Blog / Fraud analytics , Machine learning

Why there’s no single indicator for fighting fraud

There are several factors you can use to identify fraud ahead of time. We cover why there’s no single best indicator and how you should use data to better prevent growing fraud.

Why there’s no single indicator for fighting fraud

Bad-faith actors are always finding new ways of committing fraud under the radar.

In a recent survey, 38% of respondents told us that they’re not just seeing more fraud in their business but new types of fraud as well. And an increasing portion of these are committed by customers.

Pair that with the fact that fraud is costing organizations an average of $4.5 million a year and it’s clear that you need to stay a step ahead. But what’s the most important indicator or factor to identify fraud?

Well, a good fraud prevention strategy addresses the full range of factors that are available to you. To help you out, we conducted research into the factors your peers are using to identify and stop bad actors.

What factors are merchants looking at?

We surveyed merchants across the Retail, Travel & Hospitality, Digital Goods, Marketplace, and Subscription industries to get an idea of the different factors they track and which they believe are most important.

Here are some key takeaways:

  • A third of respondents believe customer profile and order content are the most significant factors.

  • Over a quarter of merchants say order content is the most important factor.

  • Almost half of all merchants say that account history is a top three factor.

There’s a clear difference between how important each of these factors are to merchants. But the truth is, you should be looking at all of them.

Let’s look into the details around each factor to better understand how they help detect fraud.

Order content

Why is it important?

Tracking order content helps make sense of what customers are purchasing and how much of it.

Take Event Ticketing merchants – tickets are prime fodder for scalper fraud, namely through bots or scripts. Most recently, tickets to the 2023 Eurovision final were sold out within 36 minutes and later resold through underground networks at more than £11,000 per ticket.

But this doesn’t just apply to Digital Goods. Reseller and bot activity has been called out as a new fraud trend by 53% of merchants across all industries. So it makes sense to keep a close eye on order content.

How can you use order content to detect fraud?

If a fraudster is using bots or scripts to target your business, this will lead to multiple bulk purchases being made very quickly.

Work closely with your Buying teams to ensure you know when high-risk items are in stock. And keep up with your Marketing team’s campaigns to know which items to watch.

Order value

Why is it important?

Order value is the total monetary value of a customer’s purchase. Tracking this makes it easy to flag high-risk orders. For example, a £2000 order from a low-cost fashion retailer is a lot more suspicious than an order of the same value from a luxury brand.

With that said, order value can be used by every industry.

How can you use order value to detect fraud?

An uncharacteristically high order value could point to account takeover (ATO) fraud or potential payment fraud. It helps here to cross-reference accounts against breached credentials databases.

You might also want to set thresholds that flag purchases over a certain amount. This amount will depend on your business and what's normal for your customers.

Location data

Why is it important?

Location data helps you track where a customer is when they make a purchase and the delivery address.

It’s unlikely that a fraudster will be in the same location as the genuine account holder. So a sudden change in delivery address could suggest ATO fraud.

How can you use location data to detect fraud?

Certain addresses can be flagged as risky if they are known pick-up spots for fraudsters, or if the billing and shipping addresses look suspicious. For example, if they don’t match, are unusually far apart, and/or frequently change.

Shared industry data

Why is it important?

Shared industry data gives you insight into fraud trends and patterns that other businesses are seeing in your industry.

Only 18% of businesses place it in the top three categories. This likely stems from roadblocks merchants face when it comes to both sharing and accessing this data.

Concerns around security and losing competitive advantage make sense — but they ultimately don’t measure up to the value of shared insights.

How can you use shared industry data to detect fraud?

Using these wider datasets helps businesses analyze fraud patterns and trends, especially when they’re just emerging.

Fraudsters rarely act alone. And when they find a vulnerability, they tend to work in droves. So having an idea of how other ecommerce businesses are under attack is key to protecting your business.

Why your customer activity might indicate fraud

Monitoring your customer activity is vital. But you need to know what “normal” behavior for your genuine buyers looks like first. It then becomes much easier to spot fraudy patterns to detect future attacks.

We looked at which customer activities online merchants monitor to prevent ATO fraud specifically. Merchants are keeping a wide range of behavior on their radar — which is a good sign.

But what’s not a good sign is just how few merchants are regularly tracking all of these activities. Merchants track password updates the most. But the global average is still only 60%.

On average, less than half of merchants monitor other metrics like activity from new devices and changes to important user features (email address, phone number, etc.).

But these are key changes you need to keep track of. A customer logging in from a new device or changing their details and immediately making an expensive purchase could be a sign of ATO fraud.

Truly effective fraud-fighting means using multiple factors to form the bigger picture of fraud threats facing your business.

Monitor what works for your business

Ultimately, there isn't one single indicator for fraud that every business can rely on in every context. And fraud analysts alone can’t realistically address all of the different factors and activities you need to take into account.

Machine learning (ML) models can help analyze and draw insights from your data, noting fraud patterns and recognizing fraudsters amongst genuine customers. All to leave more time and resources to your analysts to conduct deeper fraud investigations.

For more on how online merchants are combating fraud and what you can do to better protect your business, download our Fraud & Payments Survey.

Related content