News / As returns and fraud costs rise, integrator turns to AI for help

UPS is experimenting with AI to combat rising fraud in product returns, running trials with a number of customers using the tech to detect fraudulent returns.

With returns volumes as well as associated fraud on the rise, such tools look likely to find a warm welcome from merchants.

The pilot schemes deployed two different approaches to fraud detection. One, a risk assessment model for specific returns based on shopper behaviour and patterns, which flags suspicious returns for review before a refund is issued.

The second uses AI to compare the returned products with images from the retailer’s catalogue to look for discrepancies, such as a missing logo or a mislabelled tag – the objective to identify cheap goods being returned in lieu of purchased items.

According to the integrator’s Happy Returns arm, more than 99% of items screened with the latter method were verified as legitimate, and scrutiny of flagged returns was performed within one day.

Happy Returns said the average prevented loss of flagged items amounted to $218 per return for retailers.

Returns are an awkward issue for merchants. Consumer surveys show that return policies are among the top three reasons for shoppers in selecting a product, and that poor return experiences would cause them to stop buying from the retailer in question.

On the other hand, both the volume of returns and fraud keep climbing.

The US National Retail Federation (NRF) estimates that 19.3% of online sales were returned last year, which amounted to $849.9bn, while Parcelhero calculated that UK consumers returned £1.55bn  ($2.09bn) of goods purchased in November and December.

The NRF found 9% of these returns were fraudulent. According to the 2026 Returns Fraud Report, published by LiquiDonate, retailers estimate that as much as 15% of returns involve some form of fraud, which puts the fraud rate on returns of online purchases at 24.5%, compared with 8.9% for returns of in-store purchases.

For merchants this adds up to painful losses. To begin with, the report puts the average cost of returns to the retailer at $25 to $30, while ParcelHero found the cost of a return could be up to 66% of the product’s original price.

And on top of refunds, sellers face the costs of processing, labour, inspection, and shipping, which can add up quickly and hurt small businesses particularly severely, it warned.

Costs of returns keep climbing. One factor is the trend to place merchandise in fulfilment centres close to metropolitan areas, to shorten delivery time rather than use centralised warehousing, noted Scott MacRae, CEO of Landmark Global. As products are not always returned to the location they were shipped from, challenges can arise over duty recovery, documentation, and inventory reconciliation, he pointed out. Moreover, those facilities have higher costs than large central warehouses typically located in rural areas.

Amazon is about to deploy a strategy to leverage money tied up to cope with returns. In March, it will start a new policy called Delivery Date Plus 7, under which it intends to hold sellers’ funds for an extra seven days – a step presented as a measure to protect buyers and cover returns.

In a LinkedIn post, industry consultant Max Sigurdson-Scott commented that this put between $3bn and $6bn of funds in the hands of Amazon, creating “a rolling pool of cash” on a weekly basis, which it could invest profitably.

By shifting the cashflow curve by one week, it could create “an entirely new profit source that sellers will never see”, he noted.