Rodney Nelsestuen over at Tower Group recently wrote this piece "Fraud Management: Covering the Basics, Extending the Value". There was a lot to like in this report and, if financial services fraud is something you care about, I highly recommend getting this report from Tower. I do feel that an enterprise approach to managing decisions, especially operational decisions where fraud is an issue, would be very valuable however and Rodney's report does not really address this so I thought I would.
The report re-emphasizes the many reasons that "Fraud management continues to be a mission-critical function in the financial services industry" and points out the "enterprise-level capabilities needed for superior fraud management". Indeed it says "Best practice fraud management goes beyond point solutions, requiring a holistic, layered integration of the right technology with the enterprise's people, policy, and processes". All true. It also has a nice architectural summary (shown on the right) but there is one thing missing - enterprise-level control of the decisions that a financial services. Look at the graphic - there's a layer in there called "Back-end Monitoring/Detection" and another called "Administration/Reporting". I think these layers need to be support by an enterprise decision management layer that includes:
- Rules, not just for routing and case management but to capture fraud, enforce anti-money laundering policies and flag transactions needing review.
- Analytics, especially those using technology like neural nets that "learn" to predict fraud from complex data patterns
- Sophisticated models, like decision models, that allow for false positives to be managed and traded off against better fraud detection.
- Not just customer-centric, cross-channel data but also third party data integrated into the decision-making process so that identity fraud can be found and prevented
- An ability to load share between the computing power and the skills of people, and to allow their collaboration to find new areas of fraud for instance by allowing computers to trawl through all the data to find new patterns and for a human to say whether it is fraudulent or not
- Not just real-time monitoring but real-time action to stop transactions while fraudsters are still at the ATM or in the branch
- Infrastructure to keep learning and improving fraud-detection decisions using learning strategics and champion/challenger
- An ability for lots of different groups to supply rules and expertise to a centrally managed fraud detection environment not just an ability to share cases and reports
- Enterprise decision management to integrate fraud detection decisions into payment decisions, cross-sell/up-sell and other business areas. It is not just other processes that need to use fraud decisions but other decisions (automated origination for example) that must.
Enterprise decision management is an essential component for an FSI looking to take control of fraud management.