Netflix Recommendations Getting More Personal (NFLX)

In the first of a two-part post, Netflix Inc. (NASDAQ: NFLX) describes how it has fine-tuned its recommendation algorithms over the past couple of years and indicates the direction the company will follow in the years ahead.

The story starts with the Netflix Prize, which the company offered in 2006 to anyone or any group that could improve the company’s recommendation scheme for DVD rentals. It has evolved into a far more complicated system over the years, primarily because streaming video changed the rules. Here are a couple of interesting bits from the blog post:

For DVDs our goal is to help people fill their queue with titles to receive in the mail over the coming days and weeks; selection is distant in time from viewing, people select carefully because exchanging a DVD for another takes more than a day, and we get no feedback during viewing. For streaming members are looking for something great to watch right now; they can sample a few videos before settling on one, they can consume several in one session, and we can observe viewing statistics such as whether a video was watched fully or only partially.
…
We have adapted our personalization algorithms to this new scenario in such a way that now 75% of what people watch is from some sort of recommendation.
…
Now it is clear that the Netflix Prize objective, accurate prediction of a movie’s rating, is just one of the many components of an effective recommendation system that optimizes our members enjoyment. We also need to take into account factors such as context, title popularity, interest, evidence, novelty, diversity, and freshness.

Providing movie recommendations requires that Netflix have a lot of personal information about the viewing habits and tastes of its subscribers. Perhaps the company’s goal is to figure out a way to monetize that information, so that when it recommends a movie and a subscriber in fact watches it, Netflix can either extract a payment of some kind from the movie’s copyright owner or use its sophisticated algorithms to cut better deals for content. Whether that’s good or bad for subscribers is an open question.

Paul Ausick