Organised fraudsters have always been a step ahead in conjuring up new techniques to perpetrate fraud. Extraordinary sums of money have been lost to fraud. Several studies put the fraud figure between 3-8% of annual revenues. Equally, extraordinary sums of money have been invested in fighting fraud over the last decade. A Forrester report puts investments in Fraud Management Systems (FMS) close to 1.5bn. However, despite all the investments, evidence suggests that not all the investment has yielded the desired results.
One reason attributed to reduced returns is that most fraud tools haven’t kept pace with the changing nature of fraud. For instance, the vast majority of fraud detection tools in the market today work on high usage alarms which require unwieldy rules and thresholds to detect fraud. Our experience, working with some of the largest mobile operators, is that such rules based and menu driven systems require highly trained Analysts to use the system and substantial administrator time to get the best results. Another major drawback of such approach is the high false positives (legitimate high usage behaviour) no matter how balanced the thresholds are set. The high false positives generally take up valuable Analyst time and a high proportion of undetected cases (fraudsters circumventing thresholds), cause further losses.
Additionally most tools struggle to deal with vast quantities of data and new scenarios that need to be analysed at short spans of time; at least not without substantial infrastructure investment. Those that work on advanced models are out of reach for cash strapped service providers (licenses in excess of $1,000,000). Often new forms of fraud catch the operators by surprise and the Analysts are pressured to analyse vast quantities of new data at the drop of a hat. Now, imagine if it takes the Analyst to request for new data, wait for weeks until it is prioritised by IT, a few more weeks for it to be developed, tested and delivered (assume there is sufficient space in the hugely expensive data warehouse); the fraudster has almost certainly moved on to newer types of fraud. One could visualise the frustration of the Analyst on the lost opportunity.
The future is this: no extra licensing for extra analysis; Analysts are truly in control – they can analyse what they want, when they want without the fear of paying more to vendors; less reliance on IT for structured data; less reliance on a hugely expensive data warehouse infrastructure; analysis at speeds hitherto unimagined and interface that is as intuitive as your DVD player. We call that doing more with less. The future of telecoms fraud management is here.
