analysis. However, some statements can be done. As shown by the figures, there is a positive link between recommendations and sales, trough both direct and indirect effects. But recommendation only partly increase sales in the long tail. Niche products, as defined by Anderson, still have to find their way to the consumers.
References
https://econsultancy.com/blog/9366-ecommerce-consumer-reviews-why-you-need-them-and-how-to-use-them/
Show lesshttp://www.business2community.com/books/the-long-tail-and-what-it-means-for-content-marketing-01338858#RSYdQ4ztvusDMc5e.99
blog.beeketing.com/how-product-recommendation-helps-e-commerce-sites-increase-conversion-rates/
http://trouvus.com/blog/impact-of-personalized-product-recommendations-on-sales-volumes-on-ecommerce-sites/
http://www.business2community.com/infographics/impact-online-reviews-customers-buying-decisions-infographic-01280945#l80d9v0XRPpyvXmz.97
http://www.business2community.com/infographics/impact-online-reviews-customers-buying-decisions-infographic-01280945#20v8FyOysJbX28P6.99
http://www.invespcro.com/blog/the-importance-of-online-customer-reviews-infographic/
http://knowledge.wharton.upenn.edu/article/recommended-for-you-how-well-does-personalized-marketing-work/
http://www.inet.econ.cam.ac.uk/working-paper-pdfs/wp1520.pdf
http://www.cassknowledge.com/research/article/recommended-you-effect-word-mouth-sales
https://www.luminis.eu/wp-content/uploads/2015/08/White-Paper-Recommendation-in-e-commerce.pdf
Read more
First of all, I will give a clear definition of what the theory of the long tail involves. The theory assumes that our society and its customers are changing their buying habits. Until now the customers were mainly focused on a smaller number of mainstream products that can be considered as “hits”. Those products are situated in the head of the demand curve, the familiar market. This theory tells us that the customers are changing their habits towards a big number of products that are then situated in the tail of the curve, the niche market.
The theory is explained by the fact that overall costs decreased highly because of the new possibilities that is offered by the internet. Indeed, the internet gives access to all of the information and at any time. One simple example for this are the songs which are online which are not played on the radio.
“When consumers are offered infinite choice, the true shape of demand is revealed. And it turns out to be less hit-centric than we thought. People gravitate towards niches because they satisfy narrow interests better, and in one aspect of our life or another we all have some narrow interest (whether we think of it that way or not).” (Anderson, 2015)
In addition, we can ask ourselves if this theory is affected by the recommendations options that several platforms have put in place. The answer to this question is positive. Chris Anderson gave two conditions in order to make the niche products exploitable. The first condition is that everything has been made available and the second one is that the customers are helped to find them. The first condition is satisfied thanks to the internet’s characteristics. Those characteristics eliminated the overall access that can limit the access. The second one can be facilitated with the recommendation options. However it is important that those functions really fulfill this condition. In other words, the recommendation should not only put forward the hits products but also the niche products. Indeed, we can see that the current recommendation options work on a popularity basis and the customer’s own interest. The popularity basis could potentially erase the positive effect of the internet on the niche markets. Since we return to a system where the most classic product (the one that interest the majority of the people) has a better visibility.
Anderson, C. (n.a.). The long tail, in a nutshell. http://www.longtail.com/about.html (consulté le 24 mars 2015)
Celma, O. (2010). Music Recommendation and Discovery, 87-107.
Knowledge @ Wharton. (2009). Rethinking the Long Tail Theory: How to Define ‘Hits’ and ‘Niches’. http://knowledge.wharton.upenn.edu/article/rethinking-the-long-tail-theory-how-to-define-hits-and-niches/ (consulté le 24 mars 2015)
Show lessRead more
Let’s explain the steps in recommandation systems.
First, you have to collect data and filter them in order to have an idea of what the customers think and feel. You have two choices for that. First, explicit data gathering and active filter is the fact that the user indicates his interests to the system clearly. It’s for example a rating scale (1 to 5 stars), liking, comments. (1)
Second, the implicit data gathering and passive filter. This is the fact of looking at the customer’s behaviour without asking him clearly. For example, it’s the fact of getting the list of products that the customer bought, the music he has been listening to, etc.Both these methods have their advantages and disadvantages.
According to the website of the Carleton College : « In passive filtering, every user will be given the same predictions for a particular item. In active filtering, the system takes into account your specific history in order to make a recommendation. To put this distinction in a solid example, the news site Digg.com uses passive filtering, showing all users the articles which have received the most votes, whereas the online sales site Amazon.com uses active filtering, trying to recommend products based upon a user’s specific actions. » (2)
But I don’t think Amazon uses only active filtering, as they push informations about what other customers bought. (3)
According to Cassidy (2006), «Elaborate “filters,” such as search engines, blogs, and online reviews, which help to match supply and demand» (4) are indeed ways to make the long tail possible. But Fleder and Hosanagar (2009) said: « Recommenders can push each person to new products, but they often push users toward the same products ».(5)
Personnaly, it depends on whether the system is passive or active. I often buy make up online and seeing the opinions of different users is helpful. But I can’t help but to think that some of the consumers might be people within the company, trying to make us buy their product. That’s why I’m always paranoid about active filtering, especially when there is something in return for the user that rate, share, post of pictures of the products, etc.
My opinion is that the « length » of the tail depends on the filtering. To sum up very roughly in a sentence, I think active filtering will lend to a short tail as it will make customer aware of popular products, while passive filtering allows a long tail as it allow the customers to discover new products.
Sources:
Show less(1) http://www.podcastscience.fm/dossiers/2012/04/25/les-algorithmes-de-recommandation/
(2)http://www.cs.carleton.edu/cs_comps/0607/recommend/recommender/collaborativefiltering.html
(3)http://en.wikibooks.org/wiki/Strategy_for_Information_Markets/Collaborative_Filtering#Passive_collaborative_filtering
(4) http://webwhompers.com/the-long-tail.html
(5)http://www.stern.nyu.edu/networks/0710_Fleder_Hosanagar_Blockbuster_Culture%27s.pdf
Read more
As Chris Anderson stated “the theory of the Long Tail is that our culture and economy is increasingly shifting away from a focus on a relatively small number of “hits” (mainstream products and markets) at the head of the demand curve and toward a huge number of niches in the tail”. Firstly, I would like to evaluate the relevance of recommendation systems in people’s daily life because Anderson seems to assume that his theory applies in any case. Secondly, I will try to expand on the possible market niches a consumer might discover thanks to the theory of the long tail. Finally, I will discuss the threat of false reviews that could represent a real threat to companies using the long tail in their business strategy.
My first reaction after reading this article was to look back and to estimate my use of online recommendations. I have to admit that I do not read often the recommendations’ part when I’m thinking about buying something. However, it depends heavily on the amount I’m willing to spend on the product. In a second time, I realized that most of the time when I read comments of others on a movie it is after watching myself the movie to compare impressions and see if I missed something. Last week’s class analysed the impact of ads on people and advertisers. For my part, I hate ads and downloaded as quickly as possible adblock and still avoid websites using cookies because they upset me. This hate might result from the significant flow of ads users encounter by surfing on the Internet. Likewise, people might dislike recommendations because they are also overwhelming.
Now let’s move on to the possible market niches consumers might discover thanks to the recommendation systems of popular products. In this part, I would like to make a link with the class about net neutrality because in my opinion, Google and even YouTube are mostly not proposing the perfect match. However, we might suppose that they could because of the data and information they gather on consumers’ preferences. Nonetheless, we learned from previous scandals that Google chooses which companies are suggested first when a consumer searches for a specific product, depending on how much the firm is willing to pay to ensure its visibility online. Similarly, YouTube could propose music tailored to the consumers’ preferences but keeps to suggest the same videos with the most viewers. Therefore, I really wonder about this better access to market niches.
Finally, many companies on the Internet depend heavily on the theory of the long tail and some even build all their business around this specific part as for instance AirBnB. Firms using this kind of business model need to use it very carefully. Some of the comments from the last year already highlighted the problem of false reviews i.e. firms putting glowing comments about their service/products online. Some firms are even going further by paying some kid in Bengladesh to let a false review to make themselves known and stand out of the crowd, what makes them liable for prosecution in the US and in Europe (under consumer protection regulations). According to me, this kind of malpractice will just push users to disregard the possible comments and just generate bad advertisement to the concerned companies or individuals.
As far as I’m concerned, I prefer the last trend I observed in ‘real’ libraries (that might be existing for a long time) i.e. people letting post-its on the books with their impressions, feedbacks and books tackling the same subject than to learn how to differentiate the right matches from the pseudo suggested ones.
Sources:
Show lesshttp://www.longtail.com/about.html
http://www.theguardian.com/money/2013/jan/26/fake-reviews-plague-consumer-websites
http://www.nytimes.com/2012/11/04/technology/google-casts-a-big-shadow-on-smaller-web-sites.html?_r=0
http://www.economist.com/blogs/freeexchange/2013/03/utilities
Read more
An interesting recent development in the area of “recommender systems” is most probably the emergence of “personal assistants” such as Siri, Google Now, Cartona, S-Voice, etc. These softwares allow users to perform different tasks through a natural language user interface. Users can ask these assistants questions, get recommendations and make research on the internet.This sofwares even “proactively deliver information to the user that it predicts they will want, based on their search habits”.
For example Google Now can help you find a restaurant thanks to its recommendation system, then guide you there using Google Maps, taking into account the current evolution of the traffic on the roads, or providing you with the local subway schedules. It keeps you posted about eveything you might want to know, and sometimes even without you asking for it.
Personal assitants are developed as “machine learning” tools, which basically means the more they are used, the better they get. They base their answers and recommendations on “user input, location awareness, and the ability to access information from a variety of online sources (such as weather or traffic conditions, news, stock prices, user schedules, retail prices, etc.).”
What is very interesting with machine learning is that these tools try to emulate the human reasoning, and are developed in such a way they learn to think as the human brain. This means they dot not have “rigid” behaviours, and can actually produce recommendations which could easily be given by a human being. Machine learning is one of the many evolution of the Big Data revolution, which is currently changing the way many businessess are done.
