Date of Award
Doctor of Philosophy (PhD)
Despite the popularity of mobile and social media, email continues to be the marketing tool that brings the highest ROI, according to the Direct Marketing Association’s “Power of Direct” (2011) study. An important reason for email marketing’s success is the application of an idea— “Permission Marketing,” which asks marketers to seek consent from customers before sending them messages. Permission-based email marketing seeks to build a two-way interactive communication channel through which customers can engage with firms by expressing their interests, responding to firms’ email messages and making purchases. This thesis consists of two essays that address several key questions that are related to the management of a profitable interactive permission-based email marketing program.
Existing research has examined the drivers of customers’ opt-in and opt-out decisions, but it has investigated neither the timings of two decisions nor the influence of transactional activity on the length of time a customer stays with an email program. In the first essay, we adopt a multivariate copula model using a pair-copula construction method to jointly model opt-in time (from a customer’s first purchase to opt-in), opt-out time (from customer opt-in to opt-out) and average transaction amount. Through such multivariate dependences, this model significantly improves the predictive performance of the opt-out time in comparison with several benchmark models. The study offers several important findings (1) marketing intensity affects opt-in and opt-out times (2) customers with certain characteristics are more or less likely to opt-in or opt-out (3) firms can extend customer opt-out time and increase customer spending level by strategically allocating resources.
Firms are using email marketing to engage with customers and encourage active transactional behavior. Extant research either focuses only on how customers respond to email messages or looks at the “average” effect of email on transactional behavior. In the second essay, we consider not only customers’ response to emails and their correlated transactional behavior, but also the dynamics that govern the evolving of the two types of customer relationship: email-response and purchase relationships. We model the email open count with a Binomial distribution and the purchase count with a zero-inflated negative binomial model. We capture the dependence between the two discrete distributions using a copula approach. In addition, we develop a hidden Markov model to model the effects of email contacts on purchase behavior. We also allow the relationship that represents customers’ responsiveness to email marketing to evolve flexibly along with the relationship of purchase.
In the second essay, we apply the proposed model in a non-contractual context where a retailer operates a large-scale email marketing program. Through the empirical study, we capture a positive dependence between the opening of emails and purchase behavior. We identify three purchase-behavior states along with three email-response states. The empirical finding suggests that the customers who are in the medium relationship state have the highest intrinsic propensity to open an email, followed by the customers in the lowest and highest relationship state. Furthermore, we derive a dynamic email marketing resource allocation policy using the hidden Markov model, the purchase and email open model estimates. We demonstrate that a forward-looking agent could maximize the long-term profits of its existing email subscribers.
Zhang, Xi, "Managing a Profitable Interactive Email Marketing Program: Modeling and Analysis." Dissertation, Georgia State University, 2015.