Submitter
Scott Zoldi, Chief Analytics Officer
Company
Fair Issac Corp. (FICO)
Industry
Enterprise Software
Business Challenge
Over the past four years, fraudsters have upped the ante in payment card fraud. First, fraud patterns began changing more rapidly. This reduced the effectiveness of analytic models, which required time and expense to update with each new fraud pattern. Second, card issuers started to experience cardholder retention issues when they declined more transactions in an attempt to reduce fraud.
Analytics Solution
FICO developed a patented Adaptive Analytics technology and multilayered self-calibrating analytics for deployment into its FICO Falcon Fraud Manager platform, which protects 2.6 billion payment cards worldwide. This solution can leverage a client’s fraud experience in near real-time to adjust model weights without the need for time-consuming off-line training.
FICO also created a patented technology called Behavior Sorted Lists, which identifies an individual cardholder’s specific spending patterns at preferred merchants and preferred merchant types. Falcon models ingest this behavior, identify whether the cardholder is in-pattern, and use that information when evaluating the likelihood of fraud. This evaluation happens within 50 milliseconds of a card swipe—one-sixth the time it takes to blink.
Impact
The implementation of the innovative analytics in FICO Falcon Fraud Manager enabled card issuers and payment processors to combat the constant barrage of payment card fraud attacks. By adding Adaptive Analytics, one large U.S. debit card issuer realized an 18% improvement in real-time fraud dollars detected, and a relative reduction of 11% in account false positive ratio, saving millions of dollars per year and improving the customer experience for its cardholders. One international card issuer has experienced a 17% reduction in false-positive cases with no negative impact on the real-time fraud detection rate. For transactions that occur at a cardholder’s favorite merchants, bank clients have seen a reduction in false-positive occurrences of 35-50%, contributing to significant increases in customer satisfaction.