Machine learning is still relatively young, but you can safely expect it to completely change the recommendation system business, as it would allow for more personnal, more individualised and more advanced results. Nowadays, recommendation system are based either on a collaborative filtering approach, either on a content filtering approach. This means that we either use user’s history and compare it with people who have similar tastes to make recommendations, or that we use similarity in contents to make these recommendations. With machine learning, we could one day be able to ak our smartphones for music recommendations just like if we were asking a friend. It would open the door to a lot of “obscure contents” which we couldn’t find on our own, but that the “artificial intelligence” of the personnal asssitant could find, and it would probably be very relevant.
Maybe one day we will all have a friend who knows everything about music and movies and that could make better suggestions than any other friend. And that friend would be in our pocket.
http://www.iactiveit.com/personal-assistants/
Show lesshttp://en.wikipedia.org/wiki/Recommender_system
http://blog.laptopmag.com/siri-vs-cortana-vs-google-now
http://en.wikipedia.org/wiki/Google_Now
Read more
The long tail effect has become popular with the emergence of e-commerce and the birth of websites such as eBay and Amazon. Long tail represents a retailing strategy that sells a number of unique goods in a small quantity, in addition of the mass sale of some blockbusters. The large number of special goods that are located on a niche market can be represented by the long tail on a graph. In other words, if you sum the size all the small niche markets and compare them to the size of the blockbuster markets, there is a possible competition between both. So, does the Long Tail grow the pie or does it simply reallocate demands from blockbusters to the niche market? According to Anderson’s blog, the first effect of Long Tail will be that our choices will be reallocated to more specific products. In the future, if we are more satisfied with what we have purchased, we could be tempted to buy more. An article from the Economist confirms this statement. When people finally find what they are looking for after a long time, they sometimes are willing to pay a strong price in order to become finally the processor of the so much desire good.
The long tail effect is possible because of an infinite available shelf space on the internet. Indeed, sellers don’t need to restrict them-self anymore at the physical space available in their shop, where the incentive was to present on the shelves only the products that sold the best. Goods no longer have to compete for physical space on shelves. Recommendation systems play a crucial role, highlighting to customers goods to which they maybe even didn’t dare to dream about.
My feeling is that niche markets get a chance to be known thanks to recommendation systems. While before, many worthwhile products were simply outlawed from the shelves, many products have now more chance to be known and also to be purchased. In fact, I am wondering about the effect that it must have on the market. It must certainly be positive, but in which extend? Do many SME emerge or have more chance to last thanks to the new opportunity their products have of being known on the web? Even more, has the Long Tail effect a positive influence on innovation? According to an article of the Economist, the rate of introduction of new products and services has exploded during this last years; leading to think that company will be producing for a “market of one”.
What about the consumers? Is their utility increasing thanks to the fact that they found exactly the rare product they were looking for? The behavior of customers has changed and suppliers must adapt to that change of behavior.
As a conclusion, I think I am not wrong with stating that the internet seems definitely something that has increased the efficiency in the responding to the demand by the suppliers. While before consumers had to be satisfied with goods that did not suit exactly their preferences, they now can find what they hope for on the internet and are ready to pay higher prices for it.
SOURCES :
Show lesshttp://www.economist.com/node/12762429
http://www.economist.com/news/finance-and-economics/21638142-consumers-reap-benefits-e-commerce-surprising-ways-hidden-long
http://www.longtail.com/the_long_tail/
You raise a number of very interesting issues. Good!
Read more
Numerous studies have demonstrated that the positive impact of recommendations systems leaned, indeed, more towards the ‘’long tail business’’ by having a larger impact on niche consumers because they would have more interactions between themselves (Drane, 2007) (1). It was also justified by the facts that the profit margin would be higher for the niche products in comparison to ‘’best-sellers’’ products, which wouldn’t give as much profit : due to their popularity, the competitors would have to sell them at the same price. (Oktar, 2009) (2).
However, two conditions have to be completed in order for the long tail business to thrive. First, the products have to be readily available (no stock shortage) and second, the products can be easily found. (Anderson, 2006) (3). The later imperative is the one that causes problem. Indeed, the most popular existing recommendation algorithms are not efficient for helping ’’users to find long tail items with limited historical data due to data sparsity, even if they would be viewed favorably.’’(4)
As a matter of fact, one of the most popular recommendations model, the ‘’collaborative filtering’’ take into account our past usage and on a huge amount of usage data from other users (Youtube use this system for example) in order to recommend products to the users. The problem is that this method doesn’t benefit new products or less popular items situated at the end of the long tail. (Domingues et Al., 2014) (5).
Fortunately, these last three years, numerous studies, whose purpose is to design new recommendation algorithms, have been written in order to solve the problem and stimulate the ‘’whole’’ of the long tail. (6)
To conclude, even if the actual recommendation system does make the tail longer, the most used recommendation algorithms cannot help customers find lesser known items situated at the end of the tail, even if they would actually correspond to the users’ needs and wants. Thankfully, new studies have been done in order to optimize the recommendations algorithms and make the system more efficient so that the long tail business model can thrive as a whole.
Bibliography:
(1) A. Drane (2007). Word of Mouth and Recommender Systems: A Theory of the Long Tail. Retrieved from http://www.iese.edu/en/files/SPSP%20LUNCH%201%20abril_tcm4-23421.pdf.
(2) D. Oktar (2009). Recommendation Systems: Increasing Profit by Long Tail. Retrieved from http://en.webrazzi.com/2009/09/18/recommendation-systems-increasing-profit-by-long-tail/.
(3) C. Anderson. (2006). The Long Tail: Why the Future of Business is Selling Less of More. Hyperion.
(4) C. Chen Et Al. (2012). Challenging the Long Tail Recommendation. Retrieved from http://arxiv.org/pdf/1205.6700.pdf.
(5) M. Domingues Et Al. (2014). Combining usage and content in an online recommendation system for music in the long-tail. Retrieved from: http://www.researchgate.net/publication/239761862_Combining_usage_and_content_in_an_online_recommendation_system_for_music_in_the_long-tail.
(6) 4&5
D.Hu. (2014). Style in the Long Tail: Discovering Unique Interests with Latent Variable Models in Large Scale Social E-commerce. Retrieved from http://mimno.infosci.cornell.edu/info6150/readings/p1640-hu.pdf.
A. Bhuvaneswari Et Al. (2012). Improving diversity in video recommender systems and the discovery of long tail. Retrieved from http://www.jatit.org/volumes/Vol37No2/10Vol37No2.pdf.
Show lessRead more
Recommendations systems are increasingly present on many websites. We particularly cited non-exhaustively the most famous like Amazon, Netflix, Pandora, etc.
This topic is quite complex and involves many mathematical concepts and probability. I think that the data collect is one of the important topics to discuss when we talk about the recommendation system. The data collection can be done in two ways: implicit and explicit. Implicit data collect is rarely used for websites because it is more binding on the consumer. Indeed, the consumer must explicitly show these preferences in relation to a set of products or services. This may for example be through a ranking system where the user class products / services based on these preferences.
The implicit collect of data, meanwhile, is more complicated to implement because it uses more complicated algorithm but is much less restrictive over the consumers. It is this type of collect is most often used for websites. This may for example be based on the consumer objects seen on the website. Other criteria, less common, such as interests, the places of residence, the age of the consumer, etc. can also be took into account.
Of course, each of the data input to the algorithm must make a difference in importance according to criteria defined by the website. Example, we can easily imagine that the data “last purchase made” or “product in the basket” have far more important than ata “product page visited.”
To maximize the effectiveness of recommendation systems, a latent semantic models is often used. Professor Saerens said in his courses: “We say that semantically related words are used when they are in the same context “. For example, the words “baby” and “newborn” have semantic similarities. This allows, for example, that the recommendation system provides consumers with product recommendation with description as “baby pushchair” while the consumer was looking for “newborn pushchair.”
Source :
Show lesshttp://fr.wikipedia.org/wiki/Système_de_recommandation
http://www.cs.umd.edu/~samir/498/Amazon-Recommendations.pdf
Slides of the professor Saerens for the course of Data Minning (UCL, actuarial curriculum)
Read more
The long tail approach is a way to describe niche markets and how they work on the Internet. At the very beginning the physical limitation of stores didn’t allow them to expand their supply, hence they preferred to offer records, books, movies, and other items that would have been successful. Nowadays with the Internet revolution and online shopping development this system is outdated because there’s no need of a geographic location. So Amazon can sell obscure books, Netflix can rent obscure movies, and iTunes can sell obscure songs. But the same concept can also be applied to Google, in the sense that as Anderson said they are “long tail advertisers”. Rather than focusing on being the most popular page, in the niche market it is better to appear in a lot of pages.
Show lessAccording to me this kind of approach has for some websites many limitations. For example the recommendation system of Youtube should be improved otherwise you find yourself in a loop, in which you feel annoyed by the same related music videos . So because I consider inefficient their selection I use another app called 8tracks. The main difference is that with some tags this app will always find some items that respect my preferences, at the same time are innovative and I haven’t ever listened before.
In conclusion, this approach has his limitation because from one side we have a broader range of selected products, but on the other side the community that ranks and evaluates the product is smaller (and the long tail approach is valuable only if you have a broad range of quality reviews). For example customers that evaluates Amazon.fr amounts only for 12% of total transactions. Hence we have biased recommendation. The trend is that many reviews are given by adolescents (self-selection because online reviewing is a social practice empowering reviewer’s self-estime and ego).
References:
Six Degrees of Reputation: The Use and Abuse of Online Review and Recommendation System, David & Pinch, http://papers.ssrn.com/sol3/papers.cfm?abstract_id=857505
http://www.fritscher.ch/hec/_media/si/reputation-systems-finale.pdf
http://google.about.com/od/googleforbusiness/f/longtailfaq.htm
https://www.techdirt.com/articles/20081123/1230162928.shtml
Read more
When the Internet was first created, its main purpose was to serve as a platform of information that people could just as well inject as collect. With the years passing by, the quantity of information the web contained grew tremendously.
