How to Handle Ecommerce Returns

US consumers sent back $849.9 billion in merchandise during 2025, or 15.8% of retail sales, per the National Retail Federation and Happy Returns. Online, it’s 19.3%. Sell apparel and it’s higher still.

Most advice on this starts when the box hits your dock, which is backwards. The value you’re trying to recover starts draining the moment a customer clicks the return button, and the decisions that protect it happen well before anyone touches the item.

What does one return really cost you?

You’d be wise to run this number before you touch anything else. It’s hard to judge a policy change, a software subscription or a 3PL quote until you know what you’re already spending.

Reverse freight runs $8 to $12 on a typical parcel. Restocking labor adds $5 to $8. Holding depreciation takes $2 to $4. Customer service handling costs $2 to $5. All in, that’s $17 to $29 per return before a single markdown, based on 2026 figures from WarehousingCosts.com.

Outsourced processing rests inside that stack. Standard returns processing at a 3PL runs $3.50 to $6.00 per unit, with inspection and grading adding $2 to $5. Electronics needing functional testing run $4 to $12.

The routing penalty is real. McKinsey found that pushing returns through the wrong channel adds $5 to $6 per item in avoidable expense.

Now hold that against your average order value. A $28 cost on a $45 AOV is a different business than $28 on a $200 AOV. That ratio, not your return rate, tells you how much this deserves your attention.

Stage	Cost	Share of total
Return label and reverse freight	$3.00 to $8.00	35% to 50%
Receiving and intake	$0.75 to $1.50	8% to 12%
Quality inspection	$1.00 to $3.00	10% to 20%
Restocking and putaway	$0.75 to $2.00	8% to 15%
Refund processing	$0.50 to $1.50	5% to 10%
Disposal or liquidation	$0.50 to $2.00	5% to 15%

Your category sets the baseline. Apparel runs 20% to 40%. Footwear lands at 17% to 30%. Auto parts sit near 19%, home and furniture at 15% to 23%, accessories and jewelry at 12% to 15%, electronics at 8% to 15%, beauty at 4% to 12%.

Why is speed the part everyone underprices?

A returned unit is a depreciating asset. The clock starts the second a customer decides to send it back.

Researchers at Vanderbilt, Indiana University and INSEAD measured it. Studying returned Hewlett-Packard inkjet printers and Bosch power tools, they found printers lost about 1% of value every week and power tools about 1% a month. On short-lifecycle goods, delay through a slow reverse chain destroyed more than 30% of product value. The paper is older, published in Management Science, but the physics haven’t changed.

What has changed is how much slower most operations run than they realize. Corso’s 2026 benchmarks put the average return at nine to ten days from initiation to refund. Best-in-class clears it in three to five. A few do it inside 48 hours.

Picture a jacket returned in early November. Back on the shelf in four days, it sells at full price during peak. Cleared in mid-December, it misses the window and becomes a January markdown. Same unit, same condition, most of the margin gone.

Speed buys goodwill too. Most customers expect a refund within a week of shipping the item back, and 52% will file a chargeback when a merchant is slow. Every day you add costs you twice.

Stage	Best-in-class	Average
Request to label generated	Same day	Up to 24 hours
Label to warehouse receipt	3 to 5 business days	5 to 7 business days
Receipt to inspection complete	1 to 2 days	3 to 5 days
Inspection to resolution issued	Same day, automated	1 to 2 days, manual
Total cycle	3 to 5 days	9 to 10 days

4 decisions that set your recovery rate

Every returns operation makes the same four calls on every item. Most make them one box at a time, at the inspection bench. The ones recovering the most value make them in advance, as rules, before the item is anywhere near the building.

Refund now, or after inspection? Bring the item back, or leave it with the customer? Restock it, or route it elsewhere? Ship the replacement now, or wait?

McKinsey’s February 2026 research put average value recovery around 50% across re-commerce channels, rising to roughly 75% when goods get routed well and moved in season. That 25-point spread is the whole argument for deciding in advance.

Should you refund now or after inspection?

No single policy resolves this cleanly.

Customers want the money immediately. NRF found 76% are more likely to choose a retailer offering an instant refund or exchange. Waiting for a box to land, get scanned, inspected and approved is a week of their money sitting in your account, and they notice.

