Data from Happy Returns, a UPS-owned reverse logistics company, reveals that almost 10% of retail returns in the U.S. involve fraud, leading to an annual loss of about $76.5 billion for retailers. To combat this issue, Happy Returns is trialling a new AI tool designed to identify suspicious returns before processing refunds.
Understanding Return Fraud
Return fraud may seem innocuous at first glance. A customer might ask for a refund on a legitimate item but return something completely different, often cheaper or damaged. Retailers frequently issue refunds without inspecting returns, which enables such fraud to occur more easily and drives up costs.
Industry statistics project that retailers will manage around $850 billion in returned goods by 2025, representing nearly 16% of total retail sales. The same sources estimate that about 9% of these returns could be fraudulent, with many shoppers admitting to abusing return policies. However, Happy Returns employs in-person product verification alongside AI-driven auditing to keep its fraud detection rate significantly lower than industry averages.
The Shift to Boxless Returns
Happy Returns has established roughly 8,000 drop-off points inside stores like Ulta Beauty and Staples, where consumers can return items without boxes or shipping labels, often receiving instant refunds. While fraud attempts can still occur, the face-to-face nature of these returns, combined with verification and reporting, lowers confirmed instances of fraud compared to broader industry metrics.
By eliminating the need for boxes, Happy Returns effectively targets common fraud tactics such as fake tracking numbers and empty returns. As a company statement puts it, “If you haven’t touched the product, you can’t really know if what’s returned matches what was sold.” Moreover, physical handling has been observed to deter some potential fraudsters.
Utilizing AI to Mitigate Return Fraud
Amid rising return volumes this holiday season, Happy Returns is trialling its AI system called Return Vision with select retailers, including Everlane and Under Armour. This tool begins assessing returns as soon as a shopper initiates the process online, analyzing patterns related to timing, frequency, and location. If certain signals align suspiciously, the return is flagged for further review.
At the drop-off centers, workers can scan product barcodes and compare them to photos of the original item. Any discrepancies can be addressed immediately. Flagged packages sent to Happy Returns’ hubs are meticulously inspected by auditors who capture detailed images for comparison against official product images.
Initial AI Performance
Although still in the pilot phase, early results from Return Vision are promising. Less than 1% of revenue displayed within the network is classified as high risk, with approximately 10% of returns later confirmed as fraudulent. The average loss averted per confirmed case stands at just over $200. The system aims to focus on cases posing higher risks, allowing most shoppers to return items without any delays.
The Retail Landscape and AI’s Role
Happy Returns isn’t alone in leveraging AI to combat return fraud; other companies like Amazon and FedEx are adopting similar strategies. Across the retail sector, about 85% of merchants surveyed employ AI or machine learning for fraud detection, although opinions on effectiveness vary. Happy Returns contends that combining behavioral analysis with physical verification can address shortcomings often seen in systems reliant solely on data.
Concluding Thoughts
The dynamics surrounding retail returns—and fraudulent activities related to them—are evolving. The convenience of easy drop-offs and immediate refunds benefits customers but presents new risks for retailers. Happy Returns believes that integrating AI with hands-on inspections can help restore balance, with initial results hinting at potential success, though it’s not a complete solution. As fraud tactics become more sophisticated, retailers must also adapt more swiftly.
What do you think? Should retailers prioritize instant refunds to enhance convenience, or would it be wiser to slow down to minimize fraud? Share your thoughts.
