The Idea:Forecast how much customers will use a service and whether they will renew a contract by considering underlying motivation.
Customer loyalty is a particular concern for businesses that rely on contracts or memberships for revenue, such as gyms, arts organizations, and cell phone service providers. While previous research has provided useful estimates of which customers are the best targets for retention, few tools help businesses develop long-term forecasts of how much customers will use a product or service or how much revenue renewing customers will generate. Nor is the connection between these behaviors and customers’ underlying motivations well understood.
Professor Eva Ascarza worked with Bruce Hardie of London Business School to create a new model in which a single trait — commitment — simultaneously drives customer usage and renewal behavior. For instance, consider a customer who pays a fee every month for unlimited use of a gym’s facilities. Commitment is reflected both in the number of times she goes to the gym each week (usage) and her monthly decision of whether or not to continue her membership (renewal). Using a combination of modeling and estimation techniques, the model predicts all future revenue from active customers, hence detecting which customers are more valuable both in the short and long runs. The model identifies customers at risk of not renewing a membership by observing changes in how much a customer uses a product or service as a contract expiration date approaches. Furthermore, the model provides a way to classify customers on the basis of their commitment to the organization.
The researchers tested the model by analyzing four years of transaction data — periodic information on how many tickets each customer bought and whether she renewed her membership — for an arts organization’s membership program.
Ascarza and Hardie used their model to predict the frequency and rate of future transactions and how likely each member was to renew the membership. The researchers compared these predictions with records of actual member behavior as recorded by the organization, finding they were able to forecast future behavior with at least 97 percent accuracy. They compared their model’s predictions with those obtained through existing methods widely used in practice. Their approach significantly outperformed key benchmarks of current methods used to make similar predictions.
Additionally, the model was able to accurately classify customers on the basis of their commitment to the organization, proposing a new approach to segment the customer base.
Marketing managers, customer relationship managers, operations managers, direct marketers
You can use this research to identify and strategically target customers based on how likely or unlikely they are to renew contracts or memberships, as well as those who appear most committed to a product, service, or organization. You can also use this research to more easily calculate expected revenue over a customer’s lifetime, to aggregate data to predict sources of profit, and to plan future marketing efforts.
Publication type: Working paper