Americas

  • United States

Asia

Oceania

Mary K. Pratt
Contributing writer

How Visa fights fraud

Feature
Jun 28, 20226 mins
AnalyticsArtificial IntelligenceFinancial Services Industry

The financial services company has made massive investments in data and analytics to better detect and prevent fraud.

Dustin White, chief risk data officer, Visa
Credit: Visa

For Visa, successful fraud detection must happen in mere milliseconds.

Its Visa Advanced Authorization (VAA) scoring service, one of the company’s most prominent offerings, uses artificial intelligence and machine learning techniques to score the likelihood that a transaction in progress is fraudulent.

It does so in just 300 milliseconds, allowing customers and merchants alike to transact with confidence in real time.

VAA does its job extremely well, too, with the company reporting that VAA prevented $26 billion in fraud on its network in 2021 alone.

Figures like that, says Visa Chief Risk Data Officer Dustin White, prove that the company’s investments in advanced analytics, machine learning, and AI enables “the safe, reliable and fast movement of money between entities that’s powering today’s global economy.”

Visa’s investment in this area is significant.

The company has spent upwards of a billion dollars in recent years to further secure its digital transactions. Some $500 million of that has gone into building out the company’s data infrastructure and developing its AI capabilities.

The goal, White says, is “leveraging data and intelligence to increase the integrity, security, and resiliency of the payment ecosystem.”

Billions lost to online fraud

Such efforts have become increasingly critical in the 21st century as more commerce moves online and the world adopts a digital-first economy.

That shift puts exponentially more transactions onto Visa’s network. (According to company-published data, Visa handled 232.5 billion payments transactions from October 2020 through September 2021.)

The growing digital economy also gives fraudsters more opportunities that they can try to exploit to their advantage, White points out.

Statistics from the Federal Trade Commission give some indication of the massive scope of online fraud. According to the FTC, consumers reported losing more than $5.8 billion to fraud in 2021, an increase of more than 70% over 2020. Online shopping scams were the second most commonly reported type of fraud, accounting for $392 million in reported losses from consumers in 2021, a hefty increase from the $246 million that consumers lost in 2020.

Meanwhile, the Federal Bureau of Investigation in its Internet Crime Report 2021 said it saw a 7% increase in complaints about such crimes over its 2020 figures (which were up 69% over the total number of complaints filed in 2019).

Visa’s systems have held strong against the rising threats even as the pandemic pushed more and more into the digital sphere.

“The pandemic validated the investments we were making,” White says. “They allowed us to have the infrastructure in place and the capabilities in place as commerce moved online. Our tools were able to adapt and scale to those new behaviors.”

How data helps Visa move money

Given its industry and its line of business, Visa has a long history of developing anti-fraud technologies, and data and data analytics are core to that work.

“It all starts with data, and nobody has more data than Visa in the world of digital payments,” White says.

White has been part of those efforts since he joined the company in October 2013 as director of strategic operations and analytics. He became chief risk data officer in March 2020, after moving up through several other positions during his tenure at Visa.

Advanced data analytics and AI power more than 60 different services, including VAA, all of which—in White’s words—“make the movement of money safe and fast.”

Visa also pioneered the use of tokenization, leveraging both static and dynamic information that connects a card user to a specific transaction, to reduce the risk of fraud while also boosting the approval rate for card-not-present transactions (a particularly important capability during the pandemic when online sales and digital transactions spiked). This technology has helped reduce fraud by 28%; that engenders trust in the transactions, which in turn boosted approval rates by 2.5%, White explains.

The company is also using deep learning to study false declines (i.e., when a purchase is erroneously flagged as fraud), aiming to reduce those by 30% and also to refine its understanding of legitimate versus fraudulent activity.

The value of trust

White says such activities are critical because they not only build trust in the security of the transactions but they support commerce and a healthy economy.

As he explains, fraudulent activity costs more than just the money lost on problematic transactions. 

“There’s a larger cumulative cost,” he says, pointing to research that has found that 89% of cardholders who experience a fraud event either abandon or reduce the use of their reissued credential (i.e., credit or debit card). Customers may revert to cash or checks or stop shopping with certain merchants.

“They’ll engage with you less after a fraud event. [That’s why] the name of the game is staying a step ahead,” White says.

And it’s the reason why data is central, as Visa’s data-fueled technology initiatives allow the company to “play offense more than defense.”

Driven to constantly adapt

To that end, White says investment in data technologies not only allow Visa to prevent fraudulent transactions and reduce false declines—both essential for building trust with both consumers and merchants, they also help Visa evolve its services as the fraudsters themselves evolve their tactics.

“We want to make sure our tools adapt to their novel approaches. Fraudsters test and learn, and when they find something they exploit it. So we’re making sure that their learning doesn’t get a foothold and scale,” he says.

Fraud on the Visa network is at a historic low: about $0.07 for every $100 of payment volume with more than 2 million daily attempts to harm Visa infrastructure, White says.

To drive that figure down even further, White says his team must also evolve the company’s analytics capabilities to support the new ways that consumers and merchants are doing business.

Consider, he says, how consumer behavior has changed in recent months as pandemic restrictions eased, with many boosting in-person shopping while also maintaining the high level of digital transactions they had at the height of the pandemic.

“The digitalization of commerce is creating new ways to spend money, and our solutions need to constantly adapt to that,” he adds.

White acknowledges the challenges that he and his team face in doing that work—as well as the necessity of doing it successfully.

“Over the last two years, we have witnessed a seismic shift towards the digital movement of money, with e-commerce growing at 20% a year. And as more and more of our financial lives exist in the cloud, it’s imperative we are proactive at deploying the most advanced technology to ensure customer data is not vulnerable no matter where transactions take place—especially as hackers look to exploit vulnerabilities in card-not-present transactions, where 90% of all fraud globally resides (up from 84% in 2019),” he says.

“Success at Visa is measured based on how well we deliver on our promise by keeping our customers safe and fraud rates low.”