How Fraud Analysts Identify Emerging Fraud Trends & Improve AI Decision-Making

At NoFraud, our powerful decision engine is more than a set-it-and-forget-it solution. It’s constantly learning. It is trained on massive datasets from diverse sources encompassing historical transactions from our vast network of merchants, public records, third-party databases, and behavioral analytics. This helps create rich profiles that improve the accuracy of pass/fail decisions on every transaction.

While fancy algorithms are great at detecting suspicious patterns in large datasets, the decision engine also relies on human judgment. Analysts use their experience to interpret the data and make informed decisions. They might reach out to confirm a questionable order with quick email verification. Or if it’s a stolen credit card being used, they’ll act fast to protect you and prevent the fraudulent purchase from going through. Plus, when fraud analysts find emerging fraud patterns, they use that information to teach the decision engine so it prevents future attacks.

In this article, Anthea Hansen, a fraud expert and Australia Fraud Operations Team Lead at NoFraud, who reviews more than 55,000 transactions per year, shares how her team:

• Investigates transactions 
• Identifies emerging fraud trends
• Improves the AI-powered decision engine to approve more orders

“It’s important to string together all the datapoints to get context on each transaction.”

Fraud analysts don’t just focus on the final purchase. They examine the entire order journey to better understand customer buying patterns. 

“I studied journalism. So, I love the investigative side of researching transactions,” Anthea shares. “My job is all about dissecting transactions and using datapoints to tell stories.”

Fraud analysts meticulously scrutinize every order detail. They compare billing and shipping addresses for inconsistencies, check email addresses to see if they match the account information, and analyze order history to understand each shopper’s typical buying patterns. Even the type of products purchased and the device used to order can raise red flags if they seem unusual.

When an Order is Flagged for Review, But Approved

“Rigid fraud prevention solutions might have a lower threshold for risk and would automatically fail these transactions.”

Anthea shares an example of an order flagged for review where the transaction details didn’t line up, but the order was approved. “Our decision engine is pretty powerful, so orders flagged for review are handled on a case-by-case basis.” Whether it’s two or 10 datapoints that don’t align and indicate potential fraud, human expertise is required to make the final decision. “It’s important to string together all the datapoints to get context on each transaction,” says Anthea.

Anthea notes a high-value order, flagged for review, that did not link to the shipping address. Upon further review, she found the shopper did link to a company in the area. A builder connected to a local construction company placed an order for synthetic turf, so the transaction was approved.

“Rigid fraud prevention solutions have a lower threshold for risk and automatically fail transactions that are actually legitimate. Context is so important, and that’s why human review is essential for merchants,” Anthea emphasizes. “Otherwise, you’re failing orders, disappointing good customers, and missing out on legitimate revenue.“

Here’s a helpful resource on how rigid fraud tools can cost you good sales.

“The best defense is proactivity, so we’re constantly combing through suspicious activity.”

Fraudsters constantly devise new tricks to target vulnerabilities within a merchant’s eCommerce operations. They’re looking for workarounds that won’t trigger alerts from a fraud prevention solution. Fraud analysts stay on top of the latest trends and emerging threats by examining transactions for early signs that may indicate a fraud attack.

“The best defense is proactivity, so we’re constantly combing through suspicious activity,” Anthea shares. “We scrutinize every detail to identify activity that may indicate a fraud attack is coming; and once we find the trigger, we develop a hypothesis and test it.”

When Bots Visit Your Shop: A Precursor

In a recent case, the NoFraud team found that bots were adding items to cart, but not checking out. This tactic is known as cookie jarring and is a precursor to a fraud attack. Fraudsters know that adding items to an online shopping cart builds credibility in the eyes of merchants, and will use this tactic to build credibility before they card test or attempt an account takeover (ATO).

While AI excels at identifying patterns in vast datasets, it needs fraud analysts to feed it data. Analysts use their expertise to create and refine specific rules for the decision engine, allowing it to leverage AI’s capabilities effectively. This collaborative approach is essential in the ever-evolving world of fraud. As organized crime rings increasingly rely on bots for large-scale attacks, analysts stay on top of the latest trends and emerging threats. By feeding this intelligence back into the AI model, they empower it to adapt and proactively identify even novel threats before they cause damage.

To learn more from our experts, check out:

• How to Keep Fraud From Spoiling Your Customer Experience
• Fraud Expert Shares Why Basic Fraud Prevention Isn’t Enough for eCommerce
• Combating eCommerce Fraud: Essential Strategies From NoFraud’s Anti-Fraud Squad
• 4 Fraud Analyst Strategies to Fight Fraud and Reduce Chargebacks

Fraud analysts are essential partners in the AI learning process. By providing feedback on false positives, identifying emerging threats, and developing new fraud rules, their human discretion continuously improves the decision engine’s ability to approve legitimate orders while stopping fraudulent activity in its tracks. Discover how NoFraud’s powerful decision engine helps you safely approve more orders — book a demo.