Predictive Modeling of Customer Response Behavior in Direct Marketing
- —Young H. Chun, Louisiana State University
- —Yoonhyuk Jung, Ulsan National Institute of Science and Technology, Korea
Abstract
Using the records of customers’ responses over time in direct marketing, many authors have proposed various curve-fitting models to describe and predict the number of responses received after the launch of a direct marketing campaign. Some of those models are based on simplifying assumptions that are not realistic in many practical situations. In this paper, we first propose a probabilistic response model that has many desirable properties. Our geometric response model has three meaningful parameters: (1) an ultimate response rate of recipients, (2) a daily delay rate of respondents, and (3) a total delivery time of the request and responses. We then show that these parameters can be estimated by the maximum likelihood method. Finally, we test our response model by using mail survey data to show its superior performance. One of the advantages of our response model is attributed to the Poisson delivery time that adequately describes the delivery and processing time of customer responses.