Karthik Ramanathan, Senior Vice President, Cyber & Intelligence Solutions, Asia Pacific

Much has been said about how technology has enabled the rise of digital payments in Asia Pacific. However, until recently, technology has struggled to tackle a related and growing challenge: the volume of fraudulent charges and the rate at which consumers experience false declines. The latter has stayed stubbornly around twenty percent of all transactions for many years.

This is finally changing now, though, through the incorporation of artificial intelligence (AI) and machine learning into fraud detection systems. These new technologies are both improving fraud detection while cutting the false decline rate in half. Being able to address these challenges simultaneously is a transformative change for the digital payments industry.

The positive impact this is having on businesses and wider economies is significant. The cost of false declines is 13 times higher[1] than the cost of actual card fraud, and has been estimated to be US$118billion in the US alone[2]. A lower false decline rate translates directly into greater merchant revenues.

At the same time, a lower rate also significantly reduces consumer frustration with the transaction experience. This increases their satisfaction with digital payments and their transaction volume.

As consumers, merchants and our partners experience and become more aware of these benefits, it is worthwhile understanding how AI is delivering them and will continue improving the payments experience.

The power of AI lies in its ability to self-learn, which brings flexibility and greater accuracy into traditional rule-based fraud detection systems. This allows for the personalization of fraud detection.

Traditional rule-based fraud-detection systems can do this only to a certain degree because they use algorithms that group consumers together based on pre-defined data such as location, purchase amounts and other attributes of past behavior. Any unfamiliar charge—such as a high ticket purchase or transaction from an unrecognized location—is compared against that consumer’s history and the attributes of his or her ‘group’. As a result, this system cannot fully capture individual behaviors.

AI systems, in contrast, mine historic and real-time data from a wider array of touchpoints and identify patterns so that fraud detection systems are able to create consumer profiles based on a segment of one. This means that AI-developed algorithms can more accurately assess changes in an individual’s spending patterns.

Further, AI is able to do this without compromising transaction speed. Fraud detection systems that include AI, such as Mastercard’s Decision Intelligence, still analyze and monitor thousands of transactions per second while also incorporating AI to discern patterns across real-time streams of geographic information, anonymized consumer information and transaction data.

The next evolution of AI will be made more robust with behavioral biometrics. We are already seeing a wider use of device based information given the growth of mobile commerce. The next evolution here is behavioral biometrics. These will include the way an individual angles or holds a smartphone or swipes a finger across a tablet screen. These specific and unique individual data points will allow fraud detection models to assess any “abnormal” use of the device, further personalizing their analysis and results.

Expect transaction experiences to continue to improve. Most shoppers who have their cards declined may stop spending on that card or spending at that merchant outlet; this impacts revenues and reputations of merchants and Card Issuers. In the online world, consumers – particularly those seeking to make high-value purchases – have frequently found themselves frustrated at checkout by false fraud detection alerts. By reducing this frustration, AI contributes to higher transaction volume for merchants and greater consumer satisfaction and trust in digital transactions. As the sources of data available to analyze grow and improve, so will the benefits from AI.

For more information, please contact Karthik at Karthik.Ramanathan@mastercard.com.