FinTech Behavioral Analytics Use Case Success: Predicting Fraud and Reducing Charge-offs
ForMotiv engaged with fraud analytics leaders at a publicly-traded FinTech company that issues personal loans in Q2 2019. The VP of Fraud Analytics and Director of Scoring and Analytics were looking to reduce fraud and delinquency rates by adding alternative datasets to their existing fraud and scoring models heading into the busy holiday season.
Prior to the launch of ForMotiv’s Behavioral Intelligence solution, both teams met to review key metrics and outcomes for the first six months of the engagement. Both sides decided the client should follow the ForMotiv Operational Model which is Collect, Detect, Predict.
Detect. During the “Detect” Phase, ForMotiv used advanced machine learning models to narrow roughly 150 unique behavioral features collected to 7 highly-predictive high-risk behavioral signals unique to this particular customer. These behavioral signals were presented to the customer as signals that could be used internally to understand anomalous user behaviors.
Predict. The “Predict” phase was broken up into two parts, “offline” and then “integrated”. The first being, as fraud and delinquency outcome data matured, the ForMotiv data science and client fraud analytics teams collaborated on this narrowed feature importance and developed predictive models to deploy and incorporate directly into the client’s broadening fraud models.
This initial predictive analysis was done offline with the client until both parties were comfortable deploying a model and signals that could have a significant predictive impact. Once the impact was realized, the Behavioral Intelligence model was deployed to production.
Once the “integrated” phase was achieved and deployment of the predictive model and signals finalized, the client now uses ForMotiv’s automated feedback loop to feed outcome data as it becomes available back into the existing model to improve its predictive power. The client and ForMotiv meet monthly to review any additional findings.
Results. The most expensive source of credit default is malicious intent fraud and lenders struggle to develop models to identify these applications without targeting high false-positive leads. ForMotiv’s in-session behavioral data is assisting lenders in enhancing models and is resulting in a reduction in malicious charge-offs with minimal increases in application rejections.
To date, the model has reduced charge-offs by 12.7%. The customer is approving roughly 100,000 credit cards per month, 10% of which are fraudulent or delinquent after 120 days. ForMotiv’s data allowed the customer to remove roughly 1200 bad policies per month. With an average charge-off amount of $1,250, the customer is now saving roughly $1,600,000 per month.
Schedule a demo to see how your business can accomplish similar results.