Nowadays, it is possible to find thousands of answers to a simple question. In my opinion, we have reached a level of knowledge so high that the efficiency of the search engines such as Google, Yahoo, etc. has been compromised. Bertram Gross defined the phenomenon of “ information overload”, in his book ” The Managing of Organizations”, the following way: Information overload (also known as infobesity or infoxication) refers to the difficulty a person can have understanding an issue and making decisions that can be caused by the presence of too much information.
Recommendation engines are therefore very useful tools, since they scan and classify the billions of data according to the desires and preferences of each individual.
However, it is not relevant to advertise products that may have already been easily discovered by the user, being famous products falling in the taste category of the consumer. A popular product does not need a boost to be visible to the consumer.
It can even be annoying to perceive the obvious individual advertising on some websites.
I believe the power and value of a recommendation system lies in its ability to introduce us to new things. Products that suit our interests, but that we would’ve never discovered without the recommendation data research engine digging it up for us. Indeed, it results in giving us the impression of finding a rare unknown pearl among the other products.
Therefore, a solution to make the tail longer could be to sensitize and boost the algorithms of recommendations systems. In order to accentuate their research performances and increase their capacity to dig deeper in the giant ocean of information the Internet is today, and brings new products to the consumers.
As conclusion, I think it is important to promote the application of the long tail theory. If we are using Internet platforms to purchase objects, it is to have access to a range of varied personalized proposals and not to be found in the case of a local business.
References:
http://en.wikipedia.org/wiki/Information_overload#cite_note-1
Show lesshttp://www.podcastscience.fm/dossiers/2012/04/25/les-algorithmes-de-recommandation/
Gross, Bertram M. (1964). The Managing of Organizations: The Administrative Struggle. p. 856
Read more
Making recommendations to the online consumer seems to take place in almost every industry present online. Even Walmart.com makes personalized item recommendations and keeps working on increasing the quality and frequency of its recommendations. It is interesting to note that Walmart uses information both from customers’ online purchases, as well as store purchases. Walmart wants to see its online business do well, mainly because of Amazon’s increasing success, stealing away share from physical retailers (reference below).
Another interesting, and perhaps a little scary, example comes from the gaming industry. GSN Game is a company designing mobile games. The company collects a huge amount of information from its users and uses this information to a very wide extent. I found interesting to learn that they use this information to track customers’ preferences, but also to customize services and games accordingly.
The illustrate this, technologyreview.com says the following : « If two people were to download a game onto the same type of phone simultaneously, in as little as five minutes their games would begin to diverge—each one automatically tailored to its user’s style of play. » (reference below).
However, the company doesn’t limit its use of information to preferences and customization of services. GSN Games also induces its players to play longer and try out more games, by observing the player’s behavior, from the frequency to the speed of pressing on the screen. This way, the company is able to foresee players getting tired. The company will therefore suggest new games, in order to extend its customer’s « stay ».
Influencing people’s behavior and attitudes through « persuasive technology » has been around since the 90s, but today we can observe a new trend : companies start using data on customer behavior to design products that « are not just persuasive but specificially aimed at forging new habits », as explained by Ms. Byrnes (technologyreview.com).
Habit formation used to be the breadwinner of addiction-inducing product manufacturers like the tabacco industry and casinos. The great progress in behavioral economics and psychology of the last decade, allows many industries to create persuasive technologies.
Currently, companies seem to be focussing their efforts on understanding how people make choices, in order to stimulate activity (that is, encourage daily visits or induce longer visits as in the GSN case).
The use of online consumer data is subject to debate. Regarding this new trend in “persuasive technology”, a question to debate is the following : what are the appropriate limits of such persuasion and habit formation ? I might add to that, what will be the impact of such persuasion on our conception of freedom and freedom of choice?
According to Ms. Byrnes, there is a need for regulation. Transparency and disclosure, or even asking for a user’s permission to persuade him/her, could be a first step towards it (Nanette Byrnes, technologyreview.com).
Sources :
Show less1. http://www.geekwire.com/2014/walmart-com-borrows-ideas-amazon-offering-product-recommendations-fast-check-outs/
2. https://plus.google.com/+GregLinden/posts/YXQZVre3jk1
3. http://www.technologyreview.com/news/535826/technology-and-persuasion/
I’m not sure to see the link with the long tail hypothesis.
Read more
In my opinion, recommendation systems make the “tail” longer.
According to Chris Anderson, the theory of the Long Tail is that our culture and economy is increasingly shifting away from a focus on a relatively small number of “hits” (mainstream products and markets) toward a huge number of niches in the tail.
With the fast growth of internet market, people get access to enormous amount of products. As the costs of production and distribution fall, especially online, there is now less need to lump products and consumers into one-size-fits-all containers. In an era without the constraints of physical shelf space and other bottlenecks of distribution, narrowly-targeted goods and services can be as economically attractive as mainstream fare. I, for an example, am an user of Netflix, and I find that the recommendation lists are especially useful. I love comedy shows, but I only watch the popular, or mainstream comedies; however, Netflix recommend many good comedies to me, some of them I would never thought about before. With the recommendation systems, people get more access to the niche products.
In today’s music industry, music consumption is biased towards a few popular artists. For instance, in 2007 only 1% of all digital tracks accounted for 80% of all sales. Similarly, 1,000 albums accounted for 50% of all album sales, and 80% of all albums sold were purchased less than 100 times. Current music recommendation systems try to accurately predict what people demand to listen to. Current music recommendation systems try to make accurate predictions about what a user could listen to, or buy next, independently of how useful to the user could be the provided recommendations. There is definitely a huge need to assist people to filter, discover, personalize and recommend from the huge amount of music content available along the Long Tail.
Quote:
Show lesshttp://www.longtail.com/about.html
http://mtg.upf.edu/node/1242
http://mobblog.cs.ucl.ac.uk/2007/06/15/recommender-systems-and-the-long-tail/
Read more
The client is an ambassador of the product. Collaborative marketing is largely based on the good opinion from satisfied customers. The marketer’s intervention is useful to organize the recommendations to make them have a greater impact and to moderate critics by answering them with moderation in order to give credibility to these consumer’s opinions. Publishing customer recommendations has become a popular manner to promote online businesses and establish good reputation. An ambassador represents the brand. Strategically speaking, the brand has to identify the key opinion leader, create a community and turn it to good account. (Bernard, Fèvre, Kovacs, 2012).
This function is taken by the community manager (Synthesio, 2010) who becomes the spokesperson of the brand on the web. His or her function consists in following the brand into its marketing strategy in building a real community of ambassadors.
A concrete example of community management is the Gatorade’s experience of Command Center. Led by a team of six community managers, the Command Center conducts the community of fans and meets the consumers’ demands or claims.
Unfortunately, this comments’ management system has also been applied negatively on some booking websites of, for example, restaurants or hotels. This opportunity has been taken by companies who offered positive comments in exchange for payment for these advertisers to give a fictitious web-reputation.
Manjul Gupta and Joey F. George (Iowa State University) analysed this issue in a round table presentation called ‘the effects of fictitious customer testimonials across different cultures’ and concluded that “relatively little is known about how customer testimonials influence perceived deception and trust among web users ” (Gupta, George, 2013). Studies still have to be launched in order to analyse more deeply the impact of fictitious recommendations on web-reputation and consumers.
However, when the recommendation is spontaneous, the client becomes a business partner who needs neither a gift nor a payment but only that the real product’s quality meets the promised quality. A satisfied customer with a great friends’ network allows to quickly convincing new targets with a minimal marketing effort.
Sources:
Show lesshttp://aisel.aisnet.org/amcis2013/HumanComputerInteraction/RoundTablePresentations/12/
https://synthesio.com/corporate/wp-content/uploads/2010/11/SYNTHESIO-Community-management-ou-relation-client-en-ligne1.pdf
http://www.e-marketing.fr/Marketing-Direct/Article/COMMENT-TRANSFORMER-LES-CLIENTS-EN-AMBASSADEURS-DOPER-LA-REPUTATION-DE-SA-MARQUE-43070-1.htm
http://www.exacttarget.com/sites/exacttarget/files/10-Examples-of-Social-Media-Command-Centers.pdf
http://mashable.com/2010/06/15/gatorade-social-media-mission-control/
I’m not sure to see the link with the long tail hypothesis.
Read more
First of all, the recommendation system ables to “predict the rating or preference that user would give to an item thanks to the user’s past behaviour” (Wikipedia). This system permits to reveal the truth about what consumers want and in some cases, helps some items to be selled again (Touching the Void of Joe Simpson with his book that increases their sales decades after his release). But is that recommendation system increases the long tail effect ?
I think it depends on the sector, the type of the product and the algorithm used by the website. I’ll take some example. If you use an algorithm which recommends you a book or a film with the same subject it could be very interesting and it makes the tail longer. With « Allocine », when you read details and the test of a film, the website provides other films that « you could like » on the same subject (and not only best-sellers or films from the same producer). From my own experience, I really like film of « prisons ». When I read the test of « The Green Mile », I discovered a lot of really interesting films on this subject and which was not really known by people because of the lower marketing resources of the producers.
On the contrary, Youtube recommends most of the time « videos » or « music clips » that are the most viewed (« Vevo » recommebds each time the most popular songs). We also realize that some « videos » are all the time recommended. Youtube favors the biggest creators with more money who pay to be seen. In this case, recommendation make the tail shorter for the small budgets.
In order to improve his recommendation system, Netflix launch a competition « the Netflix Prize », where the best improvement was rewarded of 1000000$. The winner of this competition improved the algorithm just 10% better but finally, they never implemented this solution in their company. WHY ??
Because the « additional accuracy gains that we measured did not seem to justify the engineering effort needed to bring them into a production environment » (Forbes.com).
To conclude, the approach of your recommender system will depends on your algorithm and what you want for your website. The first approach is the Collaborative filtering “which can be constructed from a single user’s behavior or also from the behavior of other users who have a similar behavior.” The second one approach is the “Content-based filtering which constructs a recommendation on the basis of a user’s behavior” (ex : historical browsing information). You also have hybrid approaches which is a mix of these both approaches and which could be the best thing according to your personal offer.