But paying before you verify is the hole fraud walks through. NRF puts fraudulent returns at 9% of all returns. Appriss Retail, working with Deloitte, puts fraud and abuse across returns and claims at 15%, worth more than $103 billion a year. Those numbers aren’t in conflict. NRF counts outright fraud on physical returns. Appriss counts fraud plus abuse, and folds in item-not-received claims. Know which one you’re quoting.

Loop’s analysis of 23.4 million returns found 11.4% of return value flagged high risk, with an average flagged return of $120.

So the answer is a threshold, not a policy. Instant refunds for known customers under a value ceiling. Inspection first above it, or on any account with flagged history. That rule lives in your system and fires before the label is even generated.

Should the item come back at all?

Sometimes the cheapest thing you can do with a return is nothing.

When freight plus processing runs $17 to $29 and the item’s recoverable value sits below that, shipping it back spends money to destroy money. Low-value consumables, opened personal care, heavy goods on thin margins: the arithmetic frequently says leave it.

Returnless refunds went from an Amazon quirk to standard practice, and the logic is purely financial. Refund the customer, skip the freight, skip the intake labor, skip the disposition call. You’ve already lost the sale. Now you’re choosing whether to lose the shipping too.

Two guardrails. Keep the threshold account-level rather than published, because a discoverable rule gets gamed within weeks. And pair it with donate-in-place or local liquidation where the item has real value to somebody, just not enough to justify a trip back to your DC.

Where does the unit go once it lands?

Five roads out of receiving: restock as new, refurbish or repackage, grade down to B-stock and liquidate, donate, or recycle and dispose. Most operations default to the first and the last, which is how warehouses fill up with items nobody has priced.

Standardized grading turns this from a judgment call into a repeatable decision. Grade A goes back to sellable stock. Grade B goes to refurbishment or a secondary channel. Grade C goes to liquidation, donation or recycling. Written thresholds mean the same item gets the same treatment whether your most experienced tech handles it or a temp hired for January.

Price each path honestly. Refurbishment and repackaging cost $3 to $10 per unit. Functional testing on electronics runs $4 to $12. Disposal and recycling cost $0.50 to $3. A $9 refurb on an item you can resell for $14 isn’t a recovery. It’s a slower write-off.

Timing sits underneath all of it. The same graded unit is worth more in October than in February, which loops right back to cycle time. Fast disposition isn’t a warehouse efficiency metric. It’s a margin lever.

Do you send the replacement now or later?

A return that ends in an exchange keeps your revenue. A return that ends in a refund gives it back.

That one line should reshape how you measure this. Return rate is set mostly by category and customer behavior, and you barely control it. Refund rate responds to how you design the flow, and it’s the number sitting between your returns operation and your P&L.

The market has moved already. Loop’s data covering November 2024 through October 2025 shows 73.6% of merchants offering exchanges, and 51.7% of those running a shop-now flow add bonus credit averaging $11.28. Spending eleven dollars to hold a hundred-dollar sale is a trade most finance teams take every time.

Advance exchanges go further. Ship the replacement when the return is initiated rather than when it arrives, and a refund event becomes a fulfillment event. It carries risk on customers you don’t know, which puts it in the same tiered-trust framework as instant refunds.

One warning. An exchange that crosses a border isn’t one transaction. It’s two customs events, and the duty exposure doubles.

What happens when a return crosses a border?

With de minimis gone, every cross-border return is a customs event. Almost no fulfillment setup is built for that, and the money involved isn’t small.

Three mechanisms are worth knowing. HTS subheading 9801.00.10 lets previously exported goods re-enter the US duty free within three years, provided they come back unchanged. Filing needs an importer’s declaration and a foreign shipper’s declaration on each shipment, plus documents tying the return to the original export. Duty drawback under 19 U.S.C. 1313 works the other direction, recovering up to 99% of duty paid on imported goods later exported. And in the EU, the €150 duty threshold is gone, replaced by a flat €3 per declaration line on low-value imports.

Here’s where merchants get stuck. In the UK, EU and Canada, the consumer is frequently the named importer of record on a low-value shipment. That gives them legal standing to claim a refund and no practical capacity to file one. The party with the capacity, your logistics provider, has no standing. Nothing gets claimed, and the duty stays paid. Non-resident importer and importer-of-record structures close that loop, but they get set up before the first order ships, not after the first return.

