Netflix 用大數據揭客群迷思

Mia，用大數據揭客群迷思，Netflix 產品副總：地區、年齡和性別一點都不重要！，Inside，2016/4/1

是同一套演算法，對不同族群幾乎一視同仁，而且 Netflix 產品副總 Todd Yellin 認為，重點在於品味，而非常用的性別、年齡或地區因素。「地區、年齡和性別？我們把這些東西都丟到垃圾堆了，」他說。

Brian Barrett, Netflix’s Grand, Daring, Maybe Crazy Plan to Conquer the World, Wired, 2016/3/27.

“There’s a mountain of data that we have at our disposal,” says Todd Yellin, Netflix’s VP of product innovation. Netflix has a well-earned reputation for using the information it gleans about its customers to drive everything from the look of the service to the shows in which it invests. “That mountain is composed of two things. Garbage is 99 percent of that mountain. Gold is one percent… . Geography, age, and gender? We put that in the garbage heap. Where you live is not that important.” ...

Freedom from worrying about signals like geography, gender, and age allows Netflix to hone its recommendations more sharply, and against less obvious criteria that for competitive purposes it doesn’t divulge. It’s what lets Netflix group its titles into a couple of thousand “clusters” based not on where people live, but what they like. 

Netflix assigns each subscriber three to five of these clusters, weighted by the degree to which each matches their taste. “When you have more than 75 million people around the world, you can get really specific about who’s your taste,” says Yellin.

For all the thousands of titles in the company’s catalog, the average member only sees 40 to 50 options in a typical visit. Clusters, which can comprise anywhere from tens of thousands to millions of subscribers, are what help ensure that those members see the right 40 to 50. You also aren’t limited strictly to your cluster assignments—maybe you’d love Peaky Blinders, given the chance—as the algorithm occasionally offers glimpses outside your silo. 

“We used to be more naive. We used to overexploit individual signals,” says Yellin. “If you watched a romantic comedy, years ago we would have overexploited that. The whole top of your screen would be more romantic comedies. Not a lot of variety. And that gets you into a quick cul-de-sac of too much content around one area.” 

Getting that algorithmic balance just right is important to Netflix on a day-to-day basis, but absolutely critical when entering a new market. If geography were weighted heavily, if it mattered as much as we might assume, Netflix would have had a much harder time knowing what to show a person in a region it had never been before. Instead, it can use other proprietary signals as its guide.