George Dafermos

His Blog

Collaborative Filtering

Collaborative filtering: one of the same or marketing from the market's perspective

In a thorough study of the leading e-tailer, Sandeep Krishnamurthy (2002) chronicles Amazon’s business evolution and attributes its success as a bookseller to a wide spectrum of factors, most important of which being that Amazon turns its inventory faster than bricks-and-mortars, has reduced the book return rates, passes on cost savings in the form of cost reductions to consumers and enables unknown authors - ‘outsiders’ – to reach a global audience. But most crucial above all, in Krishnamurthy’s view, is Amazon’s vision: to become Earth’s most customer-centric company and biggest store; essentially a “platform on which you can do a lot of things” as’s founder, Jeff Bezos puts it (Hof 2000).

Amazon has not revolutionalised the bookselling industry because it offers such a vast collection of books, many of which are cheaper than at high-street bookstores. Neither because it has laid the ground for cross-selling opportunities by deploying highly sophisticated CRM technologies which track and record every single customer click. What is so special about is that it invites readers to send reviews of books and to rate them on a five-star scale along with a commentary expressing their thoughts and opinions on the book. Authors have the right to reply and other reviewers can comment on how useful the review was to them but they cannot change the review.

This process or technology – called collaborative filtering, weblogs or social navigation - has drawn quite some attention[1]. “Among marketers, the hope is that such computerized recommendations will increase demand...It means that people might read more, or listen to music more, or watch videos more, because of the availability of an accurate and dependable and reliable method for them to learn about things that they might like” (Gladwell 1999) former Microsoft executive and co-founder of Net Perceptions (, a firm specialising in collaborative filtering, Steven Snyder says. According to Net Perceptions’ website, “This is an example of where technology really is helping us provide better service and sell more — and that is the name of the game”. And e-commerce specialists Fingar, Kumar and Sharma (1999) claim that “collaborative filtering is at the heart of one-to-one personalisation and community building...the technology can be used to collect ratings about items available in an I-Market…Once an I-Market has collected a critical mass of ratings, it can respond to customer enquiries with recommendations that are tuned to customers’ preference patterns. Customer-driven collaborations and buying patterns tracked by collaborative filtering facilities can also be used to achieve advertising precision never before possible – the ultimate tool for up – selling and cross-selling”.

While Fingar, Kumar and Sharma (1999) regard ‘collaborative filtering’ as a community building and personalisation tool that aims at seizing cross-selling and up-selling opportunities, Malcolm Gladwell is far more radical. He believes that ‘collaborative filtering’ gives the chance to under supported (in terms of traditional marketing promotion) ‘outsiders’ to beat the ‘blockbuster’. It gives the opportunity to the ‘no-brand-name’ to outperform the established brand (Gladwell 1999).

Gladwell (1999) explains that collaborative filtering “works as a kind of doppelgänger search engine. All of us have had the experience of meeting people and discovering that they appear to have the very same tastes we do—that they really love the same obscure foreign films that we love, or that they are fans of the same little-known novelist whom we are obsessed with. If you and your doppelgänger love the same ten books, chances are you'll also like the eleventh book he likes. Collaborative filtering is simply a system that sifts through the opinions and preferences of thousands of people and systematically finds your doppelgänger--and then tells you what your doppelgänger's eleventh favorite book is”.

Chris Locke is sure that collaborative filtering is neither just a “sophisticated version of “Would you like fries with that?” [nor a technology which simply] automates cross-selling and up-selling opportunities” (Locke 2001: 113). As he says, the view looks radically different from a high vantage point. “Collaborative filtering works bottom-up by feeling out the edges of emergent micromarkets based on personal tastes and interests, in effect defining potential online communities” (Ibid). Locke reminds us that the ideal form of marketing is world-of-mouth or ‘markets are conversations’ and all these kids that rush to the Harry Potter dedicated pages at do not see ‘collaborative filtering’; they see ‘conversations’ among people sharing the same interest: Harry Potter. So, “They are beginning to talk to one another. They understand. No one had to explain it to them” (Ibid:109). So, “what’s the commercial benefit?” The fact that the forthcoming but yet unpublished Harry Potter book is from now amazon’s number one bestseller should be indicative of these conversations’ commercial value.

Locke contends that “’s real innovation was to create a marketplace where customers, not advertisers and marketers, could access the value of products…the really interesting marketing action is not how this information is being used to pitch products – “would you like War and Peace with that?” – but in how it’s being used to hook people up and get them talking with each other. “Hey, I just read War and Peace, and man, I gotta tell ya, this Tolstoy dude rulez!”” (Ibid:114). And does whatever possible to enrich its relational space through member pages, wish lists, purchase circles and discussion boards; essentially encouraging online communities to coalesce around topics of interest, such as books, movies and music.

Still not convinced? A friend recently told me that he’s never paid much attention to customer reviews at amazon since he reckons it’s all fake (however I always do) so I owe to provide him with another story: the story of Andrew Sullivan’s “Unfit to Print” Weblog. Andrew Sullivan, former editor of New Republic, is a blogger. And his blog has an online following well exceeding 500,000 people. So what Andrew Sullivan does in order to stimulate some discussion between him and his readers is to choose a book every month and encourage his readers to discuss it with him throughout the month by means of posting notes and opinions, raising objections, asking questions, recording their musings and reflections at his Book Club which is a category under his weblog. Apparently, the Book Club is not only a category of Sullivan’s Weblog but a rather vibrant community as within days from founding the book club, his first book-to-be-discussed, Robert D. Kaplan's Warrior Politics, skyrocketed up the Amazon charts into the top five. And the last selection of the Book club, Christopher Hitchens' new book, "Why Orwell Matters," went in a few hours from 1074 to number 3.

My friend can rest assured that none of Sullivan’s readers are being paid for contributing reviews, they’re just in it for the fun of blogging along with others. For the fun of discussing a book those others too have read. Intelligently, Sullivan came up with a way to make some money out of these book discussions: for every book that’s discussed at his weblog’s book club, there’s a link to the amazon’s online bookshelf and Sullivan gets a 15% percent on sales if you buy the book by following the link to Amazon from his weblog.

One might counter that what Sullivan’s Book club does is nothing new as it is just an example of a prominent opinion leader and that’s what opinion leaders do: influencing the masses into making certain decisions or buying certain products. Of course, this is a top-down process of ‘marketing coercion’, from the opinion leader to the masses. I would certainly agree with the assertion that Andrew Sullivan is an opinion leader. But I firmly believe that it’s the power, the passion and the energy embedded in his weblog that boosted the sales of Kaplan’s book rather than the compulsive opinion of a single man. Had not been for the weblog, all Sullivan could have done, as an opinion leader, is to suggest a book for others to buy. That’s not what he does though. What he does is to invite others to join the conversation and I am sure that his readers do not see an opinion leader but someone who’s keen on having a chat with them.

So, you may now think that weblogs work for ‘pushing’ and boosting the sales of books, records, films and other products which Malcolm Gladwell calls “taste products”, but that need not mean that weblogs will also work for companies other than those in the bookselling or music business. OK then, you’re not selling books or CDs or videos. Let’s start from betting your company on a weblog.

[1] Bluntly speaking, the main difference between the collaborative filtering system that Amazon has implemented and any ordinary weblog is that in the former case the Amazon system generates computerised recommendations while in the latter people are responsible for navigating through the opinions and views expressed by other people. In short, a typical weblog does not make recommendations but nonetheless Amazon’s system is essentially a weblog. --

your Amazon recommendations - Jahsonic - early adopter products

Managed Hosting by NG Communications