Retention of health insurance policyholders

Predicting which policies will lapse and which interventions are likely to retain the policy

>50%

reduction in lapse rate

Results

91%

Accuracy in predicting which policies will lapse in the next 3 months

5.1%

increase in embedded value (EV)

Benefit

Reduction is insurance lapse is a large contributor to insurance profitability. By identifying and retaining good customers, substantial premium income is saved.

Context

A large multinational life insurer was challenged by high policy lapse rates for one of its health insurance portfolios. Although it had a retention team and strategy, its activities were entirely reactive – at the point that a policyholder contacted the insurer to cancel a policy, the team would attempt to persuade the policyholder to stay. This had limited success. However, by predicting lapses with longer reaction lead time, it was hoped that more policyholders could be retained.

Methodology

More than 400 pieces of data for each policyholder at each historical point in time were used to train an algorithm to predict which would lapse 3 months into the future. Once this model had achieved sufficient accuracy, a further model was used to determine which intervention would be most likely to retain the policyholders who were predicted to lapse. These were actioned by a dedicated retention team and call centre.

  • Emerge ML Twitter
  • Emerge ML LinkedIn

© 2020 by Emerge