As a digital platform, Netflix shares the same two problems that most digital platforms face: how to retain customers and how to monetize their offerings. Netflix’s path has had a few significant bumps. Starting out as a DVD rent-by-mail service, Netflix wasn’t originally an online platform. The company was founded in 1997 by Marc Randolph and Reed Hastings as an alternative approach to DVD rental, which at that time was dominated by local storefront rental companies. By 2000, Netflix moved to a flat-fee unlimited rental policy without due dates, late fees, or shipping and handling fees. They were learning to expand by offering what the market wanted, and what local shops were not delivering.
In 2007, Netflix began offering online video-on-demand. It was the beginning of their streaming video service. Netflix online service grew while rent-by-mail DVD sales fell and then, Netflix got into producing original content. First they licensed and distributed independent films, which served as their springboard into content production. All along, from 1998 onward, Netfix collected and compiled information about individual user’s entertainment preferences. This information is at the very heart of the company’s growth model. Their strategy is to attract new customers and then retain them by offering a continuous stream of movies and shows that fit the new subscriber’s taste and preferences. Thus by cultivating satisfied customers Netflix opens the opportunity to raise prices and improve their profits. That’s the basic model.
In 2011, the model failed. Netflix attempted to make two business moves at the same time. First they wanted to separate the DVD mail division into an independent company and secondly they wanted to raise the price on their monthly online cloud movie service. The strategy may have been good but the tactics for execution were disastrous. Within the first week over 70 thousand members quit the online movie service, so soon there after Netflix called off their price increase. That was in 2011, now in 2014 a new price increase has been announced, and Netflix is confident their strategy and tactics are in alignment this time around.
Success at raising prices is all about offering a service that people want even if the prices are raised. The service has to be something people want to do more than they want to do other things with that money. As Reed Hastings, Netflix’s CEO said, “What we’re really competing for quite broadly is people’s time.” For people to give you their time and pay to give it to you, the service has to already know the customer’s likes and dislikes. This is where Netflix has excelled in the last few years. A recent Atlantic Monthly article reverse engineered the Netflix movie algorithm in order to get an idea how Netflix goes about offering each customer a unique selection of movies, documentaries and TV shows, custom selected for that customers likes and dislikes.
Designing Movie Preference Algorithm
The algorithm is rigorously aggressive at micro-defining entertainment categories. The broad categories used in libraries to sort out books is way too general. Netflix’s research team broke down categories into almost 77 thousand genres. To retain subscribers, Todd Yellin, Netflix VP of Product Innovation says, “Members connect with these genre rows so well that we measure an increase in member retention by placing the most tailored rows higher on the page instead of lower.” Clearly, the better Netflix knows your likes and dislikes the more likely you are to stick around.
One of the first steps in building this algorithm was an open competition for a million dollar prize to anyone who could improve on how Netflix makes suggestions about what a viewer might like to watch based on what they have already watched. The prize was eventually won, but the winner’s solution was never used. Instead Neflix hired Todd Yellin to develop an algorithm based on very specific tagging of movie endings. The movie ratings were constructed around a one to five number scale. The highest rating was a five and the lowest a one. The tags are broken into microtags, which get extremely specific about the movies content. From these tags a whole syntax is built for the genres that are used to describe movies. So by working with tags like: feel-good, foreign comedy, from the 1970s, you can build a very accurate genre description of a movie.
The algorithm plays a vital role in the overall business strategy for retaining subscribers at a level which allows price increases, but there’s more to it than merely offering movies a subscriber might like. The algorithm also allows Netflix to know what they should invest production money towards. Netflix has gone beyond merely being a purveyor of Hollywood and independent movies. At this point the company has several original franchises of their own, such as the House Of Cards, although the House Of Cards was first a British television series. When Netflix launched their Americanized version of House Of Cards they were producing what their research told them would be a smash success. If there’s any way of evaluating their research it’s probably by looking at how their own productions fare in the marketplace. This far they are dong very well – The House Of Cards is the smash hit they predicted.
Yet another side to their research is the way they use movie micro-descriptors as hooks on any individual subscriber’s emotional taste in entertainment. These micro-descriptors are how suggestions are made, but it goes well beyond just matching entertainment in a particular genre to a subscriber’s tastes. With the micro-descriptors they can offer movies and shows based on just the foundation micro-descriptor. For example, if a subscriber is known to like “feel-good foreign comedy from the 1970s,” why wouldn’t that same subscriber like other genres of comedy? They are likely to, but how to go about selecting from such a broad descriptor as “comedy”? Here’s where a working knowledge of that subscriber’s choices help fill in the blanks about how else to hone in on the particular types of comedy that will be most liked by this subscriber. If they also like newsroom dramas, then it’s a good bet they will like newsroom comedies. It’s a matter of testing a variety of choices and then offering what the combined micro-descriptors indicate.
By going deep into entertainment genres and micro-descriptors Netflix has gained control of their digital niche. Their algorithm can be imitated, but with their first in market advantage and their newly tested ability to raise prices as needed, they have control that late-comers are not likely to get hold of.