Case Study: Atmosfy boosts D1-D7 retention with Shaped

Atmosfy boosts D6 retention by 79% with Shaped

Atmosfy aimed to boost user retention and long-term growth by improving its content discovery feed. Their challenge was replacing a static, logic-based system that required manual adjustments with a dynamic, personalized recommendation engine. To achieve this, they partnered with Shaped to implement a real-time personalization engine that could better align content with individual user preferences.

Shaped built a foundational recommender using weighted engagement signals and its ELSA autoencoder to create flexible video embeddings. The solution was integrated with real-time data streams and refined through rigorous online experimentation. This implementation by Shaped drove significant increases in retention, including a 79% boost in Day 6 retention and a 15% increase in Day 1 retention, achieving Atmosfy's core goal.

Atmosfy