Roblox uses an AI-driven recommendation system to personalize the games and experiences surfaced to users on its home screen, operating the 'Recommended For You' algorithm that serves over 10.5 billion impressions daily. An improved version of the algorithm was globally released in December 2025.

The recommendation system uses a multi-stage process: a retrieval stage that identifies a candidate set of experiences relevant to a user's profile, followed by a ranking stage that personalizes ordering based on long-term retention signals such as return play frequency, social interactions, and spending patterns within experiences. Roblox's research team has published work on using large language models to generate structured game profiles from in-game text to improve content-based recommendation signals. Roblox also launched Product Intelligence APIs in late 2025 (RankProductsAsync and RecommendTopProductsAsync) that allow creators to personalize in-experience discovery of game passes and developer products for individual players.