Sources :
Show lesshttp://www.ibm.com/developerworks/library/os-recommender1/
http://en.wikipedia.org/wiki/Recommender_system
http://archive.wired.com/wired/archive/12.10/tail.html
http://en.wikipedia.org/wiki/Netflix_Prize
http://www.forbes.com/sites/ryanholiday/2012/04/16/what-the-failed-1m-netflix-prize-tells-us-about-business-advice/
Read more
Many recommendation systems exist and evolve over time. Consider first the example of the digital music player Spotify. Spotify has decided to change its recommendation system into a more intelligent system. In fact, before it used a collaborative recommendation system. That is to say that if two people often listen to the same kind of music, they will say they have the same musical tastes. It is their behavior on the musical platform that are analyzed. But this technique has a big disadvantage it is that it often tends to recommended songs that are already known because they are often the ones that come first. But the music library likes to make the user discover new songs that he has never heard before.
Therefore, Sander Dieleman, a PhD student and intern Spotify, decided to completely change the system and use deep learning to recommend songs. This is to analyze the type of song, for example: gospel, French rap, etc and propose other songs in this same range of songs, new ones that are completely unknown to the user, but could turn into a beautiful discovery. I think this system is well done because it avoids to only offer songs already known and allows the user to expand its horizons.
Another model of recommendation is the “peer-to-peer”. A first user is given a list of movies, music, books, … and should see if he could possibly recommend this list to a friend. If this is the case, the friend will receive this list by saying that the first user recommends this selection. But this selection doesn’t only include the elements of the selected list, but also other elements that have been selected by a recommendation engine. The fact that a friend recommends a product, that can be a book, a song, a movie or something else, pushes us to pay more attention because we trust him. And the fact that there are additional products that are available is also a good idea because it can lead to discoveries.
Each recommendation system has its limits and this is why hybrid systems are often proposed to overcome the disadvantages of each system and try to take advantage of their benefits. For example, there is a hybrid system that consists of a combination of collaborative system and content-based methods.
Finally, I think that social networks can play a very big role in recommendation systems, especially in peer-to-peer. Social networks can help to improve recommendation systems. For example, if two people are friends, we can consider that they have some things in common and it is not necessary to try to look for other means of similarities. Finally, social networks put a lot of information available about users. This is a real gold mine for data collection. Thus for example through notices “like” a system can already identify potential interests of its user.
Sources:
Kanna Al Falahi, N. M. (2014). Computational Social Networks. Chapter 18: Social Networks and Recommender Systems: AWorld of Current and Future Synergies. Springer.
http://benanne.github.io/2014/08/05/spotify-cnns.html
Robert Bodor, A. W. (2014). Patent n° US8825574 B2. United States.
Show lessRead more
For many platforms recommender systems play a role as important as the Search, Advertising and Networking systems; for example, for Google News recommendations generate 38% more clickthrough, 35% of Amazons sales comes from recommendations, 2/3 of the movies watched at Netflix are recommended and 28% of the people at Choicestream would buy more music if they found what they liked. As available data is become bigger and bigger the algorithms used to build Recommender systems are becoming more complex and accurate meaning that the importance of user and data mining is going to be a growing trend in many areas for such us platforms.
According to “Recommender Systems Handbook” authors, there are 2 “traditional” recommender systems: Collaborative Filtering (recommend items based only on the users past behavior ) which can be user-based if the recommendations depend on what similar users to me prefer or item-based if they depend on similar items to those that I have previously liked. The second one is Content-based Recommendations (based on items features).
There are also some “novel” methods including Personalized Learning to Rank (treating recommendation as a ranking problem), Context-aware Recommendations (recommendations based on user features, specially demographics) and Social recommendations based mostly on users similarity and trust. What most platforms put in practice is a Hybrid method that combines any of the above methodologies depending on the domain and particular problem the platform is face to or in order to improve results in specific cases such as the “cold-start” problem those platforms face when they don’t have yet an important users base. The most common Hybrid methods are Weighted where outputs from several techniques (in the form of scores or votes) are combined with different degrees of importance to offer final recommendations, Switching (depending on situation, the system changes from one technique to another), Mixed (several techniques at the same time), Feature Combination (features from different recommendation sources are combined as input to a single technique), Cascade (the output from one technique is used as input of another that refines the result), Feature augmentation (The output from one technique is used as input features for another) and Meta-level where the model learned by one recommender is used as input to another. According to Amatrianin (2014), in the general case it has been demonstrated that the best isolated approach is Collaborative Filtering.
As a practical example. In Netflix recommendation approach they go from the rating system (main input) with which they build a ranking and finally a “page generation” (grid with rows and column). Everything is personalized, for example the genre rows are generated in an implicit way (based on user´s recent plays, ratings, and other interactions), explicit way (taste preferences) and combining the both (hybrid way). Thanks to their methodology the go from a “page generation” of 10,000´s of possible rows and thousands of possible videos for each row to one personalized page of 10-4o rows. They try to have a balance between accuracy and diversity, discovery and continuation, depth and coverage and Freshness and stability; this is in line with the 3 rules described in The Long Tail, especially when having a balance between Hits and less well-known films.
From the Netflix case is interesting how they don’t put so much emphasis on the social side since they argue that if they know enough about their customers, social information becomes less useful; they track what users get shown and their interactions but they recognize that from an infrastructure perspective it is very costly. Also regarding the context they had to improve their recommender systems, Netflix own offline studies showed the wining models were too computationally intensive to scale so the expected improvement was not worth the engineering effort (they also acknowledged that their focus had shifted to other issues that had more impact than rating prediction).
Sources:
Show less“Recommender Systems Handbook.” Ricci, Francesco, Lior Rokach, Bracha Shapira, and Paul B. Kantor. (2010). http://www.cs.bme.hu/nagyadat/Recommender_systems_handbook.pdf
http://www.slideshare.net/xamat/recommender-systems-machine-learning-summer-school-2014-cmu
http://misrc.umn.edu/wise/2014_Papers/134.pdf
http://en.wikipedia.org/wiki/Long_tail
http://archive.wired.com/wired/archive/12.10/tail.html?pg=3&topic=tail&topic_set=
Read more
What an extraordinary news for artists, writers or even designers who previously needed to reach a certain amount of potential consumers in order to be launched. This amount was the minimal number of copies that would be sold to make the producer and the distributor profitable. Now, it is still true but this number is way lower as the distributor’s costs have become lower as well. Thereby, a huge creative potential has been unleashed. The diversity of products is exploding. To me, this phenomenon is the greater advantage of the long tail theory.
However, this amount of new products on the market is creating an subsequent problem: how to promote the not yet known creations?
From my point of view, various recommendation systems are the solution. It enables consumers to discover little known artists they will enjoy even more than very popular ones. In order to give a personal example, I will share my experience with Spotify, a music streaming platform. Traditionally, the recommendation system used by Spotify is a collaborative filtering approach [1]. This technique do not use any of the content characteristic. It is exclusively based on the consumption habits. However, this recommendation system is not very appropriate to make the tail as longer as possible. Indeed, since it is only providing recommendations of listened songs, it is more likely to be recommended famous songs than creations unknown from the general public. Therefore, Spotify has started to get involved in content-based approaches. There are numerous informations that can be learned from the song such as the artist previous featuring and music style, lyrics and especially the music signal in itself. The content-based approach enable the members to discover completely new songs since it doesn’t even need to be rated or previously listened [2]. Unfortunately, the use of the music signal is very complicated and needs to be improved in the coming years.
Even if the recommendation system of Spotify isn’t perfect and isn’t exploiting yet the content-based approach at its best, I think that it enables users like me to discover many more artists and thereby increases the diversity of discovered songs.
Sources :
[1] http://benanne.github.io/2014/08/05/spotify-cnns.html
Show less[2] http://www.cp.jku.at/people/seyerlehner/supervised/seyerlehner_phd.pdf
Read more
According to Chris Anderson « The theory of the Long Tail is that our culture and economy is increasingly shifting away from a focus on a relatively small number of “hits” (mainstream products and markets) at the head of the demand curve and toward a huge number of niches in the tail. As the costs of production and distribution fall, especially online, there is now less need to lump products and consumers into one-size-fits-all containers. In an era without the constraints of physical shelf space and other bottlenecks of distribution, narrowly targeted goods and services can be as economically attractive as mainstream fare »(1)
In my opinion, recommendation systems can in certain cases make the “tail” longer but it depends on certain factors.
In order to understand how online product recommendations could influence consumers’ online choices, we should refer to the article of Sylvain Senecal and Jacques Nantel (2).
In their article, they aim to investigate consumers’ usage of online recommendation sources and their influence on online product choices.
In their study, they focus on three determinants that could influence the impact of computer-mediated recommendations on consumers’ online product choices.
The three determinant that they investigate in their study are: the nature of the product recommended, the nature of the website they are recommended on and the type of recommendation source.
The first determinant underlines the fact that goods can be classified as possessing either search qualities (those that the consumer can determine by inspection prior to purchase) or experience qualities (those that are not determined prior to purchase).
The second determinant underlines the fact that the nature of the website can also influence the impact of a given recommendation.
They are three kinds of websites: sellers, commercially linked third parties and non-commercially linked third parties.
The last determinant sorts the type of recommendation sources in three categories: other consumers, human experts and expert systems.
After testing several hypothesis on these assumptions, they were finally able to show that consumers would be more influenced by recommendations for experience products than for search products which seems logical as it is difficult or even impossible to evaluate experience products before purchase, consumers should then rely more on product recommendations for these products than for search products.
But also that no relationship exists between the type of website and subjects’ propensity to follow product recommendations
And finally, that the recommender system (expert system) is more influential than other consumers and human experts.
It appears that a recommendation source providing personalized information to consumers is more influential than a recommendation source providing non-personalized information.
To sump up, I think that the article of Senecal and Nantel help us understand that the theory of the long tail is not sufficient to itself and that other factors must be taken into account if we want to evaluate the influence of a recommendation system.
Indeed, I think that a recommendation system could make the tail longer but it will mostly depend of the type of product and the type of recommendation system.
Having said that, I think that the theory of the long tail doesn’t apply with the same intensity in every sectors or industries.
Then, we cannot generalize a positive impact of the recommendation systems.
(1) http://www.longtail.com/about.html
Show less(2) Senecal,S., Nantel,J. (2004). The influence of online product recommendations on consumers ‘online choices, Journal of retailing, 80, 159-169.
(3) http://archive.wired.com/wired/archive/12.10/tail.html?pg=5&topic=tail&topic_set=
Read more
Currently, all the recommendation systems are taking more and more importance on Internet. Indeed they are simply omnipresent. Here are some examples:
• Online purchasing website knows what are your preferences and give you suggestion once you bought a certain product.
• Social networks manage to find friend you probably know or “fan-pages” you probably like,
• Broadcast video website (like Youtube) propose related videos (based on the content of what we just have seen) .
It is clear that the main advantage of this is that it can be very helpful to make discoveries of new artists, persons … Besides, it’s not new that people refer to others in other to gather information. For instance, before watching a movie, many of us check on the net the critics on specialized website.
However, there’s something that really bother me and raise an important issue. By getting all those “suggested” products or services, I personally think that we inhibit our free will and choices. I have the feeling that, due to those recommendations system, it’s making us like robot that follow what the firms claim good to offer. It is sad that our choices rely that much on mathematical algorithms and are not especially based on the personal willingness to discover the unknown.
Show lessRead more
A first update worth mentioning is the distinction between user recommendations and system recommendations. This distinction is around for a wile now, however, only recently these different types are empirically investigated, moreover the difference in effects of both was studied.
On the one hand we have system recommendations, which are most commonly generated by the collection of all user preferences. With these preferences the firm will suggest products to consumers. A very well known example of a system recommendation is the co-purchase recommendation of Amazon. Here, recommended products will be listed as “customers who bought this item also bought”.
On the other hand we have the user recommendations or also called user-generated word-of-mouth (WOM). They display consumer reviews of products whereby the review is based on the volume and the rating of the product.
This division of recommendation systems into two different types is necessary according to the paper of Zhijie Lin because according to her, both recommendations will generate different impacts. According to Lin, user recommendations are more effective than system recommendations in driving product sales.
Recommendation systems are already very popular and in my opinion they will only increase during the years. I based this comment on some striking results I found in a recently conducted survey:
– 85% of customers consuls online reviews before purchasing
– 65% of consumers claim that there purchasing decision is influenced by online reviews of other customers.
To further optimize the recommender systems firms need to increase the amount of consumer feedback that they receive. This because customer feedback will improve accuracy and can lead to a learning curve for the firm to notice a pattern in consumer preferences.
Literature has extended its knowledge about the efficacy of the recommendation systems quite a lot. Literature found that there are some differences in efficacy in relation to the sector/ type of products sold but also in relation to the characteristics of beneficiaries and in firms.
A recent paper from Harvas-Drane (2015) found empirical prove for the fact that recommendation systems do not work as efficient in all sectors. He argues that these recommendation systems are very useful for the following products: music, films, books and video and game entertainment.
He also makes a distinction in efficacy of the recommendations according to the characteristics of beneficiaries. According to him, customers with less prevalent preferences will create a long tail effect and the concentration of sales will be reduced. Consequently products that have a smaller share of total sales, otherwise called niche products, will generate a long tail. This finding is still in accordance with the long tail theory, firstly discussed by Anderson in 2004.
The firms that have lower inventory costs will benefit the most of the long tail effect because they can increase the depth of their assortment more easily and to a wider extent then others. As a consequence these firms are better positioned to serve demand for niche products (that leads to the long tail effect).
A recent study argues that word of mouth can contribute to the explanation of the long tail phenomenon.
Harvas-Drane (2015) links recommendation systems with word of mouth, which led to some interesting findings. He argues that recommendation systems and word of mouth could benefit all consumers (because of the lower search costs), but they have a larger impact on those consumers that have less prevalent preferences. However, there are also benefits for the firms: the market share of niche products in the sales distribution will increase, leading to a longer tail. But also a reduced concentration of sales within the firm’s assortment is noted.
References:
Hervas-Drane, A. (2015). Recommended for you: The effect of word of mouth on sales concentration. International Journal of Research in Marketing .
Zhijie, L. (2014). An empirical investigation of user and system recommendations in e-commerce. Decision Support Systems (68), 111-124.
Show lessRead more
Before answering the question, it may seems interesting to see the evolution of the « tail » over time. As it is said in the article « Goodbye Pareto Principle, Hello Long Tail: The Effect of Search Costs on the Concentration of Product Sales » roughly 10 to 15 years ago, the Pareto Principle was prominent.
This principle is quite simple : 20% of the products available on a market generate 80% of its sales. The distribution of the sales was therefore highly concentratred and a lot of niche products were never sold at all.
However, with the internet this trend may disappear, being replaced by what is called « the long tail ».
The authors identify two causes to this transformation of the market.
First, internet catalogues can offer much more choice than retail stores, as they do not need to have physical restrictions (stock size can be much bigger : eg. Amazon’s warehouses size).
Second, the customer can access much more information at lower costs and therefore be interested in niche products.
To answer the question and see if recommandation systems still allow to create a « long tail », we will first speak about personnal experience.
The first example we have is Spotify : it has been a few years now that we have subscribed and since we have discovered a lot of new songs. These songs are related to the kind of music that we listen, share and search, but is unkown and off the charts, there is a real feeling of « personnal assistant » looking for the songs we may like.
The second example is Netflix, that is available in Belgium since a few months. According to the article of the Huffingtonpost in the references, the firm is making changes in the market of entertainment by using a long tail strategy.
Indeed, the firm buy licenses for a lot of small shows that attract users and spur them to renew their subscription : these are the shows that create the « long tail » and that are suggested to the user according to his preferences. With the gains of the subscription, they pay for blockbusters as « House of Cards ».
From our personnal experience, the suggestions offered by Netflix are less « exclusive » than the ones offered by Spotify. The films, shows, series suggested are already known most of the time and although we also realize that there is some personnalization, we have the feeling that it is less advanced.
Maybe because of the size of the catalogue in Belgium for the moment, or because of a different algorithm.
Finally, it is quite difficult to find recent information (less than 2 years) about the subject, excepted from the paper of Oesterreicher-Singer and Sundararajan.
Nevertheless, considering the fact that the strategy of most services today on the web is a personnalized targeting and that most people (us included) want personnalized content, we can guess that recommandation algorithms developped by the firms will go this way and make the tail longer and longer over time.
Sources:
[1] http://pages.stern.nyu.edu/~goestrei/LongTail.pdf
Show less[2] http://www.huffingtonpost.co.uk/christopher-goodfellow/netflixs-long-tail-is-for_b_4716228.html
[3] http://hdl.handle.net/1721.1/74642
Read more
I have notice that we can find a lot of things on the long tail phenomenon and the recommendation systems. The first paper I would like to mention has been written by Bin Gu, Qian Tang and Andrew B. Whinston and deals with the effect of online information on the long tail phenomenon. In this paper the authors mentions two opposite beliefs concerning the effect of information on the formation of long tail. The first belief considers that the existence of online information available for customers encourages the formation of long tail. On the other hand, the second belief thinks that the customers ignore the available information and that they base their choices only on their personal beliefs.
The three authors have tried to determine which belief is the right one. Their study showed that positive comments improve the sale of popular product more than the sale of niche product. Consequently it means that the existence of information on the Internet has a negative impact on the formation of long tail.
I find this conclusion interesting because personally I always read comments of other customers before watching a film for example. Moreover I think those comments have more and more importance on the web. Indeed many people decide to make comments and therefore their influence should be increasing.
Huanxing Yang has developed in 2013 the second analysis I would like to mention. He has created a search model to explain the long tail effect. In his theory, the long tail effect is due to the decrease in the price of searching or an increase in the search targetability. What does the author means by the price of searching and search targetability? Today with the Internet, customers have easily access to a large number of products. They are note limited to the products available in the store they visit. Consequently the distribution of sales has become flatter and niche products have gain market share. So the long tail effect is encouraged by the decrease in the price of searching and the increase in the search targetability.
Finally I have noticed that several techniques have been developed in order to maximize the formation of long tail. Techniques such as adaptive clustering, hybrid music recommender system which combined usage and content data, ect.
References:
Jeyshirii, S., Thivakaran,T.K. (2014). Precise Recommendation System for the Long Tail Problem using Adaptive Custering Technique. International Journal of Innovation Research in Computer and Communication Engineering, 2(4).
Gu, B., Tang, Q., & Whinston, A. (2013). The influence of online word-of-mouth on long tail formation. Decision Support Systems, 52, 474-481.
Yang, H. (2013). Targeted search and the long tail effect. RAND Journal of Economics, 44(4), 733-756.
Show lessRead more
In order to articulate my point, I’m going to use the on demand music streaming service Spotify and its recommendation system.
Spotify is a rather new service that takes the tangent compared to its offline service counter parts like iTunes, Winamp and foobar2000. With the ever-increasing ease to have an internet access through WiFi and 3G/4G, Spotify has managed to make its service a leading force in the music software industry. Aside from being in the right place at the right time, Spotify has also made a name for itself thanks to its recommendation system that goes much deeper and further than any other in the genre.
Spotify has mostly relied on collaborative filtering which means that the users’ preferences come from historical data usage. While this system works and might be enough in most services (Amazon, imdb …), in the case of music, it has its flaws. Because it relies on usage data, the more popular and item is the more likely it will be recommended. Unpopular songs are therefore barely recommended which results in make the tail shorter.
Spotify have been developing what they call ‘Deep content-based music recommendation‘. Sander Dieleman, one of the authors or this method describes it as a way to “tackle the problem of predicting listening preferences from audio signals by training a regression model to predict the latent representations of songs that were obtained from a collaborative filtering model.” Of course we are not interested in the technical side of things but the fact that one of the leading It companies in the field of streaming service is looking to improve its recommendation system to lengthen the tail speaks volume.
