Full-service retail banks have long known that one possible strategy to competitive advantage lies in making better use of their data—and in particular, data aggregated to the customer level.
As a case in point, I'll share a few details from a successful proof of concept project we completed for a large European bank. The objective was to improve credit limit decisioning by finding the best allocation of preferred limits for each customer across four retail lending products: credit card, unsecured loan, mortgage and overdrafts on money transmission account (DDA).
To achieve this objective, we used customer-level, multi-product decision modeling and optimization. We built action-effect models that assessed the impact of different limits on a wide range of profit drivers (affordability, take-up, utilization, good/bad, time to bad, etc.), allowing us to understand the balance between revenue, credit risk and attrition/no take-up risk.
In fact, we built action-effect models for each product, which, in turn, became components of an overall decision model for the customer-level optimization.
The optimization identified the best product credit limit offers to make to each customer that would maximize profit while meeting business constraints such as take-up levels, risk exposure and loss levels. It resulted in a "basket of offers" for each customer, which the bank could potentially use to facilitate its cross-sell and up-sell efforts.