Amazon data mining to find customer tastes
November 3, 2013 | 10:54 pm
The Wall Street Journal has an interesting piece on Amazon’s original television programming pilot selection process, in which it winnowed down a long list of pilots using user ratings and feedback to pick the ones that were most likely to succeed to base series on. The WSJ compares this approach to that taken by Netflix, who didn’t even require a pilot for shows like Orange is the New Black or House of Cards.
While this doesn’t have a whole lot to do with books, it does bring up one of the most crucial aspects of Amazon as a process. I found the piece linked from The Passive Voice, which points out that Amazon has built its success on data mining to understanding what its customers want, rather than trying to dictate from on high what they should want. Writes blogger Passive Guy:
[Amazon] has no pretenses about being a taste-maker that dictates what is customers should like. Nor does it appear to be particularly interested in what self-appointed arbiters of various artistic endeavors think. That sort of attitude drives the arbiters mad.
Those of us of a certain age remember what TV used to be like. It was largely one-way. The networks made, and we consumed. There were Nielsen Ratings and surveys and studies and things, but they weren’t exactly fine-grained data. The networks muddled on and did the best they could. And by the same token so did book and magazine publishers, record labels, and every other producer of media.
Then along came the Internet. And the most successful companies on it, like Amazon and Google, have figured out that they can do a lot better job of selling stuff if they shape the stuff they sell to fit what the customer wants, rather than try to reshape the customer. And for the first time they can actually know what the customer wants, instead of making guesses, because they have access to so much more information about the customers than they used to.
Every time we buy something, every time we click a link on their site, it’s recorded, tracked, collated, and analyzed. And while this does have the darker side of potential privacy invasion, it also means that they can sell us exactly what we want, because they can figure out exactly what it is. And that’s what Amazon does. That’s one of the reasons why it’s done so well with e-books, because it has all that sales data it can use to tailor its offerings—and is notoriously stingy about sharing it with anyone else. Lacking the same kind of popular storefront themselves, the publishers aren’t as able to keep up—though even they nonetheless have more data than they used to.
Amazon, Netflix, and other streaming video offerings seem to be putting the traditional TV networks in the same place e-books put traditional publishers. While you don’t quite get “self-publishing” with these outlets, you do get independent programming that could entice people to watch the shows there and give traditional TV a miss. It’s going to be interesting to see how they deal with that.