A Connected Economy Requires Connected Risk Management

Fraudsters aren’t contained to a 9-to-5 workday, and they’re not off on holidays and weekends. Instead, they’re working all hours of the day and night.

“So from that perspective, you need to have a solution that, if there is something that goes bump in the middle of the night, it can pick up on it, and it can make adjustments,” John Winstel, senior director of Fraud Product at FIS, told PYMNTS.

Artificial intelligence (AI) tools spot those things that go bump in the night — or more specifically, payments risks — by picking up on anomalies. To do that, the AI tools need to be fed with data.

The AI tools can then pick up on anomalies, make sense of the data, and then go in and adjust the fraud management strategies. By doing so, they save humans the need to do the tedious process of going through reported fraud and suspicious activity, allowing the humans to focus on other components of fraud management instead.

“What we have found, is when you take an assisted AI approach, where you combine the power of AI along with the knowledge of one of our fraud analysts together, that’s where you start seeing the best authorization rates, the end user experience starts to get better, and we start seeing where you’re really optimizing your overall offering while at the same time mitigating those fraud losses,” Winstel said.

Spotting Anomalies, Setting off Alarm Bells

Card testing is a fraud trend that’s hot right now, with fraudsters getting a hold of card data and trying to figure out what’s valid. That’s an area where AI and machine learning (ML) can intervene. Since it knows what the merchant’s normal authorization rate is, the AI tool can set off alarm bells when it suddenly sees 1 million transactions per day where it normally sees only 1,000.

“You shouldn’t have to have a person that’s going and intervening,” Winstel said. “That’s where AI can even help to identify those type of anomalies and help to shut it down.”

Another trend over the last several years has been about the goals of those who use fraud-management tools. They used to be most concerned about keeping fraud losses low, but fraud mitigation that is too stringent and declines too many good transactions can cause merchants to lose not only that sale, but also the customer.

“It’s going to cost you even more in the long run,” Winstel said. “The cost of a false decline can be huge for everybody in the ecosystem.”

Sharing Data to Maximize Authorization Rates

One solution is to bring together merchants and issuers as we look at ways to improve approval rates while mitigating payment risk. For example, merchants can share that they know a transaction is good and should be approved because they know the customer is a good customer.

“The market’s going to continue to go there — we’re not the only ones that are doing that, I realize that,” Winstel said. “There’s going to be more and more, from a data sharing perspective, to provide those signals from both sides of the transaction so that we can really start to maximize our overall authorization rates.”

Looking ahead, Winstel said that as more and more consumers shop both in-store and online, merchants will need to make sure that all their fraud prevention tools talk to each other.

For example, if the merchant has a good customer who makes purchases on a regular basis in the store, that information should be shared with the merchant’s eCommerce fraud solution, so the customer has a good experience when buying online and picking up curbside.

“I think over the course of the next year or so, the next few years, it’s going to be all about cross-channel fraud management and bringing all that data together,” Winstel said.