Improve Customer Retention
Improve the ability to identify policyholders who are most likely to move their insurance policies to a competitor at renewal. By identifying more of these policyholders, a carrier can implement retention strategies to help keep them as customers.
Build New Model and Assess Benefit
Working with top 25 Insurance Carrier, a selection from the over 200 Powerlytics variables was tested to determine if any were highly predictive of customer retention. A model was built inclusive of these variables. An assessment was made to determine which model, the current retention model (excluding Powerlytics variables) or the retention model including Powerlytics variables, delivered the greatest improvement and value to the insurance carrier.
Incremental Benefit of $3.8 million
While both models predict well, the model using Powerlytics variables provided greater lift, less predictive error and more incremental value on the holdout dataset. Extrapolating the incremental value of using the Powerlytics model to a carrier with 2 million policies, a carrier could anticipate an increased premium in the Homeowners line of insurance of approximately $3.8 million.
Improved Lift (Figure 1a versus Figure 1b)
By scoring a blind sample, the current model is able separate the policies into customer groups that are most loyal (92.4%) to customer groups that are most likely to leave the insurance carrier (55.6%). The lift derived from the current model is 36.8 points.
By scoring the same blind sample of policies through the model using Powerlytics variables, we are now able to improve our ability to identify the most loyal customer groups (94.0) and those most likely to leave the insurance carrier (54.3%). This model provides a greater lift of 39.7 points.
Assessment of Incremental Value (Figure 2a versus Figure 2b)
To decide which model to use prospectively, the incremental benefit that the Powerlytics model provided was assessed. The blue bars in graphs below show the number of policies that the model is able to identify as ‘at risk’ of losing at renewal. As shown in Figures 2a and 2b, you can see that the Powerlytics model is able to identify a greater number of ‘at risk’ policies in the lower bins. If an insurance company implemented identical proactive and assertive retention strategies around this segment of business, it was estimated that the Powerlytics model would result in a 0.19% incremental policy growth above the current model.
Extrapolating this result to a larger population of policies, the incremental benefit is significant. An average Top 25 insurance carrier has approximately 2 million Homeowners policies, and as a result, could save an additional 3800 policies using a retention model inclusive of Powerlytics variables. At an average premium of $1,000, that equates to an incremental benefit of $3.8 million dollars.