CRISP: Customer response based iterative segmentation procedures for response modeling in direct marketing

We present a system of empirical segmentation procedures called CRISP (Customer Response-based Iterative Segrnentation Procedures) for simultaneously deriving market segments and estimating models of customer response in each of these segments. While the common practice in response modeling is to estimate a single response model for all customers in the database, we allow for customer heterogeneity by calibrating response models for different (unknown) customer segments. We describe a system of iterative segmentation procedures that simultaneously estimate the number of customer segments, the sizes of each derived segment, the values of segment-level response parameters, and their statistical significance, all in a maximum likelihood framework that can accommodate various types of commonly collected response data. To illustrate the CRISP system, we discuss an empirical application entailing typical binary response data for a large number of households for a mail subscription offer from a major magazine publisher. We describe the specific implementation of CRISP to this particular problem of list segmentation, and discuss its Potential usefulness to direct mail marketers. We conclude by discussing the general uses of the CRISP system for response modeling in other direct marketing contexts besides list segmentation.

[1]  Gary J. Russell,et al.  A Probabilistic Choice Model for Market Segmentation and Elasticity Structure , 1989 .

[2]  H. Bozdogan.  Model selection and Akaike's Information Criterion (AIC): The general theory and its analytical extensions , 1987 .

[3]  W. DeSarbo,et al.  An Empirical Pooling Approach for Estimating Marketing Mix Elasticities with PIMS Data , 1993 .

[4]  Jay Magidson,et al.  Improved statistical techniques for response modeling , 1988 .

[5]  A. F. Smith,et al.  Statistical analysis of finite mixture distributions , 1986 .

[6]  W. DeSarbo,et al.  A Disaggregate Negative Binomial Regression Procedure for Count Data Analysis , 1994 .

[7]  H. Akaike.  A new look at the statistical model identification , 1974 .

[8]  D. Rubin,et al.  Maximum likelihood from incomplete data via the EM - algorithm plus discussions on the paper , 1977 .

[9]  Paul D. Berger,et al.  Direct Marketing Management , 1989 .

[10]  Paul D. Berger,et al.  The effect of sample size and proportion of buyers in the sample on the performance of list segmentation equations generated by regression analysis , 1992 .

[11]  G. Schwarz.  Estimating the Dimension of a Model , 1978 .

[12]  Rosana P. Thrasher.  CART: A recent advance in tree-structured list segmentation methodology , 1991 .

[13]  W. DeSarbo,et al.  A maximum likelihood methodology for clusterwise linear regression , 1988 .

[14]  Michel Wedel,et al.  A Latent Class Poisson Regression Model for Heterogeneous Count Data , 1993 .

[15]  William R. Dillon,et al.  LADI: A Latent Discriminant Model for Analyzing Marketing Research Data , 1989 .

[16]  H. Bozdogan.  Determining the Number of Component Clusters in the Standard Multivariate Normal Mixture Model Using Model-Selection Criteria. , 1983 .

[17]  Behram J. Hansotia,et al.  Sample size and design of experiment issues in testing offers , 1990 .

[18]  Wayne S. DeSarbo,et al.  A latent class probit model for analyzing pick any/N data , 1991 .

[19]  Robert C. Blattberg.  Research opportunities in direct marketing , 1987 .