The music streaming service market has hit a point where users need more and want to discover new things, sticking with a somewhat out-dated system like manual curation for Songza and Beats or manually tag attributes for Pandora or even audio content metadata for echonest doesn’t cut it anymore.
In conclusion, the actual technical method used for the recommendation system can lead to short or long tail. It seems the more mature an industry is, the more precise and well developed the recommendation systems are which leads to a longer tail.
Sources:
Show lesshttp://www.slideshare.net/MrChrisJohnson/algorithmic-music-recommendations-at-spotify
http://research.microsoft.com/pubs/115396/evaluationmetrics.tr.pdf
http://grouplens.org/site-content/uploads/evaluating-TOIS-20041.pdf
http://benanne.github.io/2014/08/05/spotify-cnns.html
Read more
Firstly we have to say that the recommendation system is a very useful tool to help the customer for finding the genre(s) and style of products they like. But it has to be used effectively. Indeed, like said in the article this concept of recommendation system is not new anymore and a majority of customers rely mainly of this kind of system in order to find other product related to their taste.
The fact is that the utilisation of this system is so fast with internet now that it will be (for a lot of persons) a waste of time to do research of their own. We can point the example of ebay,alibaba or even amazon like it was presented in the text. The recommendation system in the site are just under the image of the command and then really easy and fast to find.
However eventough the recommandation systems have their avandtages we can easily point out bad consequences of those kind of processes.
we can explain those consequences in 2 main points.
The first one is what we call “collaborative filtering”. This represents the process of filtering information by using techniques involving collaboration among multiple agents. The main point is to help the customer to find popular films, musics, books or whatsoever related to their taste. The fact is that this system doesn’t take into account a lot of works made by newcomers or simply unadvertised works. Indeed the fact is that the customers can only see 5 or 6 other works in this kind of recommendation system, this increases then the inequality factor between those kind of works.
The second point is called “collective intelligence”: which are the cognitive abilities coming from the interactions in a community between the members. This knowledge of those members is then limited because their vision is highly restricted due to the fact that they share the same vision of the environment (world) and are then not fully conscious of the surroundings.
We can apply this to the case because a lot of recommendation system create this kind of community and then close the possibility of the members to expand their vision in the genre they like.
We can then conclude that only using recommendation system will favor only few works and then don’t give the chance to a lot of potentially good contributions. The only use of recommendation system will then create a shorter tail for the market with a huge difference between blockbusters and other works.
sources:
Show lesshttp://en.wikipedia.org/wiki/Recommender_system
http://en.wikipedia.org/wiki/Collective_intelligence
http://www.businessdictionary.com/definition/collaborative-filtering.html
http://en.wikipedia.org/wiki/Collaborative_filtering
Read more
I do not believe recommendations make the long tail longer for every type of products. When it comes to books or movies, their tail is longer because of their creative/artistic aspect that sometimes lives forever. A good book remains good, even after the pick of sales, and the tail will, in my opinion, never end. I believe the question to ask in this case is if recommendation makes the tail ticker. The answer to this, in my opinion, is that it should. When I watch a movie, I like it, and I am recommended a similar one “You might also like …”, I most certainly purchase the recommended one, out of curiosity. A book or movie that was recommended to me and that I did not find extraordinary, I usually put the blame on the difference in taste, but never regret the decision to buy it.
But there are those products that become out of fashion/trends, and, in my opinion, no recommendation will make the tail longer for them. Let’s take for example mobiles phones: six years ago no recommendation system would inspire any type of consumer to buy a Nokia phone. Nokia missed the trend of touch-screen, went from being the world’s most respected mobile phones brand to being forgotten, and only reappeared slightly on the market with the Windows Phone. For this type of products, the recommendation system works only while they are in the pick period. A study on the influence of users’ review on Smart phones purchase suggested that people would not purchase a product that does not have reviews, but that the familiarity to the brand is more important that other customers’ reviews (Taskin, 2013).
Furthermore, I believe recommendation has to contain more arguments and explanations when the purchase decision is not easy (for important amounts of money). For me, the fact that the review is not anonymous gives it more credibility. What is more, the efficiency of a recommendation is also influenced by the previous experience with recommendation: if I bought something based on an online recommendation and I was not satisfied, I will have the tendency to not get influenced again by recommendations to avoid deception.
All in all, I believe recommendation has an impact on the tail, provided the tail was still positive.
Taskin, D., (2013). “Online users review: influences and influencers” (Master thesis). Louvain School of Management (Catholic University of Louvain), Louvain-la-Neuve.
http://www.theverge.com/2014/9/22/6826051/nokia-saw-the-future-but-couldnt-build-it
Show lessRead more
According to Anderson, our economy has shifted from a small number of highly popular products (the head) towards a high number of niche products (the tail). Indeed, in traditional economies, popular products were favored in comparison to niche products because the physical room on the shelves was limited and expensive. With the internet revolution, it has become in the contrary easier to build a business on the long tail given that inventories are not limited and that niche products are easier to find.
In his book, Anderson explains that 2 conditions lead to successful long tail business: making everything available and helping the customers to find the niche product. The second condition is fulfilled with the recommendation tools develop by players in long tail business. Recommendation systems design aims at making the consumer select the most important item for him.
What is at stake in this post is the effect of recommendation systems on sales. I will consider the 2 issues that are the most present in the literature: the impact on sales volume and sales diversity. Diversity is directly linked to the long tail phenomenon. Indeed, recommendation systems can have two distinct impacts:
Helping consumers discover new products that fit better with their preferences, hence increasing the diversity;
Reinforcing the popularity of already famous products, hence leading to more concentrate sales and thus less diversity.
It had been proved that recommendation systems have a significant impact on sales volume but the impact on the diversity of purchase seems to be more controversial.
This post was originally written in 2012 and I would like to update it with a paper written in the context of the 35th international conference on information systems in 2014 by Lee and Hosanagar.
In this paper, the authors analyze the impact of different recommender algorithm on both sales and diversity since, according to the authors, different algorithms have different impact on these two variables. The debate in the literature on the effect on diversity by recommenders systems is due to these differences. Recommendation systems are classified by Schafer, Konstan and Riedi in 6 types: the collaborative filtering (consumers who bought this item also bought that item), the content-based (according to your consumption history), the hybrid (combining the first two types), the social-network (what your friends bought), the popularity recommender (the most purchased items) and the recently-viewed (the most viewed items). Lee and Hosanagar analyze the impact of different recommender systems by measuring the diversity at the item and genre level with the use of the Gini coefficient mentioned in this post. The Gini coefficient has a range of values between 0 and 1 and measures the level of concentration diversity. A coefficient of 0 means that all products have the same amount of sales, while a coefficient of 1 means that a number of popular products account for most of the sales.
As explained before, the kind of algorithm used has an influence on sales volume and diversity. I am going to compare two of them: the collaborative filtering and the recently-viewed algorithm. One of the results of the study is that a collaborative filtering algorithm increases the sales volume and the individual diversity but decreases the overall aggregate diversity. Indeed, it seems that individuals discover a higher diversity of products through links provided to other items that can match with their preferences. However, the algorithm introduces a popularity bias since consumers move towards the popular items rather than towards niche products, which introduces a concentration in the diversity at the aggregate level, and so making the tail shorter. As far as the recently-viewed algorithm is concerned, it decreases the aggregate diversity and doesn’t influence the individual diversity. In fact, this algorithm, by showing you the recently viewed items, decreases the views on other less popular items and so decreases the diversity.
To conclude, in this comment, I have analyzed some issues related to niche and popular products and the influence of recommendations on them. However, one further investigation could be to analyze to what extent the previous analysis applies to products that are classified as neither popular nor niche item.
References:
Anderson, C. (2006) The Long Tail: Why the Future of Business is Selling Less of More. Hyperion.
Schafer, J. B., Konstan, J., & Riedi. (1999). Recommender systems in e-commerce. Proceedings of the 1st ACM conference on Electronic commerce, 158–166.
Lee, D., & Hosanagar, K., (2014) Impact of Recommender Systems on sales volume and diversity, 35th international conference on Information Systems.
Shanthi, K., & Kalimuthu, M., (2013) Improving diversity using contend based approach in recommender systems, Research Journal of Computer Systems Eengineering, 4.
Show lessRead more
Due to the exponential growth of the online market, the competition entered in this field. The owners of the sites need to attract the customers by providing attractive facilities. One of them is the Recommender Engines that attempt to recommend books, movies or articles based on our past actions. For example, Netflix offered a $1,000,000 prize to researchers who could improve the quality of their recommendations.
Gaurav Arora, Ashish Kumar, Gitanjali Sanjay Devre and Amit Ghumare analyse the recommendations system for movies to users. The existing system is based on individual users’ rating that can be useless for users who have different tastes. The new system compute the similarities between different users and the recommendations are based on these rating given by users of similar tastes. It provides more precise recommendations.
The system proposed by the authors it the batter system than any other existing system, according to them. “This system has added the positive features of existing systems and has overcome the drawbacks of existing systems. The system uses all the existing algorithms i.e. content based, context based and collaborative based algorithms. All these algorithms are combined to give more precise result.”(Page 768). There are four different modules developed:
1. Admin: this system allow to add movies in a database, view movies and update it;
2. Recommendation Engine: compute the similarities between the different users. It will recommend movie to a user;
3. Movie Web Service: this service allows user to rate movies, comments them and show the movie recommendations to the users;
4. Android user: the user can rate, comment and see movies.
This new system of recommendations is a hybrid one that improves the performance by overcoming the disadvantages of traditional recommendation system.
Based on: Gaurav, A., Kumar, A., Devre, G. S. and Ghumare, A. (2014). Movie recommendation system based on users’ similarity. International Journal of Computer Science and Mobile Computing, 3(4), 765-770.
Regarding the purchase, some surveys have been conducted. Here is a summary of the different results:
• According to Metail, 56% of consumers say they would be more inclined to use a retailer if it offered a personalised experience;
• An Infosys study on consumer behavior showed that 59% of shoppers who have experienced personalization in their shopping agree that it affects their buying behavior;
• A study launched by the eTailing Group show that recommendations system could improve purchases: 77% of online customers admit to have made additional purchases based on personalized product recommendations;
• 74% of online shoppers get frustrated with sites that shown them content that has nothing to do with their preferences or past buying behavior (study of JanRain).
