Insurance renewals and lapse prediction

Back to White Papers

The problem

Logic dictates that longstanding policyholders be rewarded for their loyalty with cheaper premiums relative to new customers. The opposite is the case - their premiums are slashed overtime. In some cases, the disparity can be markedly significant in order to attract new policyholders. The FCA estimates that 6 million people are being overcharged, especially policyholders enrolled in auto-renewal, eliminating the opportunity for a policy review. The result: policyholders leave and take a better deal elsewhere. In some cases, motor insurance companies are losing up to 50% of their customers.

What does this mean for our models?

One of the issues with this is the disruption to machine learning lapse modelling. A lapsed model identifies policyholders who are most at risk of lapsing, considering their demographics and interaction with the insurance company - these include: additional products, add-ons, as well as the number of years a policyholder is with the company. The issue is that with such massive savings on offers from other insurers for new policyholders, they are attracted to savings from a cheaper plan elsewhere. This inevitably weakens lapse prediction models to the point of them being worthless - someone’s likelihood of lapsing is massively correlated to cost saving and little else.

Change is coming

A recent initiative by the FCA changes this completely. These plans mean insurers will have less flexibility on price setting and will be obliged to put loyal customers onto cheaper deals in line with what they offer new policyholders.  If the FCA decides on implementing these restrictions, existing customers will be less likely to search for new deals elsewhere. This leaves customers, old and new, on a level playing field.

What does this mean for lapse prediction modelling?

Insurance companies will ultimately have significantly more accurate models – the random component of existing customers leaving, for better deals, would be significantly reduced. At MM our models will now be more powerful and better able to explain why customers are leaving. Inevitably in order to try fix the problem of customer retention by using marketing solutions.

In effect, with the reduction in premium for existing loyal customers and the subsequent improvement in the power of lapse modelling, the net effect with be an increase in customer retention.