2016-12-01

This story was delivered to BI Intelligence "Payments Briefing" subscribers. To learn more and subscribe, please click here.

MasterCard will incorporate machine learning and artificial intelligence (AI) into fraud prevention through a new tool, called Decision Intelligence.

The platform, which the firm calls a “comprehensive decision and fraud detection service,” gathers information including value, time, device, and merchant data, and uses an algorithm to make a real-time decision about a purchase's legitimacy, ultimately cutting down on both false declines and actual fraud.

The tool can also learn from each transaction and apply that knowledge to later purchases, therefore improving its abilities over time. Decision Intelligence, which will launch in all markets, marks MasterCard’s first global implementation of AI.

That’s critical for both merchants and Mastercard as they look to solve a complex problem.

Card fraud continues to rise, especially when it comes to e-commerce. Both the number and value of fraudulent transactions rose in the US in 2016. Much of that fraud is shifting online —  from October 2015 through April 2016, e-commerce fraud attacks rose 11%. And cost per dollar for digital fraud grew by up to 12% year-over-year, compared to just 3% growth for physical fraud, which illustrates how costly this is for merchants. That’s led players to invest in robust fraud prevention offerings, including extra verification steps and increased manual review.

But these efforts may actually create more problems. Adding steps or increasing manual review can make purchasing challenging, which can lead to cart abandonment and lost sales. But even when customers do complete a purchase, these tools aren't always effective. Stringent purchase monitoring increases false declines, or purchases that are declined because they look fraudulent, but aren’t. Fifty-eight percent of declined transactions were false positives in 2016, according to LexisNexis. BI Intelligence estimates this cost US e-commerce merchants $8.6 billion last year, exceeding both actual fraud losses and the amount saved from prevented fraud.

Using smarter tools could help mitigate some of that impact while improving purchase security. The most effective way to prevent fraud while limiting false declines is through “multilayered” products,which can bring down false decline rates by up to 13 percentage points, according to LexisNexis.

Decision Intelligence fits into this category, and could therefore help MasterCard's many partners limit false declines without further impacting the consumer purchasing experience. Such a tool could ultimately increase purchase volume and limit losses, which helps issuers, merchants, and Mastercard alike in the form of revenue.

E-commerce merchants are dealing with rising fraud, and in response, they're putting stronger safeguards in place to try to protect against these unlawful transactions. However, e-commerce companies often over-correct for the threat of fraud, leading to false declines, also known as false positives, which occur when a legitimate transaction is rejected.

These false declines are becoming a costlier problem than actual fraud — US e-commerce merchants will lose $8.6 billion in falsely declined transactions in 2016, according to our estimates. This amounts to over $2 billion more than the $6.5 billion in fraud they will prevent, meaning that false declines are undermining these merchants' ability to effectively combat fraud.

BI Intelligence, Business Insider's premium research service, has compiled a detailed report on false declines that looks at the rising cost of fraud and how false declines are actually a larger direct cost for merchants than fraud itself. It also identifies the reasons why e-commerce and mobile commerce companies are particularly vulnerable to these trends. And lastly, it lays out some of the major causes of false declines and the most important solutions being put in place to try to combat this problem.

Here are some of the key takeaways from the report:

Fraud is rising in the US, costing merchants 1.47% of annual revenue in 2016, up from 0.51% in 2013. As fraud eats into revenue, merchant processors and acquirers are seeking to blunt its impact by enforcing strict rules to block suspicious transactions.

False declines — valid transactions that are incorrectly rejected — are unintended consequences of e-commerce merchants' fraud prevention strategies. False declines, also called "false positives," will cost e-commerce companies $8.6 billion in 2016, according to our estimates. This eclipses the $6.5 billion in prevented fraud, meaning false declines must be reduced in order for merchants' fraud prevention strategies to be cost effective.

Causes of false declines fall into three buckets. False declines can be caused by identity-related, technical, or structural issues. Examples of causes include conflicting shipping and billing information, outdated card information, and differing risk appetites among issuers and merchant acquirers/processors.

There are solutions for two of the types of causes. E-commerce merchants can solve identity-related problems by requiring their customers to authenticate themselves through more accurate means, such as 3D Secure and biometrics. Merchants can solve for technical issues associated with false declines by using smart routing, card updaters, and local domains.

But structural problems are limiting the effectiveness of these solutions. Each issuer, acquirer, and processor makes decisions on fraud using their own set of standards. This makes it difficult to contain the problem of false declines because stakeholders can't control each other's criteria.

In full, the report:

Estimates the cost of false declines compared to fraud losses and prevented fraud.

Determines the effectiveness of e-commerce merchants' current fraud prevention strategies.

Categorizes and explains the various causes of false declines.

Uncovers the potential solutions to solving false declines.

Provides guidance on how merchants can minimize the issue of false declines going forward.

Interested in getting the full report? Here are two ways to access it:

Subscribe to an All-Access pass to BI Intelligence and gain immediate access to this report and over 100 other expertly researched reports. As an added bonus, you'll also gain access to all future reports and daily newsletters to ensure you stay ahead of the curve and benefit personally and professionally. >> START A MEMBERSHIP

Purchase & download the full report from our research store. >> BUY THE REPORT

Show more