Improved predictive underwriting for a life insurance provider
Case Studies / Insurance / Predictive Underwriting
An existing predictive underwriting solution at a large global life insurance provider was not performing up to expectations and needed to be replaced with a better performing solution.
Our goal was to develop a replacement model to make automated rating decisions and mimic underwriters’ decisions with the existing data and infrastructure constraints.
Instead of only identifying low risk applicants, we modified the analytical framework to minimize cost of misclassification. Learnings from advanced modeling techniques were used to build a generalized linear model that could be implemented in client’s technical environment.
Transparency in the modeling solution led to higher acceptability by underwriters and the business. -. Once the model was implemented, the client saw an 8-10% improvement in mortality slippage (and doubled volume at the same level slippage).