Return abuse costs retailers roughly six times more than outright return fraud. Most stores still aim their return policy at the fraud.
That gap is the central finding in the 2026 Total Retail Loss Benchmark Report from Appriss Retail, a returns and loss-prevention vendor. Its figures come from transaction analytics across 250 million unique customer identifiers plus a survey of 1,020 US shoppers run in December 2025. The report separates returns loss into two buckets. Fraud, meaning fake receipts and returning stolen merchandise, accounts for 2% of all returns and $14 billion. Abuse, meaning excessive but technically legitimate returns, accounts for 12% of all returns and $86 billion.
The thing everyone writes policy about is the smaller number.
The honest caveat: the numbers do not all agree
Before anyone builds a plan on one statistic, the industry data conflicts.
The National Retail Federation and Happy Returns, a UPS company, reported in their 2025 Retail Returns Landscape that 9% of all returns are fraudulent. That release, published 15 October 2025, was based on surveys of 2,006 consumers and 358 ecommerce professionals at US merchants with over $500 million in revenue. Appriss puts pure fraud at 2%.
The two are measuring different things. NRF asked retail professionals what share of returns they believe are fraudulent, which lumps in a lot of behaviour that Appriss classifies as abuse. Appriss scored actual transactions. Neither number is wrong, but the practical read is the same either way: a meaningful slice of your returns is not a normal customer changing their mind, and most of that slice is not criminal.
Scale, for context. NRF and Happy Returns estimated retailers would see 15.8% of annual sales returned in 2025, totalling $849.9 billion, with online return rates higher at an estimated 19.3%. Appriss put 2025 returns at $706 billion. Different methodologies, different totals, same order of magnitude.
Why abuse is harder to catch than fraud
Fraud has a signature. Empty boxes, counterfeit swaps, receipts that do not exist. NRF found that retailers tracking these incidents reported increases in overstated quantity of returns (71%), the empty box or “box of rocks” (65%) and decoy returns such as counterfeits (64%).
Abuse has no signature. It looks exactly like a good customer.
Someone orders three sizes intending to keep one. Someone wears a dress once and sends it back. Someone returns 60% of what they buy and genuinely believes that is fine, because your policy says it is fine. Appriss found just under half of shoppers, 45% in the NRF survey, consider “bending the truth” acceptable when making a return, especially if they are unhappy with the item.
Here is the part that should change how you think about this. Appriss reports that processing a return costs an average of 30% of the item’s value, which works out to roughly $211 billion across all returns, fraudulent and legitimate alike. You are not just losing the margin on abusive returns. You are paying handling costs on every return, including the honest ones.
What to do: stop tracking “return rate” as one number. Split it into return rate by customer, by SKU, and by reason code. Abuse hides in the average.
The blanket policy trap
The instinct when returns costs rise is to tighten the policy for everyone. Shorter windows. Restocking fees. Receipt required. Appriss argues that is precisely backwards, and the customer data is the reason.
According to the report, the top 1% of customers generate up to 50% of total sales and return 8% less than the average shopper. But because they buy more, they hit policy tripwires more often. Receiptless limits, short windows, ID checks. A blanket policy designed to catch abusers ends up interrogating your best accounts.
They do not argue. They just stop buying. Appriss found 47% of shoppers have skipped a purchase because of return policy concerns, rising to 56% among frequent returners, and 13% have switched retailers over inconsistent policies between channels.
That last one matters for anyone running a store alongside a physical location or a marketplace listing. If a customer is flagged in one channel and waved through in another, you have not built a policy. You have built a gap.
What to do: before you shorten your return window, model what it costs you in lost first purchases. For most small stores that number is larger than the abuse it prevents.
The middle option almost nobody offers
Most return systems have two outcomes: approve or deny. Appriss makes the case for a third, which it calls “warn and approve.”
The mechanic is simple. Low-risk returns go through with no friction. High-risk patterns get approved and get a notification explaining that their return rate is outside normal range. Only returns hitting real fraud or extreme abuse thresholds are declined.
Appriss reports this three-tier approach can reduce abusive returns by 90% without measurable loyalty damage, and cites a 12% drop in in-store returns and a 6.5% drop in online returns across its client base. Treat those as vendor-reported results from the company selling the solution, not independent findings. The logic behind them holds up regardless: most people committing abuse do not know they are doing it, so telling them is cheaper than punishing them.
Customer appetite for this appears to exist. Appriss found 61% of shoppers are open to technology-driven return eligibility checks. They want to understand how the decision was made, not to avoid the decision.
What to do: add one automated email that fires when a customer’s return rate crosses a threshold you set. Not a threat. A plain note that their returns are unusually high and that you want to help them order better next time. This is a straightforward job for a workflow tool if your platform will not do it natively, and our comparison of Zapier, Make and n8n for online stores covers which handles conditional order logic best.
Where instant refunds fit
The pressure runs the other way too. NRF found 76% of consumers say they are more likely to choose a return option offering an instant refund or exchange, and 82% cite free returns as a major purchase consideration, up from 76% the previous year.
Instant and returnless refunds solve a real problem. When shipping a $22 item back costs $9 to process, the return costs more than the write-off. Shopify’s own returnless refunds documentation frames it as a cost decision rather than a generosity one, and that is the right frame.
The risk is that refunding before inspection moves cash out the door before anyone knows what is in the box. If you offer it, cap it by item value and by customer history, not as a blanket setting.
What to actually do this quarter
- Pull a return rate per customer report. Sort descending. Look at the top 20. You will probably recognise two or three names, and at least one will be a customer you thought was great.
- Add reason codes if you do not have them. Without them you cannot tell a sizing problem from a product description problem from abuse.
- Check your return rate by SKU. A single product with a 40% return rate is a listing or sizing fix, not a fraud problem. That fix is usually cheaper than any policy change.
- Set a returnless refund threshold. Compare your actual inbound processing cost against item value. Below that line, refund and let them keep it.
- Close the channel gap. If a customer is flagged anywhere, they should be flagged everywhere.
If step one shows you need real tooling rather than spreadsheets, our roundup of return-management software options compares what is available at small-store price points. Stores already running automated decisioning on other parts of the funnel have an advantage here, and the pattern is consistent with what we found in stores running their own AI agents.
One more note on the fraud side. NRF found 85% of surveyed retailers are already using AI to detect or prevent return fraud, which means fraudsters have moved on to the stores that are not. Small merchants are the soft target now, the same dynamic covered in our piece on how stores became the internet’s top bot target.
The takeaway
Your return problem is probably not a fraud problem. It is a measurement problem wearing a fraud costume. Six dollars of every seven lost to returns comes from customers who think they are behaving normally, and the cheapest intervention is telling them they are not.
Sources
- Appriss Retail, The 2026 Total Retail Loss Benchmark Report (vendor research; methodology based on 1,020 US consumers surveyed December 2025 and analytics from 250 million unique customer identifiers)
- National Retail Federation and Happy Returns, Consumers Expected to Return Nearly $850 Billion in Merchandise in 2025, 15 October 2025
- National Retail Federation, 2025 Retail Returns Landscape
- Shopify, Returnless Refunds: How They Work for Retailers (platform documentation)
Last reviewed: 30 July 2026.
Affiliate disclosure: e-commPartners may earn a commission if you purchase through links on this site, at no extra cost to you. This does not influence which tools or findings we cover.