Based on: Pratik Dholakiya, P. (2014). How do product recommendations influence buyer behavior? On line on Econsultancy site https://econsultancy.com/blog/65866-how-do-product-recommendations-influence-buyer-behavior/
Show lessRead more
My comment will be structured in three parts. I will first give my opinion about long tail theory, then about the recommendations systems and I will finish with the relation between the two previous.
From my opinion, the long tail theory doesn’t apply to every sectors or at least it doesn’t apply with the same intensity. In some industries, it’s true or more true that a large part of selling products distribution came from a large number of products each sold in low quantity. As Chris Anderson show it with the examples of Amazon and Netflix. I think that it’s also apply in the book sector more than in the music or film sectors. I see two distinct reasons. Firstly, production studios as Warner or Universal exert a massive brain-washing and they promote a lot a few products. They have a larger impact on people than publishing houses. Secondly, I guess people reading books regularly have more critical and like to make their own judgement. More than that, I think readers choose more wisely their books because it asks them an effort and it’s a real engagement where watching a movie take only 1 or 2 hours. For these two reasons I think that in the book sector the tail is longer, I mean that a larger part of the books sold doesn’t come from best sellers (in comparison with the movie sector).
Concerning the recommendations systems, I think they are useful and more present than before. Nowadays, people are more connected (by trivial relation sadly) and they are asking for the opinion of others. As an example, in the app-store or on Android, the apps are ranked by their quotation and the top ranked, the more recommended are the first you can see.
Now the question is to know if such recommendations systems impedes the long tail phenomenon or not… Taking the example of Netflix, I think they do. Indeed, I am more likely to recommend a movie or a series which is not very famous than the new blockbuster. For example, I would find it ridiculous to promote the 12 grammy’s series Breaking Bad, because I suppose every body knows it. I am more likely to recommend a new series for example. So from this supposition that recommendations systems promote unknown products, it surely impedes the long tail theory.
But can I say that it’s always true, probably not. It depends a lot of the type of recommendation system. On downloading websites, here is a ranking of the most downloaded products. If we accept that as an indirect recommendation system, we have to say that in some cases they reduces the long tail effect because it promotes the “best-sellers”.
In conclusion I would say that the long tail effect mainly depends of the type of products and of the way they arte purchased (on-line vs. off-line,…).
Sources:
Show less-http://archive.wired.com/wired/archive/12.10/tail.html
-http://fr.wikipedia.org/wiki/Longue_tra%C3%AEne
-http://fr.wikipedia.org/wiki/Liste_des_distinctions_de_Breaking_Bad
Read more
Nowadays, recommendation systems have an undeniable importance on Internet as they can be seen nearly everywhere: social networks where they help you find people you might know or pages on subjects that might interest you, online buying platforms (“you have bought that product, you could like this other product”), or simply on Youtube where you are proposed related videos and online news sites (where you can’t look at related articles). This mechanism can be very helpful to exapnd your personal tastes (Youtube, Spotify,…) or discover hings that you maybe didn’t even know existed, but I don’t think it has only positive aspects…
From my point of view, there are two situations and my opinion on recommendation systems will vary considerably depending on the actual situation.
Show lessThe first situation occurs when there isn’t money “in the balance”. For this situation, we have examples like Facebook or Youtube. In this case, I enjoy the presence of recommendation systems. I don’t say all the propositions are interesting (because actually a big proportion of them isn’t…) but it enables me to discover new things related to things I already like. And because money isn’t involved, I don’t really care about “trying” what is suggested to me.
On the other hand, we have all the recommendations about things you might be interested to buy. For this category, I feel much less at ease with recommendation systems. The main reason for this is that I don’t really trust online recommendations whether they are made by consumers or automatically because I feel like I’m being oriented in the direction the site wants (because a particular product doesn’t have the expected success or any other reason). As far as money is concerned, I strongly prefer to rely on the experience of people I know (and trust) or on my own research on the subject rather than on any recommendation with an online source.
Read more
The article is based on the theory of long tail. The theory was developed by Chris Anderson and describes the fact that the internet sites sell a high proportion of niche products in small quantities. This theory is often applied in e-commerce because the logistic costs are reduced and the centralization of products allows to manage products in small quantities.
Before, the economists referred to another law : the law of 20-80. The law of 20-80 led to place priority on customers that represent 20% of the total customers who contribute to over 80% of sales. With time this idea changed. Now the law of the long train is in place. With this law, the idea consists of paying attention to customers who represent 80% of the total mass and 20% of sales. Now, it is possible thanks to internet.
Instead of offering only the 20% of references that make 80% of sales, it is possible to be profitable by offering 80% of references that make traditionally 20% of sales. It is the revolution of internet, we can approach the niche markets. Firms sell small amounts of products but when you add the sales, they sell a lot. Amazon, the example of the article, makes 57% of sales with unknown references.
The reason being that the uncasual users want more than popular contents. The role of recommendations is to satisfy and retain these customers. For that, the site promotes niche contents of which the profit margin is more important than the popular ones and allow to diversify the offer.
After a few years, the results are not convincing. The long train effect is small and does not have any impact on the diversity. Today, we are witnessing the opposite phenomenon, some writers, musicians buy up all of demand. Now, we pay attention to « the superstar effect ».
According to Marcello Vena, responsible of numeric department RCS Mediagroup (books), the concentration of numeric market is the crux of the problem. The concentration of retail digital titles around a few global players makes the promotion of editorial diversity impossible. The monopoly is the problem. But Marcello Vena also says that the recommendations motors are truncated with their recommendations. To sell more they offer you another title of the same author than the original title of other books that you do not know. But it is the same problem with the social recommendations because the motor offers you the most recommendations shared and not recommendations with small views that could be an interesting discover. There is also a risk of media bias. Some recommendations systems could be associated with writers or musicians and generate a bias like we studied last week.
Nevertheless, I think that the recommendations are still important for e-commerce. For me, they replace the sellers in the physical shops. Indeed, in a physical store, the sellers help the customers to make their choices, but on the internet, there is no seller. Thus, the recommendation is used to help client for e-commerce.
So the recommendation system seems to act like the word of mouth. Indeed, we can consider the human beings like the best tool to communicate thanks to the interactivity and personal opinions. Before, the sites tried to have comments from authors or specialists in the field. But now, the sites try to attract recommendations from users and create a sort of community. This community votes and gives opinions. A book can be appreciated by the experts in the field and not be found interesting by the users. That is why, with time, the recommendations of users have emerged and become essential.
With the appearance of social networks such as Facebook, Twitter or GetBlue, the sense of community is reinforced. GetBlue allows users to indicate where they are but also recommendations for cultural goods (books, music,…). GetBlue gives notifications on what your friends do, what they listen or read and you can follow the activities of users with an option « friends to follow ». The friends give recommendations and the system allows to select its own sources of recommendations on the public profile of publications or tastes.
Finally, I think that everyone is sensitive to recommendations because these respond to our expectations. We want to be sure that the purchase that we will do is a good deal, we want to reassure ourselves that we made the right choice.
Sources :
• http://www.definitions-webmarketing.com/Definition-Longue-traine
Show less• http://www.conseilsmarketing.com/autres-conseils-marketing/les-business-modeles-la-longue-traine-walkcast-plan-marketing-partie-65
• http://www.markentive.fr/blog/4-raisons-dutiliser-la-longue-traine-pour-votre-strategie-content-marketing/
• http://www.cairn.info/revue-d-economie-politique-2010-1-page-141.htm
• http://www.cairn.info/revue-reflets-et-perspectives-de-la-vie-economique-2008-2-page-95.htm
• http://lafeuille.blog.lemonde.fr/2014/06/24/pourquoi-la-longue-traine-ne-marche-pas/
• http://tech-insider.org/file-sharing/research/acrobat/0807.pdf
• http://www.longtail.com/about.html
Read more
I did not find any good readings proving or not what was only assumptions at the moment the article was written, but this comment will show some new ways marketers found to make us buy what we did not know whe needed.
Indeed, the article only spoke about « automatized » recommendations systems (by that I mean not taken care by an employee of the company website you are visiting), that directly appears while checking the page of a product. But same year as the publication on the blog was published, another one was posted by « Fortune » (link at the end), which stated that « Amazon also doles out recommendations to users via email. […] the company provides some staffers with numerous software tools to target customers based on purchasing and browsing behavior. But the actual targeting is done by the employees and not by machine. If an employee is tasked with promoting a movie to purchase like say, Captain America, they make sure customers who have viewed other comic book action films receive an email encouraging them to check out Captain America in the future. » I do not have information to assess if that sytem is still used nowadays, but I found it quite interesting to see that it was not only « computerized ».
Also, with the well known success story smartphone had, it seemed pretty obvious that it will have a rôle to play in the recommendation systems. In 2013, a project was being studied and tried, consisting on the implementation of an eye tracking system directly implemented in ordinary glasses, and connected to a smartphone in order to play the same role as an ordinary recommendation system, and also help the consummer in his decision process. According to the article i also linked in the end of my comment « instead of learning from click behaviour and past customer ratings, as it is the case in the e-commerce setting, the mobile RA (recommendation agent) learns from eye movements by participating online in every day decision processes. »
In a first pilot study they did with five randomly chosen participants in a supermarket, they tried to get a first impression of the user’s behavior in a situation where he would need such technology, and his acceptance of the technology. First results show long eye cascades and short fixations on many products in situations where users are uncertain and in need for support. They also found a surprising acceptance of the technology itself throughout all ages (23 – 61 years), but at the same time, consumers expressed serious fear of being manipulated by such a technology. For that reason, they strongly prefer the information to be provided by trusted third party.
Since the article do not give a name to that project, i wasn’t able to see the actual state of the project, but I think that some Google Glasses features may have been inspired by that kind of project.
