Netflix’s Algorithms

Did you know that over 80% of what people watch on Netflix comes from the recommendation algorithm? So, what is it that entices us to click on that new TV show, a catchy title, an interesting synopsis, or maybe the thumbnail?

If you compare your friend's Netflix account to your own (that is, if you both are using your own subscriptions), you will notice the thumbnails are quite different. Netflix tailor’s thumbnails to create a truly personalized experiences for each subscriber. According to Netflix the thumbnail is the most important piece of the puzzle to engaging viewers to hit Play since as humans we are intensely visual creatures.

Using a collection of tools and algorithms known as Aesthetic Visual Analysis or Ava for short, Netflix determines the best images for thumbnails from the tons of data it sifts through. To put the magnificence of the data in context an hour-long Stranger Things episode has almost 86,000 video frames.

AVA analyses every static frame of a video creating metadata which helps build a database. An image ranking algorithm then uses the metadata to pick out the most attractive and clickable shots. Using the algorithms selected shots Netflix's creative team then design the thumbnail masterpiece.

Netflix's algorithms will then choose the most suitable thumbnail based on the subscribers viewing habits. From the example, the thumbnail on the left showing Robin Williams, a famous comedian could be shown for subscribers who mainly engage with comedy content while other subscribers more interested in romantic titles could be shown the other thumbnail.