Customer satisfaction is essential to success in the telecommunications industry. Refund requests drive many calls to customer service and bring attention to subpar customer experiences. A text analytics model was built to “listen” to these calls and identify common pain points that lead to refund requests. Confusion making payments on the web and mobile app was a common theme, leading to accidental and overpayments. Several new UI designs were generated that kickstarted a series of AB tests evaluating a variety of payment flows and payment options. Additionally, loopholes in policy were discovered that allowed customers to game the system and receive undue refunds. The tests, in conjunction with the text analysis, highlighted a clear path forward to change policy, reduce refunds, and improve customer experience.