Our recommendation engine fuels discovery across multiple product surfaces in headset (such as Horizon Feed, Meta Horizon Worlds catalog, Horizon Store), on web, and in our Meta Horizon mobile app. While each surface has its own design and entry points, the core recommendation process generally follows two stages: Candidate Retrieval and Personalized Ranking.

To build a thriving ecosystem, we are making investments to bootstrap new, emerging worlds to a selected group of users who might be interested. We start with an initial pool consisting of worlds that meet a baseline threshold of user engagement. These worlds will then graduate from this pool and enter the next pool where they have a higher likelihood of being exposed to a broader audience. The worlds with the highest user engagement (i.e. retention, time spent, user volume) from this pool subsequently graduate into a smaller pool of “breakout” worlds that can be promoted to the broadest audience.

We recently launched an early iteration of this in Highlighted Worlds and noticed that a higher volume of emerging worlds made it to the top of the list. At the same time, we are enhancing our recommendation engine to be more responsive and are expanding this approach to other surfaces. This will help ensure that creators of all sizes have an opportunity to reach their audience and succeed on our platform in the future.

Our discovery recommendation engine is continuously evolving to boost user engagement and support ecosystem success. By providing a personalized experience and an efficient content exploration strategy, we aim to improve how users discover and interact with worlds. To make the most of these benefits, we recommend a few actions that creators can take today to improve world engagement: