Returns Are Getting Expensive: How Retailers Are Rewriting Policy in 2026

The US return rate went down last year. Almost every article on this subject says the opposite, and the gap between those two statements is where the actually useful analysis lives.

Here's the correction. NRF and Happy Returns put 2025 US retail returns at $849.9 billion — 15.8% of sales, down from 16.9% in 2024. Online returns ran higher, at an estimated 19.3%. The long-run trend is still roughly a doubling since 2019, so nobody is claiming the problem is solved. But the most recent year moved the wrong way for the "returns are exploding" narrative.

What did not fall is the cost of handling each one. And that distinction — flat or falling volume against rising unit cost — explains everything retailers are currently doing to their policies.

The number that's actually moving

Processing a single return costs somewhere between $10 and $65 once you account for return shipping, labour, inspection and restocking. Industry estimates commonly put it around 30% of the item's value.

Two of the inputs to that figure have been rising independently of return volume: carrier rates and labour. When retailers are asked why they introduced fees, processing cost and carrier shipping cost come up in roughly equal measure — around 40% citing each.

So the pressure isn't that more things are coming back. It's that each thing coming back costs more than it did, against margins that haven't improved.

That last row is why this is an October article rather than a January one. The policy you set now is the policy that gets stress-tested in eight weeks — a point that connects directly to the 90-day Black Friday timeline and to the demand picture in what early signals say about holiday spend.

The finding that should settle the argument

Return Economics 2026, a benchmark of the UK's top 100 fashion retailers produced by Ingrid with Harper Concierge and Limesharp, contains the single most decision-relevant fact in this whole subject.

No brand that introduced return fees between 2023 and 2026 has subsequently removed them.

Not one. And fee levels have risen over the period, broadly tracking inflation.

Why the absence of reversals matters Every retailer that introduced a fee had the option to reverse it if conversion collapsed. None took it. That is not proof the fee was free of cost — it is evidence that those who tried it concluded the margin was worth what it cost them.

The adoption numbers around that: 23% of the UK's top 100 fashion retailers charged for returns in 2023, rising to 35% by 2026. More striking is where the growth came from. Affordable luxury went from 4% to 20%. Luxury went from 0% to 12% — segments that treated free returns as non-negotiable three years ago.

The pattern looks like peer-watching rather than conviction. Each segment waits to see whether the one above or below it survives the change, then follows.

Featured Recommendation AD · AFFILIATE

4.3 / 5.0

Katana

Real-time inventory and production management for makers, connecting Shopify, accounting, and your sales channels.

Best for: Scaling makers and ecommerce manufacturers

Visit Website

View details Compare

It's no longer free versus paid

The more interesting development is that the policy question has stopped being binary. Four distinct structures are now visible in the market:

• Flat fee. A fixed charge per return, usually deducted from the refund. Boohoo's £1.99 is a representative example — though even there, Premier members get one free return per order, so it isn't quite universal.
• Behaviour-triggered. ASOS monitors return rates over a rolling twelve-month window and applies fees only to customers who return 70% or more of what they buy. The majority keep the free experience; the cost is recovered from the segment generating it.
• Membership exemption. New Balance positions free returns as an explicit loyalty benefit rather than a default. The fee becomes a reason to join something.
• Tiered by method. Different pricing for drop-off versus home collection, letting the customer choose between convenience and cost.

The behaviour-triggered model deserves particular attention, because it's the one that treats return rate as a customer attribute rather than a company-wide problem. Most retailers report returns at product or category level. Very few segment their customer base by return behaviour — which means they cannot distinguish between a serial returner and a high-value customer who happens to return often, and those two people should not be handled the same way.

What a fee does and doesn't fix

This is where a lot of policy thinking goes wrong, and it's a straightforward distinction.

A return fee recovers cost. It does not reduce returns. Those are different objectives requiring different work, and conflating them produces a policy that annoys customers without solving the underlying problem.

More than half of returns are attributed to size or fit. Around 62% of consumers buy multiple sizes deliberately, intending to send back what doesn't fit. That behaviour isn't a customer failing — it's the rational response to sizing uncertainty in an environment where returns were free and shipping was free. You designed the incentive; they responded to it.

Charging £2 after the fact does nothing about uncertainty that existed before the purchase. What does:

• Fit-specific review data. "Runs small, size up" from twenty verified buyers removes more returns than any fee, which is one of the more underrated arguments for collecting and displaying reviews properly.
• Honest measurements, not just size labels. Garment dimensions, model height and size worn, material behaviour after washing.
• Imagery that shows the actual thing. Colour accuracy, scale references, video on the body — standard product page optimisation work that happens to have a returns dividend nobody measures.
• Exchange as the default path. An exchange keeps the revenue and often costs less to process than a refund. Most return flows still make refund the easiest option, in the same way most checkout flows default to the path that was easiest to build rather than the one that performs best.

