Amazon Payments Boosts Conversions Using Contextual Bandits

Amazon Payments applied a multi-objective contextual multi-armed bandit model on Amazon SageMaker AI to personalize a product acquisition funnel. In a seven-week A/B test, one customer population saw a high single-digit percentage relative lift in final-funnel conversion, while another saw no improvement, revealing the issue was content quality rather than the model. The system uses Linear UCB, which conditions decisions on customer behavioral signals rather than fixed segments, enabling personalization that generalizes across similar visitors. A code repository using synthetic data is available for hands-on exploration of the method.