Close this search box.

GUEST COMMENT Fraud-as-a-service: the underground network causing havoc with retailers’ return policies

Image: Adobe Stock

Daniel Marchetti is customer advisory, principal business solution manager, AML & fraud, SAS UK & Ireland

According to research, on average, retailers receive around 16% of items purchased back as a return. The NRF (National Retail Federation) says retailers are losing around 10% of the price of returned items to fraud. It’s costing retailers billions globally.

Networks of fraudsters are using the dark web and other methods to cut consumers a deal on products, by buying items and then returning counterfeit products in their place – which are often uncannily similar to the real thing.

These fraudsters run sophisticated operations and are getting to know retailers’ processes inside out. In the world of omnichannel and with many retailers offering free returns, fraud has become a very real crisis for retailers. And this is only set to get worse.

As consumers struggle to make ends meet during the cost-of-living crisis combined with the UK now entering a recession, it’s not just sophisticated fraudsters that are likely to take advantage but potentially the everyday person will look to abuse return policies too.

This is increasingly likely at a time of year, just past the festive season, when many people will be returning goods. In fact, according to our own research of 3,000 consumers in the UK & Ireland, 55% expected to return unwanted items after Christmas. So it’s clear the extent of just how busy retailers will be processing returns into 2023. Retailers need to have the right tools in place to help identify and prevent fraudulent returns and put a stop to organised groups performing fraud-as-a-service.

How are fraudsters providing underground services to save consumers money on products?

Returns fraud is becoming a major concern for retailers – especially at a time when they are expecting sales to take a hit due to the recession. I work closely with many major retailers and they recognise just how much the pandemic has accelerated returns fraud and they want to take action to put measures in place to reduce it and prevent it.

We’re seeing a large increase in anomalous returns and new methods used by organised groups are emerging. The whole approach is becoming increasingly sophisticated. As consumers use more online channels to purchase products, more information is being shared on the dark web on how to defraud specific brands.

Guides or manuals are being created and circulated online about how to specifically approach defrauding the brands that are being targeted. These guides include the specific steps to get through customer service, the details of their policies and how to circumvent them. The guides tell consumers what they need to do to successfully get a refund and keep the product they have bought. It is very worrying for retailers and is connected to the increase in false refunds.

Returns fraud only likely to increase

The acceleration of digital transformation to facilitate mass remote working, ecommerce and online communication is partly behind the rise of returns fraud.

It has created opportunities for fraudsters, quick to take advantage of the proliferation of new technology and new “way of life” into their scams and fraud schemes. Online methods such as phishing and smishing (text messages) have increased and the fraudsters have kept pace with the topics of the day.

However, companies need to develop a strategy that enables the deployment of appropriate tactics to manage these new or increasing risks.

How can retailers combat and identify potential return fraudsters?

Everything starts with data. The data available to retailers through their transactions and returns, as well as supplier files, employee files, and financials can be used. If this data is collected, organised and analysed correctly, it can detect the different fraud typologies that they should be looking for.

Using AI and analytics is the only way larger retailers can analyse these major datasets quickly enough to react to a possible crime being committed. Most typical AI systems work through a number of stages.

Data ingestion: This is where an AI provider predefines etls (extract, transform, loads) and data quality and data management processes. Data management has to be robust and accurate to ensure all that follows can be carried out to effectively identify and prevent fraudulent activity.

Detection: This is typically where a retailer working alongside their AI software provider will implement specific models or specific business rules over the different fraud typologies that are needed.

For example, you can have an anomaly detection model working which detects unusual activity or patterns of unusual activity.

You may also need a challenger model to target a specific fraud typology like returns fraud. So there are different packages for each kind of role typology being deployed in the detection stage.

Investigation: The user interface of your AI software will send you alerts when that unusual activity is detected. These alerts are usually generated in batch or in real time – whatever is pertinent for each fraud typology.

Reporting: This comprises dashboards, which you can review whenever you need to access information and see what’s happening. How many alerts do I have? What’s the value? What’s the volume by channel, by country – whatever you’d like to investigate. And these reports can be used by people with various roles in the business to report back on.

The best solution is a hybrid analytical approach to fraud detection, helping reduce false positives by incorporating the latest anti-fraud measures, advanced statistical models, and deterministic and associative analysis.

These solutions work by analysing components and examining links and associations among counter parties and the various elements from the individual case reporting, highlighting potentially fraudulent practices.

When a fraud referral or case is completed, the results are stored to use in future models, which improves performance and efficiency.

Depending on the fraud types, a model developed specifically for an individual retailer ensures the most suitable application of machine learning in the identification of sophisticated and evolutionary fraud.

Retailers need to make tackling fraud a priority

Fraud isn’t just a problem for retailers. It’s a real challenge individuals, businesses and government are facing. Official figures in the UK reveal 41% of all crimes against individuals in the year to June 2022 was a form of fraud. It’s astonishing and something that needs to be tackled effectively as it eats into company profits and time – and can ruin people’s lives.

Retailers need to act now to tackle fraud to protect their bottom line as we enter what could be the UK’s longest recession in generations. But it’s also about protecting the customer experience by not introducing unnecessary fraud checks, as well as minimising reputational damage.

I know from working with many retailers that this is becoming a higher priority and as fraud becomes an even bigger issue for society to tackle, it’s likely many businesses will follow suit to identify and prevent all forms of fraud affecting their operations.

Daniel Marchetti is customer advisory, principal business solution manager, AML & fraud, SAS UK & Ireland

Read More

Register for Newsletter

Group 4 Copy 3Created with Sketch.

Receive 3 newsletters per week

Group 3Created with Sketch.

Gain access to all Top500 research

Group 4Created with Sketch.

Personalise your experience on