Big Data and Machine Learning Fight Payment Fraud

Most consumers don’t think about e-commerce fraud.

Until they become victims.

Eric Christensen, Director of Payments/VP of Products at digital rivertold PYMNTS that for financial institutions (FIs) and merchants serving these consumers, the growth of e-commerce has created more opportunities for online fraud.

Although the ratio hasn’t changed much (measured as the fraud-to-transaction ratio), the money lost to bad actors is significant. E-commerce losses from online payment fraud topped $20 billion globally in 2021, up 14% from the previous year.

And we are seeing a change in who is trying to siphon off funds.

As Christensen said, “It’s not necessarily the fraud conglomerates trying new things. These are the individuals trying to figure out what they can and cannot do when adopting new e-commerce systems. »

One thing is certain: fraudsters do not take vacations.

And they’re turning to some of the newest and most popular payment methods.

Christensen said that as more consumers become familiar with the emergence of super apps and digital wallets, it will provide more avenues through which fraudsters can attack.

“We’re looking at how fraudsters are going to start attacking the buy now, pay space later,” he said. After all, it is the fastest growing payments that will attract the most attention from criminals.

For companies that have been attacked, waiting and reacting is never a good strategy.

Especially with the inherent seasonality of e-commerce, “the last thing you want to do is play around with new fraud technologies during your busiest times,” Christensen said.

Look at seasonality

Retailers, traders and all sorts of businesses, he said, should find the “low” points in their seasonal distribution – and use that time to adjust the rules so they feel comfortable with them. their fraud defenses in place when business resumes.

Regardless of when adjustments are made, Christensen said, companies need to be mindful of the balance between friction and customer experience.

“Obviously, you could stop all fraud, if you wanted to, by making the experience so horrible that customers won’t come back,” Christensen said, only a little wryly.

the better The thing to do, he said, is to build a holistic, layered approach to fraud — one that takes the rule-based guardrails of legacy systems and leverages big data to detect anomalies inherent in transactions and at the consumer level.

Read more: One-size-fits-all solutions fail to keep pace with growing payment fraud

Christensen said data related to how devices are used, consumer behavior and how data is captured online can help companies craft flexible new rules that optimize fraud-fighting efforts. .

This data, he said, can be as specific as whether consumers transact through iPhones or laptops, and even the geographic location where the devices are used.

How consumers enter their information — directly into fields or “copy-paste” — can indicate whether consumers are genuine or not.

Big data and machine learning can work in tandem to glean insights to help fraud systems make decisions without human interaction.

Rule optimization, Christensen said, gets more orders through while reducing false positives, resulting in a population of happy, satisfied customers.

As the EU has rolled out two-factor authentication, the results have been mixed.

We are now in a better position than before the introduction of the new regulations and multi-factor authentication a few years ago.

“It’s been a bumpy road,” Christensen said. There was not enough interaction between traders and regulators regarding what the rules would be and their impact. “We continue, as an industry, to look for new ways to improve engagement with the merchant community.”

Christensen thinks card brands and card processors have implemented the new rules inconsistently. Merchants were kept in a waiting stage until everything was ready between the various parties who “touched” a transaction.

“There was very little time on the merchant side to make sure it was going to be a seamless experience,” said Christensen, who added, “that’s why you kept seeing delays in the app. takes a long time to get everyone lined up, to get a payment transaction through the system.”

See also: As regulators probe BNPL practices, data privacy and localization rules come into play

“I think two-factor authentication is the right step for us as an industry. I think we need to do it more in the US and keep moving forward.

Going forward, fraudsters are not going to rest on their laurels. The question remains how long it will be before fraudsters find more effective ways to circumvent two-factor authentication.

As Christensen told PYMNTS, “we will always have to keep innovating, because fraudsters are going to keep innovating too.”



On: Shoppers who have store cards use them for 87% of all eligible purchases – but that doesn’t mean retailers should start buy now, pay later (BNPL) options at checkout. The Truth About BNPL and Store Cards, a collaboration between PYMNTS and PayPal, surveyed 2,161 consumers to find out why providing both BNPL and Store Cards is key to helping merchants maximize conversion.

Leave a Comment