ID Analytics went deep on fraud protection at Money 20/20
LAS VEGAS, Nev. – If you have ever received a call from your bank asking you to confirm you indeed recently bought golf clubs, you have seen an important part their fraud detection system first hand.
Ken Meiser is Vice President of Identity Solutions with ID Analytics, an innovative provider of credit and fraud risk solutions and analytics. We spoke at the recent Money20/20 Conference in Las Vegas.
The irony of a bank questioning anyone golfing is indeed rich, but they do not call everyone who buys golf clubs. What they know, from decades of purchase patterns generated by millions of customers from many different banks is which groups of people tend to buy golf clubs.
Your bank also knows your personal purchase patterns, and when they detect abnormal spending activity, they may give you a call to see if you indeed bought those Calloways.
ID Analytics powers such decisions by collecting and analyzing information from hundreds of companies across different sectors. By pooling that data together, everyone benefits from its collective strength, Mr. Meiser explained.
“All enterprises contribute activity to the consortium. They show how you act, did not act, and how your identity is typically asserted.”
Abnormal purchases are indeed one common sign of fraud. Another one is purchase velocity, Mr. Meiser said. Fraudsters know there is a limited usage period before fraud is detected, and they want to make good use of that window, so they make several purchases in the first hours after a theft.
That is precisely what happened to me when I had my credit card stolen a few years ago. I got a call from my bank within two hours asking me if I bought two cartons of cigarettes at the 7-11, groceries at a store I never visit, and women’s running shoes. They put a stop on my card and cut off the gravy train.
When a customer applies for credit at an ID Analytics partner, their ID Score may be accessed to provide a risk determination for that transaction. Created 13 years ago, the ID Score is used one million times each day, Mr. Meiser said.
One strength in its design is it gets stronger over time, as more purchase activity from more customers of more companies is incorporated into the data base, Mr. Meiser added. The ID Score is in its ninth iteration as the company applies additional market insight.
ID Analytics knowledge of one specific company’s clients and sales data is never shared with another, and it does not have to be for a competitor to benefit, Mr. Meiser explained. Because customer behavior in specific industries tends to be consistent, a company at risk of fraud from one particular customer benefits from all the times similar fraud attempts were made industry-wide. They also benefit from the thousands of times normal behaviors are proven safe elsewhere in the industry.
That collective strength cannot be achieved by a lone company, and they know it, so they are more than willing to share their data (which their individual customers can opt out of), because they know with more information the model becomes more accurate.
Good patterns also emerge from the data, Mr. Meiser said. Perhaps you do fit the profile of someone who purchases golf clubs, you and thousands of others. While models will never be able to predict singular behavior, companies know if they reach out to a large enough group with a common purchaser’s traits they should buy at a much higher rate than a random group of the same size. That makes it worthwhile.
The only way to come close to 100 percent accuracy is to assess 100 percent of a group, something that is both impractical and expensive, Mr. Meiser said. At some point, the increased investment provides diminishing returns.
“You have to balance the likelihood of loss with the cost to administer the fraud issue,” Mr. Meiser explained.
A company has to be willing to accept a certain percentage of fraud to concentrate on highest probability fraud areas, he added. That may mean the three percent of transactions with the highest likelihood of fraud are singled out. Because of the level of technology employed, that three percent is unlikely to notice any service disruption.
In order for a fraud detection program to make financial sense, you need a large enough population containing enough fraudulent activity. Of one percent of the three percent singled out is indeed committing fraud, then the plan was worth the time and money. A one-in-100,000-event is not, Mr. Meiser added.
In today’s society, many groups of people choose not to use credit cards, Mr. Meiser said. These groups still leave rich data sets, as they have to pay cable and utility bills.
The use of biometrics continues to generate attention, just as it has for the two decades Mr. Meiser has been in the analytics industry.
“For 20 years I have been hearing next year is the year,” he said.
His point was that the biometrics industry has significant challenges to overcome before it can be a reliable contributor.
It is, however, beginning to contribute in select areas, Mr. Meiser acknowledged. He said he is a member of the United Services Automobile Association (USAA). Because the USAA works closely with a stable customer base, voice log-in and facial recognition are valuable tools.