Insight, Impact, Innovation
HomeOur WorkCase StudiesIs your collection strategy alienating profitable slow-pay customers?

Is your collection strategy alienating profitable slow-pay customers?

Download PDF


Collections analysis helps capture $4M additional revenue

The Business Challenge

Our client was faced with the challenge of collecting outstanding credit card debt from a large number of accounts. The collections process increases costs, offsets any incremental dollars collected, and can become a customer relationship nightmare, especially with customers who normally pay on time and are likely self-cures.

Fractal Analytics recognized the potential to collect additional outstanding dollars by using an optimal collections strategy and effective schedules.

The Solution

The most important task in designing an effective collections policy is to segment and score accounts by their propensity to self-cure, roll-over or charge-off. Once a bank can classify debtors by these factors, it can optimally utilize its collections resources to maximize collections.

We deployed our product Collections RiskTreeā„¢ to address the problem of suboptimal utilization of collections resources. Collections RiskTree uses historical information on debtors' behavior and correlates that to debtors' transaction history, bureau information, card characteristics, and demographic data to segment the portfolio and predict debtors' behavior in the near future.

Our analysts built two separate RiskTree collections models using advanced statistical techniques to better classify debtors and their behavior. The first model was an early stage model, scoring customers for their propensity to roll-over from early stage delinquency to mid-stage delinquency. The second model evaluated the delinquency behavior of customers in a later bucket. While RiskTree provides key indicator variables for segmentation and model scoring, both models used customer information variables like maximum past delinquency, credit limit utilization, average pay down ratio, and balance outstanding.

The Results

Fractal Analytics' RiskTree models demonstrated that nearly 75% of the delinquent customers were likely self- cures and the bank could follow a soft collections policy for this segment. A huge savings in cost was attainable by simply reducing the number of collections resources expended on this segment. Further, the self-cure segment would also grow in loyalty since the bank would not need to bother them with collections calls.

The model also indicated that bulk of the charge-off risk was concentrated within 25% of the accounts. Redirecting resources to this 25% of the population which constituted majority of the dollar risk would ensure higher dollar collections without increasing the cost of collection. Our preliminary return on investment models indicated that the bank could potentially collect an additional $4M by simply reallocating collections resources as indicated by RiskTree.