Analysis of a variety of data sets is central to tackling online fraud without creating customer friction.
As the digital economy grows and evolves, fraudsters are presented with a growing number of opportunities to take advantage of eCommerce retailers who haven’t put up sufficient defenses.
“Scale is important for fraudsters,” Ido Lustig, head of risk and fraud at Checkout.com, told PYMNTS in an interview, adding that “back in the days you had to take a physical credit card and go do some fraud in person somewhere. Nowadays, you can run … high numbers of credit cards electronically.”
In addition to the increased scale fraudsters are able to operate at, Lustig pointed to a growing level of sophistication, with social engineering scams and the use of synthetic identities among the ascendant threat vectors of the past decade.
In the first instance, he explained how criminals use phishing emails and texts, as well as fake phone calls to trick people into sharing financial information that can be used to steal their money.
In the second, fraudsters take out loans on behalf of people who don’t exist or are minors and then divert the borrowed funds, meaning that when the time comes to collect repayment, there’s no one to go after.
Building Anti-Fraud Tools
To help protect merchants and consumers from falling victim to such scams, companies like Checkout.com process vast amounts of data to identify patterns of suspicious behavior that can then be used to flag and prevent fraud attempts.
Learn more: Preventing Identity Fraud Comes Down to Effective Use of Data
“We get information about the person who is using the credit card or is performing the transaction, we get information about the transaction, about the device,” Lustig said, explaining how the firm uses all that data to create its fraud identifiers.
And with all those data points to analyze, artificial intelligence (AI) models are able to score each transaction based on the risk that it poses, he further explained.
By using a Big Data approach based on risk scores, Checkout.com’s merchant clients are able to customize their strategies to reduce fraud without blocking legitimate transactions.
Related: Fraudsters Winning ROI Battle, But Data May Turn the Tables
Collecting Diverse Data Sets
As with any such analyses, the more data Checkout.com is able to feed into its machine learning models, the more accurately they can identify fraud. But as Lustig noted, relying too much on consumers to provide that data introduces unnecessary friction into the eCommerce experience.
Instead, he said, “What we tried to do … is use as many data points as possible that are not provided directly by the consumer but are driven from the network effect that we have from the merchant, from the device footprint, from all of those things that we can leverage to ensure that we have … low fraud rates, together with very low friction.”
He did, however, caution that AI isn’t “a silver bullet that will help us solve everything.”
As Lustig observed, just as anti-fraud technology deploys AI, so are crooks using the same technology to improve on their attacks. For example, when creating synthetic identities, fraudsters will use AI to generate names, Social Security numbers, and other fake information at scale.
With threats constantly evolving, he noted how important it is for merchants to not become complacent.
“If you do not have the ability to deploy real-time logics that would decline that type of a transaction … or slightly fine tune it to decline different types of transactions, then you’re going to have a real problem,” he said.
This even makes it more important for online merchants to rely on companies with expertise in cybercrime to successfully beat fraudsters at their game.
As Lustig said: “Do not try to fight on your own because it’s going to be a very, very hard task and that’s not your main line of business.”
For all PYMNTS EMEA coverage, subscribe to the daily EMEA Newsletter.
We’re always on the lookout for opportunities to partner with innovators and disruptors.
Learn More