Specialist urges fraud data marts

An Australian business fraud specialist advocates the setting up of special data marts devoted to the detection of dishonesty.

Business intelligence (BI) software is usually seen as a way of analysing and improving a business's performance in servicing its honest customers. But its role in detecting dubious dealings can also be a powerful contributor to the bottom line and is often underestimated, says SAS Australia fraud specialist Peter O'Hanlon.

The rise of e-commerce and the trend towards inter-business partnerships and outsourcing opens new avenues and new suspects for fraud, he says. A certain small proportion of customers have always been responsible for fraud, "but it's the staff who know the loopholes", he says and partners' staff can have almost as much knowledge of and access to the organisation's information systems.

Fraud covers a spectrum with two extremes, he says; at one end the "quick hit" involving a single large transaction, with money siphoned off, perhaps, to a fictitious entity, and at the other the "trickle flow" method, diverting a few unnoticed cents from each of a large number of transactions.

Recent large-scale accounting scandals like Enron have heightened the awareness of internal and partner-generated fraud, O'Hanlon says.

A BI system can help in two basic ways: by identifying the exceptions to normal patterns of business activity, and by making the company aware of the patterns of fraud, the ways in which one fraud or fraudster resemble another, O'Hanlon says. Ac-cumulating the statistics on past cases of fraud can help establish the particular characteristics of a transaction that may be fraudulent and stop it before it happens.

More than one retailer, for example, has told Computerworld that they take extra care over any delivery requested to a different address from that given for billing.

Internally, analysis may identify a particular person involved somewhere in the chain of a number of fraudulent activities. Without such analysis concentrating effort on the hot spots, "you'll spend a lot of time investigating transactions that don't warrant it", O'Hanlon says.

Work Cover in New South Wales used BI to identify and act on patterns of fraud among medical providers. It claims to have saved $A30 million in five years with the technique, a return on investment of $10 for each $1 spent on purchasing software and setting up the fraud detection system.

"You don't change your operational procedures or your business databases; you just access the data [with particular attention to the parameters that serve to indicate fraud patterns] and put it in a data mart [small specialist data warehouse] dedicated to that task."

Such a data mart must, however, operate on clean and accurate data, like all business intelligence.

Clearly the technique will come up with some "false positives" - apparent signs of fraud in a transaction which turns out to be honest, and there are risks of creating negativity by investigating the wrong people, O'Hanlon concedes. "But a lot of people are quite willing to be investigated, to the extent of confirming that an exceptional transaction was honestly made, because they see they're contributing to the detection of fraud," and that the organisation is making an effort.

Join the newsletter!

Error: Please check your email address.

More about EnronSAS

Show Comments

Market Place

[]