Fraud Model Optimization & Tuning
A Global Payment Processor was experiencing high losses due to fraud committed against their debit cards on their debit solutions. Matrix-IFS was hired to evaluate and create a roadmap in order to close the gaps in its fraud monitoring solution.
After a two-week assessment and development of an execution plan, our team began working to implement the plan as detailed according to the scope of work. The fraud prevention solution implementation plan detailed three stages designed to maximize and front-load returns.
Phase 1– All debit transactions were entered into the solution to allow for basic rule coverage on all debit transactions. This process closed significant holes in the client’s transactional coverage and allowed for an immediate and significant reduction in fraud losses.
Phase 2 – We added user profiles, calibration scoring, and four additional feeds which increased the rule coverage and lowered false positive ratios.
Phase 3 – Our team performed model tuning and added additional channels to consolidate the card approval process under the same fraud monitoring engine. They also created a two-way feedback loop between the fraud solution and internal customer support interface in order to automate the handling of alerts and calling customers.
- Staged deployment structure allowed the client to realize returns on investment within the same calendar year.
- Matrix-IFS provided guidance as to the hardware required to operate the system in real-time and assisted in coordinating its procurement and setup.
- Score calibration and tuning have helped the team capture fraud which was not covered by their internal data mining rules, increasing fraud solutions coverage and lowering false positive ratios even further.
- Creating user profiles based on multichannel data assisted the client in significantly increasing their fraud prevention capabilities.