Netflix reveals it personalises artwork design according to users’ viewing habits

Netflix has revealed in a Medium post that it personalises the artwork shown to each user, tailored to their viewing habits. Using the example of Stranger Things, a team from Netflix explains how a range of differently designed cover images are generated, and display to different users according to what they’ve watched before.

In another example, it uses the film Good Will Hunting, showing how, if a user watches romantic movies then the artwork will show an image of Minnie Driver and Matt Damon leaning in for a kiss. If they watch more comedies, the artwork will feature a still of Robin Williams. It also works for specific actors; using the film Pulp Fiction it shows how users might be shown different designs based on their previous viewing of Uma Thurman or John Travolta films.

This personalisation is generated by algorithms and data stored by Netflix, and isn’t always so clear cut, the team explains. “Of course, not all the scenarios for personalising artwork are this clear and obvious. So we don’t enumerate such hand-derived rules but instead rely on the data to tell us what signals to use. Overall, by personalising artwork we help each title put its best foot forward for every member.”

Netflix also shows different artwork for the same show to the same user in multiple places on the platform, to test reaction and learn their preferences. Though it admits this is part of an evolving process, and has questioned the use of too many different artworks for one show, saying it confuse the viewer and the research results.

The article on Medium also gives insight to the smaller design choices on each artwork. “Maybe a bold close-up of the main character works for a title on a page because it stands out compared to the other artwork. But if every title had a similar image then the page as a whole may not seem as compelling. Looking at each piece of artwork in isolation may not be enough and we need to think about how to select a diverse set of images across titles on a page and across a session.”