Issue
This major bank had successfully launched a low interest, no frills credit card, exceeding their initial acquisition targets. However, as the customer base matured, bad debt levels started to increase above those predicted.
Solution
The solution addressed two issues –Credit Scoring and Acquisition. By analyzing individuals who had developed into bad debt customers, we were able to build a propensity model to identify those customers most likely to become a bad debt problem. On the Acquisition side, using Transactional, Demographic and Geodemographic inputs, we were able to identify those channels which were most likely to deliver good – and bad – debt customers.
Results
The modelling improved both the credit scoring success and total response rates by around 50%. Overall response increased from 0.8% to 1.2%, with bad debt predictive accuracy increasing by over 15%.
