Netflix knows what you'd like to watch next. How?

14 Sep 2017 | By Anish Chakraborty
Netflix recommendations are not random suggestions

You start binge watching a TV show on Netflix and after a period of time, you finish watching all the episodes of all the seasons. What do you do next?

Well, it is estimated that about 80% of the users tend to make their viewing choices based on the Netflix recommendation system.

Now the question is, how does Netflix know what you will like?

In context: Netflix recommendations are not random suggestions

14 Sep 2017Netflix knows what you'd like to watch next. How?

ChoiceLooking beyond the obvious

Netflix makes use of machine learning algorithms to provide you with these recommendations.

The options, on a casual glance, may appear as a far cry from the viewers' outwardly tastes and initial show preferences. However, they prove to be actually pretty accurate.

Netflix looks into the nuanced threads within the content, rather than focusing on broad genres, to make these suggestions.

Love Tech news?
Stay updated with the latest happenings.
So how does it work?

ProfileSo how does it work?

Netflix's VP of product innovation, Todd Yellin, while explaining the process, said, "The three legs of this stool would be Netflix members; taggers who understand everything about the content; and our machine learning algorithms that take all of the data and put things together."

The video streaming platform currently has around 250 million active profiles. The work begins here.

ContentThe two legs of the metaphorical stool

The platform sieves through these profiles to collate data as to, "what people watch, what they watch after, what they watch before, what they watched a year ago, what they've watched recently and what time of day."

Netflix deploys dozens of in-house and freelance staff to watch its content and tag it. Tags are assigned to several factors i.e. plot, cast, protagonist, storyline etc.

TagThe main pillar

Now comes the critical part, "We take all of these tags and the user behavior data and then we use very sophisticated machine learning algorithms that figure out what's most important - what should we weigh."

These three factors give Netflix its 'taste communities', i.e. users around the world who share the same content viewing traits.

NetflixThis is how it knows

There are "a couple of thousand" taste groups and they drive the Netflix recommendations on your screen. The data, which Netflix feeds into the algorithms, are broadly divided into two categories - explicit and implicit.

Explicit data is what the user directly lets Netflix know i.e. if he/she has liked a show, whereas implicit data is behavioral data i.e. what the platform finds out.