Churn forecasting in insurance
Doubles predictive power
One of the typical features of the insurance sector is its seldom or null interaction with customers.
Therefore churn forecasting is a difficult task and it is traditionally based on the existence of previous signs, usually very obvious, such as presentation of a complaint from a customer or a dissatisfaction caused by bad management of an insurance claim.
Another trait of this sector is the fact that an insurance policy can indicate relationship between people. For instance, a policy owner can be a different from an insured person, and both can also own a shared insurance policy with other people.
This simple scheme allows defining relationships between customers (for example, a shared policy, a shared address, or a recommendation). These relationships allow identifying influence areas of each customer, known as communities.
Neo Metrics has implemented Social Network Analysis to insurance communities in order to forecast churn at one of the top insurance companies in Spain, achieving double predictive power as compared to traditional churn estimation models.
Analyzing to decide. Deciding to create value.