WATCHING YOU
NETFLIX HAS ASSEMBLED A TEAM OF MOVIE AND TV EXPERTS TO FIGURE OUT YOUR VIEWING HABITS.
Netflix Inc. likes to flaunt its ability to recommend what vi e wers s hould watch next, but the way it tries to predict the next binge-worthy show keeps getting more complex as it tries to get more people to view even more content.
The underlining problem with recommendations in general is genres. The streaming video service isn’t against them, but it finds they are generally too broad to help users find new content. Why settle for “drama” when you can have “imaginative time travel movies from the 1980s” instead?
Changing the way titles are categorized by becoming much more specific helps Netflix recommend quirky shows and movies that users may not find otherwise, and allows it to identify “gateway shows” to understand the path someone takes to get from, say, the gritty House of Cards to the lighthearted Unbreakable Kimmy Schmidt.
To do that, Netflix has assembled a team of movie and TV show experts, many of whom have been filmmakers, critics and writers, to place tags on content.
“They watch every piece of content and then they go through and tag them with everything to cover every aspect around that title,” Todd Yellin, Netflix’s vice-president of product innovation, said in an interview. “We can tell you how much violence or sex it has, does it have a dark ending or a happy ending... does it have a chimpanzee in it, does it have a corrupt cop, or does it have a corrupt cop who happens to be a chimpanzee.”
The average Netflix subscriber looks at 40 to 50 titles before deciding what to watch, and the company said that’s too many since it wants people to spend more time consuming content and less time searching for it.
To do that, Netflix hires data scientists, mathematicians and engineers to create machine- learning algorithms that make sense of the countless number of tags and the viewing habit data of more than 250 million individual profiles.
These algorithms have made more than 2,000 “taste communities” around the world that group together people with similar likes and dis- likes based on the tags and what was watched before or after a title so that Netflix can make more accurate recommendations.
For example, Yellin said, Netflix might recommend the new show Ozark to one taste community because it likes Narcos, a cocaine documentary, and El Chapo, but it might also suggest it to a different community that has been watching House of Cards, the People v. O. J. Simpson and Blacklist.
The same end show is recommended to different taste communities that come at it from a different way based on tags or user behaviours.
“( The second group) is obviously into anti- heroes and people fighting off their demons, but the first are people interested in the drug trade,” Yellin said.
By using taste communities to group people together by common interests, genre bias can be eliminated and titles can be recommended to a wider audience. More than 80 per cent of title discoveries come from Netflix recommendations, the company said, with specific plot points and character values helping with the accuracy.
Netflix also dug into its data to find out why subscribers who have never seen a superhero title eventually find various Marvel series, including 2017’s the Defenders.
In one case, the company found those who enjoy watching titles based on the “dark side of society” first started with Amanda Knox, moved on to Black Mirror and Easy before finally discovering Luke Cage, one of the main characters in the Defenders.
“We’re just giving you a glimpse into what we do across all of our shows,” Yellin said. “It’s really important for us to bubble up to the top the right title for (viewers), but finding the right title for them isn’t as obvious as it seems.”
In a perfect world, he added, Netflix will finish refining its recommendations when a user turns on the screen and is served just one title that’s perfect for them every time.
“I don’t think that’s going to happen in my lifetime, or at least in my career when I’m supporting my family, so I think I have a secure job,” Yellin said. “It’s a hard challenge.”