In recent decades, the use of incentive-based reward programs to foster customer loyalty and promote sales has become prevalent in various industries. While these strategies are widely accepted and implemented, there is a significant gap in empirical studies to ascertain their real-world effectiveness. This thesis embarks on a comprehensive examination into the effectiveness of an online business's reward program, utilizing data from the past five years and employing data mining techniques, including RFM (Recency, Frequency, Monetary) model and clustering algorithms; hypothesis tests are employed to further strengthen the drawn conclusions. Contrary to popular theories, the findings reveal that small incentives such as rewards did not induce significant changes in customer purchasing behavior, nor did they effectively boost sales among rewarded customers. A control group of non-rewarded top-class customers showed more robust purchasing patterns. These unexpected results challenge existing beliefs and call for a critical re-evaluation of current practices in sales promotion and customer loyalty. The research underscores the need for empirically grounded strategies, further exploration into alternative loyalty-building methods, and a recognition of the complex realities influencing customer engagement.