In conclusion, recommendation systems are getting more and more sophisticated and use new means in order to assess which product would fit each customer, so I do not really doubt that, if it is not already the case, marketers and developers will soon find a way to sell even the niche products to the customers that would be interested by them, and who would never had been aware of their existence without the RS.
– http://fortune.com/2012/07/30/amazons-recommendation-secret/
Show less-http://pub.uni-bielefeld.de/luur/download?func=downloadFile&recordOId=2578942&fileOId=2602478
Read more
I did not find any good readings proving or not what was only assumptions at the moment the article was written, but this comment will show some new ways marketers found to make us buy what we did not know whe needed.
Indeed, the article only spoke about « automatized » recommendations systems (by that I mean not taken care by an employee of the company website you are visiting), that directly appears while checking the page of a product. But same year as the publication on the blog was published, another one was posted by « Fortune » (link at the end), which stated that « Amazon also doles out recommendations to users via email. […] the company provides some staffers with numerous software tools to target customers based on purchasing and browsing behavior. But the actual targeting is done by the employees and not by machine. If an employee is tasked with promoting a movie to purchase like say, Captain America, they make sure customers who have viewed other comic book action films receive an email encouraging them to check out Captain America in the future. » I do not have information to assess if that sytem is still used nowadays, but I found it quite interesting to see that it was not only « computerized ».
Also, with the well known success story smartphone had, it seemed pretty obvious that it will have a rôle to play in the recommendation systems. In 2013, a project was being studied and tried, consisting on the implementation of an eye tracking system directly implemented in ordinary glasses, and connected to a smartphone in order to play the same role as an ordinary recommendation system, and also help the consummer in his decision process. According to the article i also linked in the end of my comment « instead of learning from click behaviour and past customer ratings, as it is the case in the e-commerce setting, the mobile RA (recommendation agent) learns from eye movements by participating online in every day decision processes. »
In a first pilot study they did with five randomly chosen participants in a supermarket, they tried to get a first impression of the user’s behavior in a situation where he would need such technology, and his acceptance of the technology. First results show long eye cascades and short fixations on many products in situations where users are uncertain and in need for support. They also found a surprising acceptance of the technology itself throughout all ages (23 – 61 years), but at the same time, consumers expressed serious fear of being manipulated by such a technology. For that reason, they strongly prefer the information to be provided by trusted third party.
Since the article do not give a name to that project, i wasn’t able to see the actual state of the project, but I think that some Google Glasses features may have been inspired by that kind of project.
In conclusion, recommendation systems are getting more and more sophisticated and use new means in order to assess which product would fit each customer, so I do not really doubt that, if it is not already the case, marketers and developers will soon find a way to sell even the niche products to the customers that would be interested by them, and who would never had been aware of their existence without the RS.
– http://fortune.com/2012/07/30/amazons-recommendation-secret/
Show less-http://pub.uni-bielefeld.de/luur/download?func=downloadFile&recordOId=2578942&fileOId=2602478
A bit scary… We’ll have the opportunity to discuss the issue of privacy in a coming lecture.
Read more
A recommender system learns from a customer and recommends products that she will find most valuable from among the available products. In this comment, I took the Spotify case to show how the recommendation system has evolved and that it could be based on other than historical data and also try to show if this new technique would make the tail longer or not.
Spotify is well known to rely on collaborative filtering approaches I.e. the process of filtering for information or patterns using techniques involving collaboration among multiple agents viewpoints, data sources, etc. This technique determines the user preference from historical usage data. Note that the collaborative filtering techniques do not use any of information concerning the items that are being recommended. However given this last characteristic, popular items will be easier to recommend than unpopular items since there is more usage data available for them making recommendation systems predictable. Moreover, the heterogeneity of content with similar usage patterns i.e. having an original song and a remix song from the same artist and having the recommendation system only taking the original song narrowing the choice of the system. However, the most important problem with the collaborative filtering technique is given that new and unpopular songs don´t have any usage data to be analyzed they will not be recommended this Is called the cold-start problem.
Thus, given these different difficulties, Spotify proposed another approach based on the content information itself. The company is trying to integrate other sources of information into their recommendation scope to tackle the problems explained previously. The other source of information could be: tags, artists, lyrics or reviews from the web and the audio signal itself. The audio signal could be the genre of the music or the instruments used. Engineers at Spotify wanted to develop recommendation systems out of audio signals. In order to do so, they made a regression model to predict latent representations of songs that were obtained from a collaborative filtering model to actually predict the representation of songs that had no historical usage data. We will not cover the details of the regression in this article but they could be found in the ´´References´´section. The result of this technique, was a built up of different playlists based on the instruments used in songs, it filtered also songs that had harmonic singer voices. This new algorithm made the platform recommend new and unpopular music which wasn´t the case for the collaborative filtering technique, making Spotify able to recommend the right music to the right public.
In conclusion, we can clearly see that these new techniques would make the long tail even taller by having new kind of products recommended to the public and by lowering the probability of listening to the same songs over and over again. This technique could also be used to other products like books but of course the variables taken into consideration would be different.
References:
Show less– http://en.wikipedia.org/wiki/Collaborative_filtering
– http://research.microsoft.com/pubs/115396/evaluationmetrics.tr.pdf
– http://www.slideshare.net/MrChrisJohnson/algorithmic-music-recommendations-at-spotify
– http://www.slideshare.net/erikbern/collaborative-filtering-at-spotify-16182818
– http://papers.nips.cc/paper/5004-deep-content-based-music-recommendation.pdf
Very interesting: Where economics and computer science meet.
Read more
In the years following Anderson theory of the long tail the literature on the subject seemed to agree that the success of retailers such as Amazon.com and Netflix is largely attributed to the “ long tail” phenomenon. Indeed niche products, that are not available at limited-inventory competitors, generated a significant fraction of total revenue in aggregate. However most of existing recommender systems were not able to recommend tail products due to the data sparsity issue. Based on that realization interested parties then focused on developing effective algorithms to recommend long tail items.
Furthermore, in his 2010 book “ Music Recommandation and Discovery”, Oscar Celma assessed that Anderson’s success theory may have been a failure but that his core idea still is an interesting way to explain the web-provoked changes, in terms of the availability of all kind of products (from hits to niches). Later he states that a recommender should focus on promoting the tail of the curve by providing relevant, personalised novel recommendations to its users. It would imply smoothly interconnecting the head and mid regions with the tail, so the recommendations can drive interest from one to the other.
In 2014 more and more tend to think that this long tail theory might have been an utopia. Daniel Kaplan said “The long tail effect is so small that it can cause no change in the market and has no impact on cultural diversity”. Some even develop contradictory theories for instance the superstar effect where blockbusters sell more than ever. Thierry Crouzet went as far as saying : ‘We are witnessing the opposite phenomenon. The curve has certainly lengthened, but it has widened enormously. It has some titles that sell a lot, and all others who are fighting for crumbs.”
In her book “ Blockbusters” Anita Elberse also shows that even companies supporters of the long tail have changed their tune. On one hand, Amazon seeks above all to subcontract to other players, and is looking with Kindle to set foot in the editing and production. On the other hand, now that Netflix business model is migrating towards online video ( Netflix used to primarily send DVDs at home at the time of Anderson’s interview), the company is obliged to secure the license of his films where previously it only had to buy DVDs. However, these licenses become increasingly expensive. Netflix evolved into a studio and invests hundreds of millions dollars in the production and promotion of its own shows like “House of Cards” or “Orange is the new black” buying into the blockbusters theory. Youtube slowly seems to follow the same path.
So my conclusion would be that even if the theory of the long tail is valid it doesn’t pass the test of reality. They are economic factors that make its realisation impossible. Another thing is that we are constantly faced with too much information, too much choices on the web and it can be a source of anxiety. So bestsellerization and blockbusters are what work right now as they also answer the need for easy access to novelty.
SOURCES:
Celma, O. (2010). Music Recommendation and Discovery. Berlin Heidelberg : Springer-Verlag.
Show lessChen, Y., Wu,C., Xie,M.,& Guo, X. (2011). Solving the Sparsity Problem in Recommender Systems Using Association Retrieval. Journal of Computer, 6(9), 1896-1902.
Yin, H., Cui, B., Li, J., Yao, J., & Chen, C. (2012). Challenging the Long Tail Recommendation. Proceedings of the VLDB Endowment, 5(9), 896-907.
http://lafeuille.blog.lemonde.fr/2014/06/24/pourquoi-la-longue-traine-ne-marche-pas/
http://www.slate.fr/tribune/84585/longue-traine-blockbusters
Read more
The idea of sharing information between each other is not something quite new. Take for instance the caveman; they already used information from others. One of them was hungry and saw a nice red fruit. His first thought is that he wants to eat it but it might be poison. As you are not as hungry as your fellow, you will let him taste first and see if it is safe. Indeed if he survive you will get the information that the fruit is eatable and you will start to do the same. Thus, you learned the lesson without having to eat the fruit yourself. Moreover, critics of movies or books have existed for a long time. Many of us like to have others’ opinion to have an idea of what will be good or bad.
Nowadays, technology has evolved and it allows us to share and communicate our opinion, our preferences and our tastes with many more people. In recent years, recommender systems appeared in a large panel of application (Google, Amazon, YouTube, etc). E-commerce applications have found this system very useful in order to deal with the information overload problem. However, generating recommendations needs to be improved. Indeed, it suffers from data sparsity and cold start. The first issue concerns the difficulty to collect enough reliable data as the active users rates a small portion of items. Plus, the cold start refers to the difficulty of finding the right information to please the cold users as they rate a small number of items. Few solutions are developed in the link below.
Relying on personalized information, many Internet users are making important decisions. Most of the time those people are not even aware that the information to them. Nevertheless, it is still hard to find good update on the positive and negative effect of recommendation system. Knowing these consequence will help to minimize the negative ones and maximize the positive ones.
References:
https://www.coursera.org/learn/recommender-systems/lecture/bWRen/introduction-to-recommender-systems
Show lesshttp://misrc.umn.edu/wise/2014_Papers/134.pdf
http://www.rasaneh.org/Images/News/AtachFile/7-10-1393/FILE635553493145844568.pdf
Categories