Merchants Face off Against Evolving Fraud Tactics

In “The Way Payments Are Now Done,” 33 payment executives discuss what payments’ “new normal” looks like. Andre Machicao, SVP, global head of product and solutions at Cybersource, explores how the COVID-spurred digital acceleration has caused a spike in fraud, and how merchants are rethinking their fraud management tactics and using machine-learning models.

As businesses expand eCommerce capabilities, more consumers are starting purchases in digital channels — and completing them there if they can. Unfortunately, higher volumes of digital transactions can also bring more fraud attempts to merchants of all sizes.

Fraudsters are becoming increasingly sophisticated with testing merchants’ systems to determine which ones are likely vulnerable or under-protected. Tactics like friendly fraud, card testing, phishing and identity theft are quickly becoming the leading means of attack in the midst of an accelerating shift toward digital. COVID spurred accelerated growth in digital commerce, and what followed was an increase in fraud attempts and fraud rates by revenue for around three in four merchants globally, with merchants in Asia-Pacific and enterprise and mid-market merchants especially impacted.

When it comes to the tools used to fight these new fraud trends, merchants are rethinking their fraud management tactics, opting to rely more heavily on a handful of widely used solutions as opposed to nice products with specific capabilities. In fact, Cybersource’s Annual Fraud Benchmark Report found that the average number of tools merchants have in place has dropped by roughly half, from 10 in 2019 to five this past year.

With the rapid growth in digital transactions, traditional manual review practices in fraud prevention can easily become overwhelmed by the volume of activity. With that threat looming, businesses have turned to advanced machine learning models to automate risk detection and increase risk scoring accuracy, all while evaluating larger numbers of transactions in real time. In an eCommerce setting, with countless transactions happening at any given time, minimizing the human element goes a long way toward improving speed, efficiency and accuracy in the fraud management process.

With the massive amounts of data occurring with eCommerce, global scale in your data set will help a machine learning model learn that much faster and have the ability to train itself on the latest technologies and consumer shopping preferences (i.e., mobile or BOPIS). Machine learning models are better positioned to maximize internal and external datasets, help manage fraud accurately and protect revenue – at a lower cost to the business.

The world is going digital. Whether it is contactless payments in-person or eCommerce in general, consumers are seeking out new ways to make everyday purchases. And as consumers change their behaviors, so too have fraudsters, opting to shift their tactics online.

Tackling fraud is an always-on effort and can prove challenging for businesses that need to balance customer experience with security and efficiency in transactions. As the world continues to embrace the shift to digital commerce, tools like machine learning, coupled with customizable business rules, will allow merchants to both fight fraud and generate a better customer experience in the process.