Customer churn prediction in insurance industries : a multiproduct aproach

Abstract

Customers in the insurance industry usually have multiple products under the same company. Due to the importance of costumers and the increase of quality and satisfaction given in services, churn prediction for multiproduct customers has become important to support customer satisfaction. An estimation was made, using survival models, to find the determinants that most affect multiproduct customer churn. Using data from a Latin American company, the estimation shows that the first product cancellation influences customer churn. The variables that have the strongest influence in churning are the number of cancelled products, the customers portfolio, the claims paid and the distribution channel. On the other hand, the variables that increase the life time of the customer in the company after the first cancellation are the maximum number of products the customer has had and having a health insurance after the first cancellation. This study contributes by identifying the significant variables that influence customer churn in the Latin American insurance industry, and estimating the time a customer will remain in the company after the first cancellation, which will help stakeholders on the achievement of a customer centered strategy.