Predicting Customer Value and e-channel Disposition from Cross-sectional Survey Data
Research Seminars
Academic Areas Marketing
Professor Sudhir Voleti, Assistant Professor, Marketing, Indian School of BusinessView speaker profile
January 30, 2013
| 12:30 PM - 2:00 PM | Wednesday
AC2MLT, hyderabad, Hyderabad, India
For ISB Community
Abstract:
Full Abstract: A customer’s value to a firm depends a lot on the customer’s loyalty to the firm, after controlling for demographic and environmental factors. We investigate to what extent the attitudinal determinants of self-reported customer loyalty to the firm help predict (i) self-reported and eventual, transaction-based customer value to the firm, and (ii) self-reported disposition towards the electronic channel of transacting with the firm. We use data from a 2001 cross-sectional survey of the users of online banking for a US credit union. Cross-sectional data in surveys offer little in terms of information or degrees of freedom to estimate heterogeneous customer response parameters. We propose and demonstrate the application of a Bayesian Semiparametric approach that uses a probabilistic clustering framework to estimate individual-level customer response parameters. We use the homogeneous parameter OLS and the latent class methods as benchmark approaches. We segment the bank’s clientele using estimates of individual-level attitudinal determinants of customer loyalty. To validate our segmentation exercise, we then use actual transactions data to test for significant segment-wise differences in customer value to the bank and customer disposition to the online banking channel. We find that our method outperforms benchmark approaches in yielding well-identified segments of customers with differential value and online banking disposition to the bank.
Full Abstract: A customer’s value to a firm depends a lot on the customer’s loyalty to the firm, after controlling for demographic and environmental factors. We investigate to what extent the attitudinal determinants of self-reported customer loyalty to the firm help predict (i) self-reported and eventual, transaction-based customer value to the firm, and (ii) self-reported disposition towards the electronic channel of transacting with the firm. We use data from a 2001 cross-sectional survey of the users of online banking for a US credit union. Cross-sectional data in surveys offer little in terms of information or degrees of freedom to estimate heterogeneous customer response parameters. We propose and demonstrate the application of a Bayesian Semiparametric approach that uses a probabilistic clustering framework to estimate individual-level customer response parameters. We use the homogeneous parameter OLS and the latent class methods as benchmark approaches. We segment the bank’s clientele using estimates of individual-level attitudinal determinants of customer loyalty. To validate our segmentation exercise, we then use actual transactions data to test for significant segment-wise differences in customer value to the bank and customer disposition to the online banking channel. We find that our method outperforms benchmark approaches in yielding well-identified segments of customers with differential value and online banking disposition to the bank.
