Netflix is one of the most prominent, using big data to segment its content into over 76,000 categories -- some broad, others very specific. This segmentation breaks down film content into component parts and helps fuel a recommendation engine, where Netflix can say with some certainty that because you liked movies A and G, you'll love N, Q, and Y.
The history of Netflix's algorithm is a fascinating illustration of analytics evolution.
Back in 2006, Netflix launched a competition to see if anyone could come up with a new way to sort and recommend content to consumers -- one that was fully automated, efficient, and accurate. Over the next few years, it awarded cash prizes (around $50,000) to the most impressive ones, eventually handing over a million dollars to a group that managed to improve the star-rating recommendation accuracy by 10%.
But Netflix never even used it. It had already grown beyond that idea, instead looking to take into consideration things like what your Facebook friends watch, and what other customers with similar viewing histories also enjoy. Ultimately, this led to the relatively long and specialized categories that we have today.
Netflix also uses analytics to drive its original programming. It greenlit two seasons of House of Cards (committing $100 million in the process) simply because of data that said David Fincher -- who produces the show and directed the first two episodes -- is popular on the streaming service, as are Kevin Spacey and the original British series. And look at the returns; the show has not only done well on-demand, but it has received multiple awards across both seasons.
The Netflix story shows how one company uses analytics to recommend more of its product to its customers, but other analytics firms are looking at TV viewing on a broader scale.
One British company, named SecondSync, has begun analyzing social networking interactions during TV broadcasts to gauge public opinion on issues from political debates to soap opera preferences. According to the owner of SecondSync, it operates by using natural language processing of TV shows to generate keywords, which are then searched for on Facebook.
Of course, with varied privacy levels, this isn't an easy task, which is why SecondSync gets in bed with Facebook itself to get the job done:
"Facebook used these search terms to calculate anonymised statistics of total discussion volumes for the search terms, aggregated at telecast airing window, daily and minute level," SecondSync explains in a whitepaper. "In addition, Facebook used these search terms to randomly select a small number of anonymised public Facebook posts that SecondSync uses to test the accuracy and quality of its search terms."
This sort of information could be used by advertisers to find out which shows have the most engaged viewers and at which point in the broadcast they're most engaged, or at which point to send out a promotional tweet related to the show.
A handful of companies are already lined up to receive the data, too. SecondSync has deals with the BBC, Channel 4, and Twitter in the UK, as well as NBCUniversal.
What SecondSync is doing for Facebook, Bluefin Labs had been doing with Twitter for over six months, eventually being acquired by Twitter in February. Nielsen's SocialGuide is another Twitter TV analytics service, which looks to differentiate itself by offering daily and weekly rankings for programming.
So it looks like the future is not only bright for analytics firms, but consumers, too. As time goes on, we're not just going to have near-unlimited, on-demand choice with our media viewing, but the content will be (robotically) hand-picked based on our viewing preferences.
What do you think, members? Has Netflix's recommendation engine turned you on to new favorites? Are you happy with the old rabbit ears? Share your opinions below.