How AI Can Help Prevent Fraud and Save Retailers Millions
One of the most pressing concerns that keeps retail professionals up at night is how to combat fraud. Retailers could lose upwards of $71 billion from fraudulent online transactions over the next few years, yet some executives feel that publicly acknowledging a fraud issue would harm their brand. As a result, there's a lack of conversation around fraud prevention and often little visibility into current operations.
What Are Retailers Worrying About
One of the most significant fraud concerns merchants face today are false positives — i.e., transactions attempted by legitimate customers that are tagged as suspicious by fraud prevention systems, ultimately leaving money on the table. Because their effect is so difficult to accurately measure, false positives are often ignored, and their cost greatly underestimated. However, a majority of retailers say that fraudulent transactions that aren't detected cost more than a legitimate transaction that's inaccurately declined, despite some evidence that the opposite is true. What’s more, relatively few companies track false positives. Thirty percent of merchants said they don’t attempt to measure them, and 42 percent don’t even know what a false positive rate is, according to our 2018 Fraud Operations Study.
In addition, one the most pervasive types of fraud, account takeover, is also one of the hardest to identify, according to 38 percent of fraud teams. As more retailers have loyal customers store payment credentials to take friction out of the checkout process, e-commerce fraud is moving quickly from the transaction level to the account level. Because consumers tend to reuse passwords across multiple online accounts, information stolen in any of the thousands of data breaches occurring each year can be used at many different retailer sites. A valid username/password can make any malicious actor look like a trusted customer. And once logged into a legitimate account, fraudsters can steal from merchants in any number of ways.
The Role of Machine Learning in Fraud Prevention
Retailers are falling behind because they're not discussing fraud challenges and their team doesn't have the bandwidth or a clear understanding of the best solution. This is where artificial intelligence- (AI) and machine learning- (ML) based fraud protections come into play.
Merchants might hear AI and ML and think of them only as popular buzzwords. Many don’t truly understand the capabilities of these technologies. On top of that, there are too many solutions to count, and it can be overwhelming to pinpoint the differences and benefits of each, as well as determining what will have the most measurable impact on the organization.
Let’s take a step back and not just throw terms around, but identify four key considerations to take into account when determining the right AI security solutions to implement to uplevel fraud protection.
Trust You Have the Right Partner
The first step when vetting AI-based security solutions is to identify the organization’s fraud prevention goals. Some of the most burning questions merchants can ask themselves when reviewing which solution to employ include (but are not limited to):
- Can the solution react quickly?
- Will it scale as my network grows?
- Is the solution data agnostic?
- How does it handle false positives?
- What types of benefit and risk analyses does it provide?
Once these questions are addressed at a high level, retail leaders should educate themselves on the specific capabilities of the solutions out there, keeping the following considerations in mind.
‘Explainability’ and How it Helps Retailers Avoid False Positives
A common misconception is that AI-based protections can be biased, resulting in false positives (e.g., an AI algorithm rejects a customer for a loan for no clear or apparent reason). This is just another example of how vital education is, and how detrimental it can be for an organization to assume machines are capable of doing things they’re not.
Immediacy is Key
Consumers demand immediacy in all retail experiences, from short wait times to easy online transactions. However, with immediacy comes risk. Fraud happens in the blink of an eye and must be detected instantaneously. The best AI solutions are built for rapid detection and can adapt to different models and systems. They learn from previous flags of fraud and identify those patterns for future detection.
Doubling Down on Deep Learning
The crux of AI for fraud protection is determining what malicious behavior looks like compared to legitimate transactions. There's no one-stop-shop algorithm that can do this. However, deep learning algorithms, including both supervised and unsupervised machine learning, can provide retailers with comprehensive protection. The former looks at historical data stored in a log (often rules marked by IT teams) to inform how it flags fraud. The latter observes real-time behavior patterns to make the call if the attempted transaction is fraud or not, essentially educating itself in the moment.
Not all AI solutions are created equal, but there are key factors to account for that will help fine-tune the decision-making process when considering which ones to implement. Fraud protection should not be a gray area. Every retail professional has a part to play — from the CIO to IT to the customer experience manager — and education is key in deciding the most effective way to protect against the common enemy: fraud.
DJ Murphy is the editor-in-chief of Card Not Present, a media brand generating original news, information, education and inspiration for and about the companies and people operating in the card not present space.
Related story: Addressing Loss Prevention Across the Retail Ecosystem