E-Commerce Fraud Prevention Requires End-to-End Transaction Visibility

This year, for the first time in history, global e-commerce will account for over a fifth of all retail sales. But 2023 will also bring another less auspicious milestone: chargeback fraud will cost merchants an estimated US$125 billion globally — a gargantuan sum that will eat into digital sellers’ razor-thin margins.

Illegitimate chargebacks — also known as friendly fraud — are a monumental problem for merchants, with half of sellers claiming that dishonest payment disputes are their largest financial drain. For small to mid-sized companies, friendly fraud could cut gross revenue by up to 1.5%, potentially making the difference between collapse and continued commercial viability.

Historically, virtually all payment fraud detection has been retroactive, taking place after a suspected attack has occurred — but beating fraud, including chargebacks, in the age of ubiquitous e-commerce requires a more intelligent approach. To stay ahead of fraudsters, brands need to use new, technologically enhanced tools to combat fraud at each stage of the payment journey.

Following are strategies for security-conscious merchants to safeguard their finances throughout the entire digital transaction process.

Take a Data-Driven Approach

Traditional fraud prevention focuses on identifying past attacks because there hasn’t been enough data available to take a more proactive and preventative approach. Today, though, that is changing.

By their nature, e-commerce transactions generate enormous amounts of data at every step of the transaction journey. New machine learning (ML) solutions and advanced analytics make it possible to collect and analyze that data in real-time, spotting patterns that betray suspicious activity to give an early warning of potential fraud.

However, it’s important to remember that ML tools work by spotting patterns. That means they get smarter over time — but it also means they aren’t always adept at managing novel situations.

Don’t put your complete trust in a “black box” algorithm. Make sure you understand what’s going on under the hood and have human experts on hand to help manage unexpected situations such as sudden (but non-fraudulent) shifts in demand patterns or consumer behavior.

Find Clues in Related Purchases

One area where ML tools can be especially powerful is in spotting purchasing patterns that suggest fraudulent behavior in the offing, as shared by my colleague Dor Bank on Medium.

Suppose a customer buys the same items at or around the same time each month. In that case, a purchase consistent with their past behavior is unlikely to result from a stolen credit card — and thus, a chargeback on that purchase is quite likely to be an instance of friendly fraud.

By the same token, if a consumer’s typical activity suddenly changes — for instance, if instead of buying one product a month, they suddenly buy two dozen high-value products in quick succession — there’s a good chance that a card-not-present attack or another form of payment fraud has indeed taken place.

Such methods can use backward-looking analysis to flag previous transactions that appear fraudulent based on subsequent behavior and use past transactions to flag later purchases for additional review preemptively.

Pay Attention to Contextual Clues

Incorporating contextual clues, such as after-sales interactions between merchants and consumers, can also enrich fraud detection analytics.

A message to customer support from a shopper who says they don’t recognize an order might indicate that traditional fraud occurred. On the other hand, a purchase cancellation request from a customer who then goes on to submit a chargeback claim leaves little doubt that friendly fraud is afoot.

Less obvious customer support interactions, like a request to change delivery details, can also be a risk factor because fraudsters sometimes order items using legitimate addresses to beat shipping verification systems, then divert packages en route.

Sometimes a degree of common sense is also needed. If an order involves shipping a bulky and expensive garage door system to a high-rise studio apartment, for instance, something strange is likely going on.

Prioritize the Customer Experience

Early in the consumer journey, it’s possible to collect valuable data relating to factors such as the amount of time consumers spend on different product pages or how long they take to enter personal details and complete ID verification checks.

But be careful; it’s essential to make such measures as hassle-free as possible to avoid degrading the customer experience. This methodology requires a sophisticated analytic approach to prevent both false negatives, which let fraudsters slip through the cracks, and false positives, which improperly reject legitimate transactions.

In digital commerce, it’s easy for customers to click away to a competitor’s website, so it’s essential to find solutions that combine a high level of fraud protection with a seamless sales process and that can reliably identify fraud without increasing friction for legitimate customers.

Be Proactive Across the Payment Journey

In all these areas, merchants need to find ways to join the dots between fraud prevention processes, chargeback mitigation processes, and the consumer experience.

It’s no longer enough to focus on one area of the customer journey or one stage in the transaction process. Merchants need an intelligent and integrated end-to-end solution to reduce fraud without getting in the way of legitimate shoppers.

Creating an effective payment fraud mitigation system is one of the biggest challenges e-commerce merchants face. The stakes are high; get this wrong, and they risk an erosion of profits, decreased customer satisfaction, higher operating costs, and the prospect of sanctions from the big payment card networks.

Fortunately, new technologies — including well-designed ML and automated analytics solutions — now make it possible for online sellers to take the battle to fraudsters and more effectively beat both traditional and friendly fraud.

The goal is to adopt an end-to-end approach and to be proactive about identifying and defeating fraud at all stages of the sales journey by preventing it before it happens. This strategy involves neutralizing new attacks in real time and implementing efficient and effective systems to counter after-sale chargeback fraud.