How Netflix determines its category names

Netflix uses an algorithm to recommend movies you should watch.

If you’ve ever wondered about the category titles used by Netflix, you’re not alone.

The titles can be overly descriptive at times and long movie categories have become the norm. In addition to the standard categories such as “TV Action & Adventure” and “New Releases”, there are some that appear oddly specific. Binge-watching crime shows might lead to suggested categories pragmatically named “Exciting Criminal Investigation TV Shows” or “Binge-worthy TV Thrillers”.

The prompts you receive regarding shows is based on an algorithm, which suggests content users may like based on information such as viewing history and whether you’ve watched a certain movie to the end or only halfway. Social media platforms including Facebook and Instagram do this as well, but Netflix groups its movies and shows into  “taste clusters”.

Taste clusters are formed by finding patterns among users with similar viewing habits. If you see a suggestion, it’s probably there because another person who watched the same shows you do watches a different show you haven’t seen yet. This prompting is similar to what online book and wine retailers do. They suggest titles or brands based on other people’s choices, who also chose what you bought.

The algorithm is only part of the process, though. After it identifies taste clusters, people known as “taggers” add their two cents' worth. Taggers are employees whose job is to watch TV for a living. These alleged entertainment experts name each taste cluster by finding the common descriptive thread that runs through every movie or TV show in that category. It could be as simple  as “Crime TV Shows”, or it might require a more nuanced approach such as “Visually-Striking Cerebral Experimental Movies”.

So next time you wonder why you’re getting suggested titles, recognise it’s partly due to what you’ve previously watched, partly due to what others with similar viewing preferences have watched and partly to do with “expert” analysis.