Every one of those is a marketing job. Which is the point of this article.

The policy is a conversion element

Here's the framing that most coverage misses entirely, because it's written for operations teams.

Around 81% of shoppers consider the return policy before adding to cart, and roughly 71% say a poor return experience will stop them buying from a brand again. Which makes the returns policy a piece of pre-purchase merchandising that happens to be administered by finance and logistics.

Two consequences follow, and they pull in opposite directions:

First, tightening the policy has a conversion cost that will not appear in the returns line of your P&L. It appears as fewer orders, and it gets attributed to something else — the ad platform, the season, the market. If you introduce a fee, instrument the conversion rate change deliberately, or you'll never know what it cost you. Watch abandonment too, since a fee discovered late in the flow behaves exactly like any other unexpected cost, and your cart recovery sequence will quietly start working harder without anyone knowing why.

Second, if the policy is a conversion element, it needs to be visible where conversion happens. Most sites bury it in the footer. In gifting season especially, an extended or clearly stated returns window reduces the perceived risk of buying for someone else — but only if the shopper sees it before deciding, which is the same argument as for surfacing delivery dates before checkout rather than inside it.

The fraud question, handled honestly

Around 9% of 2025 returns were classified as fraudulent, and roughly 85% of retailers report deploying AI to detect and prevent return fraud. Wardrobing — wearing or using an item then returning it — accounts for the majority of that.

Two cautions on acting on this.

The first is that fraud detection produces false positives, and a false positive here means accusing a genuine customer of dishonesty. That is among the most reliable ways to permanently lose someone, and the cost of it doesn't appear in the fraud-prevention business case.

The second is proportion. Fraud is 9%. Fit and sizing is over half. A programme that spends heavily on catching the 9% while leaving the 50% untouched is optimising the smaller number because it feels more actionable.

What this means if you're not a major retailer

Most of the coverage on this subject is written about and for large operators. The smaller-retailer version is different in three ways.

Your free returns are a genuine differentiator now. As more of your larger competitors introduce fees, keeping returns free becomes a positioning asset rather than table stakes. That's worth saying loudly on the product page rather than treating as a policy detail — and it's worth modelling against customer lifetime value rather than per-order margin, using something like the approach in calculating and using CLV.

Your data is small enough to actually read. A large retailer needs models to find return patterns. You can read a hundred return reasons in an afternoon and find the three products causing most of it. That's a faster route to a lower return rate than any policy change.

You can afford to be generous selectively. Free returns for repeat customers, a fee for first-time high-risk orders, exchange incentives — the nuanced structures the big players are building with software, you can run with a rule and a spreadsheet.

What to do about it this quarter

• Separate cost recovery from return reduction in your planning, and fund both. They are not the same project.
• Segment return rate by customer, not just by product. This is the single most useful analysis available and almost nobody runs it.
• Make exchange the path of least resistance, and route exchangers into a post-purchase flow rather than treating the interaction as closed.
• Fix the three products causing the most returns before touching the policy that affects all of them.
• Check what your free-shipping threshold is encouraging. A threshold that nudges people into adding a second size to qualify is funding its own return, which is worth modelling alongside the rest of your average order value work.
• Instrument conversion before any policy change, so the cost is measurable rather than inferred.
• Put the policy on the product page if it's good, and rewrite it if it isn't. Either way, stop hiding it — a returns page is a website content decision as much as a legal one.
• Plan for January now. It's the busiest return month, and the experience people have then determines whether the Q4 cohort you spent so much acquiring — see peak-season paid media — becomes a retained customer or a one-off.

The short version

The US return rate fell in 2025, to 15.8% from 16.9%, which contradicts most of what's being written. The cost per return didn't fall, and that's the actual pressure. Paid returns have spread from 23% to 35% of the UK's top 100 fashion retailers and moved upmarket into luxury — and not a single brand has reversed the decision since 2023, which tells you more than any survey. But the policy question has stopped being binary: behaviour-triggered fees, membership exemptions and method-tiered pricing all recover cost while preserving more of the experience. And a fee recovers cost without reducing returns, because returns are a fit-information problem and fees are charged after the information was needed.

The uncomfortable part for marketers: over half of this is caused by things on your product pages, and almost none of the industry conversation is aimed at the people who own them.