The effectiveness of customer loyalty programs (LPs) has often been debated despite their prevalence across many industries. Besides incorrect design, another major reason for doubting their effectiveness is the poor communication between the company and LP members. In this paper, we examined which type of e-mail is preferred by different tiers of LP members and provided insights on how to establish good communication between the company and its LP members. We took advantage of a massive dataset containing six million pieces of data, but it did lead to the ‘deflated p-value’ problem. To solve this problem, we propose using two methods, standardized chi square analysis and ROC analysis. Consistent results were generated, although standardized chi square analysis was much more eff...[ Read more ]

The effectiveness of customer loyalty programs (LPs) has often been debated despite their prevalence across many industries. Besides incorrect design, another major reason for doubting their effectiveness is the poor communication between the company and LP members. In this paper, we examined which type of e-mail is preferred by different tiers of LP members and provided insights on how to establish good communication between the company and its LP members. We took advantage of a massive dataset containing six million pieces of data, but it did lead to the ‘deflated p-value’ problem. To solve this problem, we propose using two methods, standardized chi square analysis and ROC analysis. Consistent results were generated, although standardized chi square analysis was much more efficient in terms of computation time. Therefore, we recommend using this method for overcoming the ‘deflated p-value’ problem. This paper contributes to both the e-mail literature and the LP literature, and most importantly, the proposed method provides guidance for data analysis in the big data context.