Predicting a customer’s future behavior could provide many opportunities for a firm to manage its customer relationships. It could decide whether to concentrate on those customers who have a low predicted lifetime value or on those who are already high-value customers. It could establish direct channels and direct communication with customers who indicate a willingness to conduct more business with the firm. It could also determine what special offers cause a customer to use the firm’s service or product more frequently, and how many customers will be active in subsequent time periods. Customer lifetime value (CLV) is one way to measure the relative attractiveness of a customer or a customer segment. Predicting CLV at very early stages of the customer relationship is of particular importance. Firms could identify customers with a high CLV, even before they have proven to be high-value customers. In this study, author Wangenheim develops a model that aims at predicting number of transactions per period, upgrading behavior, and, subsequently, CLV, based on information that is available to a firm early on in the customer relationship. The model uses data from customer communications, channel choices, the availability of choices from competition, and exhibited transaction behavior. The model tests data from a major European airline (disguised). The results show that it is possible to predict customers’ future behavior, early in the relationship. Regular updating of the CLV estimates improves the accuracy of the prediction. By using share of wallet (SOW) and customer satisfaction data for a subset of the customers analyzed, Wangenheim also shows that customers for which CLV is overestimated conduct a smaller proportion of their business with the firm and are less satisfied than customers for which CLV was underestimated. Hence, the model forecasts can be used to determine which customers are worth focusing on
«