Can technology detect fraud & embezzlement

In the news this week was the story of the U.S. Army Corps of Engineers employees that were indicted in a $20 million bribery and kickback scheme involving government contracts. Quick out the gate were vendors that talked about how their fraud prevention technology could have monitored and detected such scandalous behavior.

But is this true? To find out I took a little deeper dive in the area of fraud detection security solutions to see if they really work.

In general the answer is yes, for industries in which there are high volumes of transactions and available data to model typical consumer behavior and patterns.

For example, in the credit card, insurance claim, telecommunications, and banking industries we find successful application and results of deploying fraud detection solutions.

How they work is by modeling typical consumer behavior and looking for anomalies in transactions that differ from the norm. Then in cases where fraud has occurred, they analyze multiple variables associated with the crime and criminal to create predictive models that indicate the likelihood of a future crime in progress.

Fraud detection software has been used to stop millions in payments before they reached medical providers, who vary from the norm in claims submitted. It’s been used to spot fraudulent car insurance claims. A great example that one insurance provider in Germany reported is that claimants who call back shortly after filing a report and that demand a quick settlement are more likely to be cheaters.

Similar examples are found with credit card fraud. A consumer that buys gas using a credit card at a station where no attendant is present, or buys a diamond ring soon after buying gas has a higher fraud risk profile.

Purchase of two pairs of sneakers in teenage sizes  in London, New York or Miami during school holidays gets flagged as a high risk transaction, possibly involving a stolen credit card. A $100 purchase at a liquor store that sells whisky, a plane ticket outside of the country for a flight leaving in three hours and even a moped purchase by an elderly woman are various triggers based on scoring algorithms that alert companies to possible fraud.

In the case we cited above involving the kickback scheme, no fraud detection software was involved. Just someone who blew the  whistle on the crooks and good old police work. Some things technology will never replace.