What Matters in Managerial Judgments of Customer Value?

The ability of managers to utilize sales-history data to form accurate inferences about the long-term value of customers is examined. Although in recent years considerable advances have been made in the development of formal models that allow such assessments to be, statistically, in most real-world settings the identification of key customers remains an intuitive process. We examine the process that drives intuitive assessments using data from a laboratory simulation that asks respondents to identify which of two customers is likely to be the better long-run prospect based on sales histories. These data allow us to identify the features of sales-history data that appear most influential in driving intuitive key-customer predictions, and compare the nature of these decisions with those that would be made using a well-established statistical model, the Pareto/NBD. We find that respondents utilize sales data in a way that is quite dissimilar from that which would coincide with the Pareto/NBD. For example, empirical results reveal that intuitive judgments under-attend to information about the recency of transactions, and that the prediction process is altered depending on the managerial objective of the assessment task, such as whether to add a customer to a key-account list or delete the customer - a distinction that would not be recognized by the statistical model. Implications of the findings for both academic and applied work in customer relationship management are discussed