Netflix is at it again – schooling us all on what personal really means.
For a long time, Netflix has been perfecting personal recommendations on what to watch. Now it’s delivering a new feature to enhance how it makes those recommendations – personalized artwork.
So OK, that’s cool enough thinking about the thousands of titles, millions of users and all the potential key art variations needed to meaningfully personalize content. But what’s equally cool is their approach to measuring the performance of recommendations. It’s basically impossible to control for all the variables behind personalized artwork to understand what works best. So Netflix employed a methodology called Contextual Bandits.
You’re going to have to read the blog post to really understand it (and then explain it to me!) but here goes: contextual bandits are a class of online learning algorithms that trade off the cost of gathering training data required for learning an unbiased model on an ongoing basis with the benefits of applying the learned model to each member context. In other words, rather than waiting to collect a full batch of data, waiting to learn a model, and then waiting for an A/B test to conclude, contextual bandits rapidly figure out the optimal personalized artwork selection for a title for each member and context.
Anyway, it’s all pretty fascinating. And you can read more about it on the Netflix tech blog.
Why It’s Hot
Netflix takes the idea of dynamic creative to a whole new level, continuing to set the bar for 1-to-1 marketing.