Fraud detection software for credit cards has been a mainstay in banking and card services companies since it originated several decades ago. From a technical standpoint, fraud detection typically doesn’t use big data. Instead, it monitors credit card transactions from individuals and assembles a composite picture of what each individual’s credit using patterns are. If the individual cardholder customarily spends under $100 per credit card transaction within a 100-mile radius of where he lives, and his credit card suddenly shows charges coming in from halfway around the world and in higher dollar amounts than the cardholder spends, the activity is flagged and the software analytics recommend that the card be frozen. A warning flag is sent via online monitoring to the card services company or the host bank of the cardholder, and the institution immediately informs the owner of unusual activity. If the owner says that he did not make the charges, the card is immediately frozen.
Does this work well for banking and credit card companies? In 2011, merchants lost 190 million dollars in credit card fraud, so the job is far from done. Nevertheless, credit card fraud detection software prevents even greater losses. It is used by Visa, Mastercard, American Express and virtually every banking institution-although no one publicly discloses how much fraud is prevented by using the software.
Having once managed card services for a bank, I can attest that the software prevents losses, and sometimes in an embarrassing way. I can recall one instance where an unusual buying pattern was detected on a board member’s credit card, the board member could not be reached, and a staff member locked down his credit card. The board member wasn’t at home because he was standing at the checkout counter of a major home improvement store, watching his credit card get denied at the register!
Despite these occasional “blips,” credit card fraud detection analytics has worked in banking-and its history yields lessons that should be applied to big data, such as:
#1-Big Data Analytics should be as operational as it is strategic
When they start with Big Data Analytics, companies tend to expect major breakthroughs in intelligence that will transform strategy. The reality is, the most successful big data users will be those who learn how to make Big Data Analytics deliver value in their daily operations. When you start mining big data to uncover how you can save money in operations and earn revenue in sales, you are in a position to transform the company’s bottom line.
#2-Big Data must have the right operational “fit”
What made fraud detection analytics a fit in banking was that it didn’t require a great deal of analysis by the end users, who were often card services personnel with high school or AA degrees. The card decisioning software identified the potential fraud. It then gave the staff member or his supervisor a chance to notify the cardholder and freeze the card. This procedure was built into the daily operational procedures in card services departments. A similar effort should be made to simplify the daily use and application of Big Data Analytics.
#3-Big Data operations must have absolute upper management support
Fraud management analytics and detection is standardized and highly regulated in financial institutions-and card services employees cannot make their own rules. It is also supported without exception by upper management. These are important caveats that should also apply to big data. Why is this important? Because it’s still too easy for employees at many companies implementing Big Data Analytics to work around the process by insisting on using the same reports that they were using 15 years ago.
The bottom line for companies as they move to Big Data Analytics is to remember the best practices already cultivated from analytics software. You can strive for every possible strategic advantage from Big Data Analytics-but if you also take the time to firmly embed big data disciplines into the operational fabric of your organization, you can gain greater benefits than you initially anticipated.