Another customer panel courtesy of SAS. This one had Rex Pruitt from Premier Bankcard (I blogged about how Premier put predictive analytics to work and Rex presented at Predictive Analytics World), Chris Swecker formerly of the FBI/Bank of America, Cameron Jones SAS’ Chief Compliance Officer and was moderated by Ellen Joyner. Fraud, obviously, is a huge deal. With estimates of hundreds of billions of fraud worldwide and the increasing role of organized crime in fraud, organizations are getting more aggressive about reducing fraud. Analytics offers great benefits in fraud with everything from social network analysis (currently trendy and the topic of a recent product from SAS) to pattern detection with neural networks.
Rex is a portfolio analyst at Premier and considers every aspect of the customer lifecycle. Fraud modeling at Premier is important as about 4% of their applications were fraudulent and were causing charge-offs because they were not being caught at application time. Using Enterprise Miner they build a fraud score and implemented it. $9m in bottom line revenue thanks to the fraud approach – eliminating fraudulent applications reduces defaults (36% reduction in first-pay default for instance), reduces charge offs and frees up credit line capacity for good applications (credit card companies get a double whammy from fraud applicants as they often use up a limited capacity for credit). Chris likes to talk about a new paradigm in fighting fraud – fighting fraud as an enterprise, focusing on fraud as a network challenge not a one-off, transaction at a time activity. The growth of organized crime as an issue in fraud is established and the way to address this is to focus on them as a network – a global, viral, network. And far too many never get prosecuted, not least because it is increasingly internet-based and worldwide/cross-jurisdictional. The best way to find these networks is use analytics – front-end scoring to find networks not just one transaction at a time and back-end investigation of fraudsters to drive out their networks. Detecting fraud early, when it is still focused on lower value product lines, helps protect higher-value products over time as organized fraud almost always expands from low to high value products. Cameron came next. SAS has a fraud framework designed to pull together alerting, fraud scoring and compliance issues around fraud. Cameron talked about the new SAS Case Management framework, not something in which I am terribly interested as case management is a known problem and it did not seem like SAS had taken a terribly innovative approach – there was no sign of it being a dynamic environment driven by decisions for instance.
Compliance with regulations and catching fraud to save money are two sides of the fraud problem. In AML, for instance, the focus is on meeting a minimum standard while credit card fraud detection is all about the money saved. Chris emphasized that organizations need to focus on more than just being compliant with regulations and actively try to prevent problems even when they don’t seem to have an immediate bottom-line impact. Chris also pointed that the size of the networks involved in fraud are growing and it is a very networked problem – not hierarchical, not individual fraudsters but widespread and interconnected networks.
Fraud is, of course, one of the most important uses of analytics and decisioning for many organizations. Nice panel.
Comments on this entry are closed.