Microsoft Research partnered with TripAdvisor to explore how a membership model affects user engagement using causal AI. Leveraging an A/B test with a simplified sign-up, the ALICE team developed a new machine learning method to handle non-compliance and estimate varied user responses. Results showed membership boosts engagement, especially among iPhone users and those visiting vacation rental pages. The methodology, now in the open-source EconML library, enables more accurate customer segmentation and personalized strategies, helping TripAdvisor tailor experiences and improve retention.

Join for free to read