As an early project exploring the possibilities of algorithmically generated film, we set out to examine how we could push the boundaries of customization at massive scale. Through the combination of design, copy, engineering, and a whole lot of math, we were able to ship a personalized film to 800 million people simultaneously, creating an experience that felt hand-made for every single person on the Facebook platform.

How was it done? We designed a back-end algorithm that focused on “high-coefficient relationships” - that is, the friends and family closest to each of Facebook’s 800 million users - and had it surface the most poignant photos, videos, posts, and reactions from that specific part of the social graph. Then, we shot live-action plates with predefined locations that each of these pieces of content could be placed in based on their meta-data, and utilized hyper-specific language and personalization to make each user feel as if this film had been made specifically for them.