Case Study: Large Multi-National Bank achieves 29:1 ROI in 6 months and uncovers new risk predictors with FICO Machine Learning

FICO Machine Learning Yields Rapid Time to Value (TTV) and Uncovers New Risk Predictors

A national division of a large multi‑national bank partnered with FICO to test whether its vast, previously unused transactional data could reveal new predictors of credit risk and improve early-stage customer decisions (cross‑sell, upsell and originations). The bank’s internal analytics team already had origination and behavior scores for more than 9 million customers but needed to demonstrate the practical value of granular transaction streams for risk modeling.

FICO applied its Data Distillery and signal‑detection tools with machine learning to generate ~25,000 candidate features and, working with the bank, distilled them to 11 explainable predictors. The short engagement produced rapid, measurable impact—a 29:1 ROI in six months—with significant predictive lift (over 30% for key segments, 72% KS uplift for mobile‑only originations), a 21% reduction in bad rate for cross‑sell campaigns, and fully transparent, xAI‑driven models the bank could inspect and adapt.