Rise Of The Machines (In Fighting Fraud)

As the data created by online transactions continues to expand at a rapid pace, so, too, do the security elements surrounding them. Todd Clark, SVP and head of STAR Network and debit processing at First Data, explains how machine learning creates a “smarter point of sale.”

Around the world, at every moment, digital transactions are initiating tens of thousands of risk signals being analyzed to ensure their security.

This enormous amount of data — combined with the advent of machine learning — is, in the words of Todd Clark, Senior Vice President and Head of STAR Network and Debit Processing at First Data, “creating a smarter point of sale, in turn, enabling networks to chart a new path to growth.”

Clark — whose company has partnered with machine learning fraud prevention company Feedzai — recently shared with PYMNTS three things that are changing the way payment networks are keeping their participants’ data safe.

First and foremost, observes Clark: “As fraudsters get smarter and faster, so must networks’ fraud detection.”

And that, he explained, is why First Data and its STAR Network felt it was “imperative to integrate machine learning into our systems and that doing so is a significant differentiator.”

Which he also doesn’t believe is something that only they should be thinking about. Clark recommends that everyone in the payments business should be thinking about machine learning, if they aren’t already, given that “you’re only as strong as your weakest link.”

As fraudsters’ capabilities continue to improve, the industry needs to respond in turn, advises Clark, to “beat fraudsters at their own game.”

The second change that Clark points to is the industry having “entered the era of omnidata.”

The opportunities for payment fraud are unwittingly being increased as a result of the expansion of available payment channels. In response, says Clark, “omnidata capabilities” need to be delivered to customers in concert with the omnichannel consumer experience.

“This essentially means better intelligence,” he explains, allowing payment networks to “consume any data in any channel and make sense of it.”

Thirdly, Clark emphasizes the value of combining otherwise disparate data from multiple sources in combating fraud, which the STAR Network says it has received as a result of its partnership with Feedzai.

The combination of Feedzai’s machine learning software and First Data’s experience, Clark says, has made the STAR Network capable of scoring over 3,000 transactions per second.

“We were thrilled when we were able to bring these new STAR scoring capabilities online in months, not years,” he remarks. “Today, we are achieving five nines (i.e., 99.999 percent of transactions are being scored). This is an order of magnitude five times better than other scoring platforms. We’re doing this with two-millisecond latencies. To give you an idea of how fast this is: The human eye takes 250 milliseconds to blink.”

As a result, Clark explains, the STAR Network “can produce new machine learning models to meet new business needs and use cases quicker than ever. The new STAR fraud-scoring platform can ingest many kinds of data that were not previously possible. That’s omnidata at work. We have data adapters that are ready to ingest nontraditional data, such as social network and mobile device IDs, or acquiring-side data, such as basket-level SKU information.”

Clark’s view is that this added capability will allow the STAR Network to “not only enable transactions” but to “grow its network offerings and services for issuers, acquirers and merchants” — thereby providing consumers and merchants “a more positive service experience.”

Machine learning, concludes Clark, can help enable participants in a payment network to leverage its value “regardless of how their customers engage in commerce.”

All that’s left now, he adds, is for a merchant or issuer to “pick the right network.”