Bringing Heterogeneity Into Customer Equity Calculations

• Activity Number: 240
• Type: Contributed
• Date/Time: Monday, July 30, 2012 : 2:00 PM to 3:50 PM
• Sponsor: Section on Statistics in Marketing
• Author(s): Bruce Hardie*+ and Peter Fader and Yuzhou Liu
• Companies: London Business School and The Wharton School and Leonard N. Stern School of Business
• Keywords: customer equity ; CLV ; customer lifetime value ; beta-geometric

Abstract

Since the pioneering work of Gupta et al. (2004), researchers have explored the idea of estimating current-customer equity (CCE) using company-reported summary data. One key assumption of this work is that of constant retention rates. However this assumption does not hold in practice, and we know that failing to account for cohort-level retention-rate dynamics leads to systematically biased estimates of CCE.

One reason why researchers have not captured these retention-rate dynamics is the belief that we need longitudinal data for each cohort of customers in order to estimate the parameters of any formal model of contract duration (such as the beta-geometric distribution). We show that this need not be the case. In fact, under certain assumptions, all we require are data on the number of new subscribers and the total number of subscribers for each period. However, it does assume that we have these data from the time the service of interest was launched on the market, which is not met in most settings. We propose a solution to this problem of left-censoring that involves "backcasting" the arrival of new subscribers. We illustrate our methodology using data from Netflix.