Recovery also depends on data captured months earlier, at clearance. A fulfillment partner who doesn’t record and share it makes your claim impossible, and most merchants find that out only when they try to file.

Which returns are telling you something?

Reason codes are the most underused dataset in ecommerce, mostly because merchants collect them and never route them to anyone who can act.

Shorr Packaging surveyed 2,013 US consumers in November 2025. Wrong size came in at 44%, damaged goods at 31%, description mismatch at 11%, change of mind at 9%.

Read that again, because only one of the four is a customer problem.

Wrong size is a merchandising and product-page issue: fit guides, real measurements, sizing data pulled from reviews. Damaged goods at 31% is a packaging and pick-pack issue inside your own four walls, meaning a third of your returns are self-inflicted and fixable without touching your policy. Description mismatch belongs to whoever writes your PDPs. Change of mind, at 9%, is the only slice that’s really the customer’s.

Sort returns into four buckets, product, content, fulfillment and customer, and each gets a different owner and a different fix. That’s a very different meeting than “our return rate went up.”

Bracketing earns its own line. McKinsey found roughly one in four transactions now includes bracketed items, where the customer orders multiple sizes fully intending to send most back. You can’t policy your way out of that. You can fit-guide your way out of a good chunk of it.

How much should you charge for a return?

Free returns are no longer the default, and the shift happened fast.

Loop’s merchant data shows 65.2% now charge a return fee, averaging $9.04. Refund windows average 39 days, exchange windows 41. NRF found merchants pointing to operations cost (40%), carrier shipping cost (40%) and tariff or economic uncertainty (33%).

Consumers hate it, predictably. NRF found 82% still cite free returns as a major purchase consideration, up from 76% the year before. Shorr put 86% wanting free return shipping and 51% rejecting restocking fees outright.

So how are both true at once?

Customers dislike uniform friction more than they dislike differentiated policy. McKinsey found 71% said return policies that differ by product and by customer wouldn’t make them shop less. The problem isn’t charging. It’s charging everyone the same way regardless of what they bought or how they’ve behaved.

Four structures worth testing. A flat fee off the refund, simplest and bluntest. Free when they exchange, paid when they refund, which prices the outcome you want. A prepaid returns option at checkout, which turns the cost into a product. Or tiered treatment, where repeat customers go free, first-timers pay standard, and flagged accounts face tighter rules.

Windows need a decision too. Two-thirds of consumers expect 30 days. Shorter saves you little and reads as hostile. Longer, which 37% of merchants go over the holidays, costs almost nothing on items you’d take back anyway.

The returns you never see

Now the number that should worry you most.

That’s not a return you avoided. That’s a customer stuck with something they didn’t want, blaming you for it, who quietly never orders again.

NRF fills in the rest. 71% say they’re less likely to shop with a retailer again after a poor returns experience, up from 67% a year earlier. Four out of five tell friends and family.

So a falling return rate can be a warning light rather than a win. You can’t tell which without putting repeat purchase rate next to it. If returns dropped and reorders dropped with them, you didn’t fix anything. You made it hard enough that people gave up.

When does outsourcing returns pay for itself?

Returns volume doesn’t arrive evenly, and that’s usually what breaks an in-house operation.

Holiday returns spike to 44% or more of the annual total in January. The staffing, dock space and throughput you need for six weeks look nothing like what you need in April. Hire for the peak and you pay for idle capacity ten months a year. Staff for the average and you get a January backlog, which stretches cycle time exactly when unit values fall fastest.

The market has largely settled how to handle it. NRF found 49% of merchants lean on third-party logistics for holiday returns, 43% hire seasonal staff, and 37% extend windows to smooth the curve.

Where returns get processed drives cycle time too. Handling them in the same buildings that ship your outbound orders removes a transfer leg and gets graded units back into sellable inventory in days rather than weeks. Split them off to a separate facility and every handoff adds days you pay for in depreciation.

Speed Commerce processes returns inside the same North American network that ships your orders, including retail EDI returns and return-to-vendor handling for merchants selling through retail partners.

Where to start this quarter

Four moves, in order.

Measure your cost per return before you touch policy, software or partners. Pull reverse freight, intake labor, inspection time, refund processing and disposition write-offs into one number per unit. Nearly everyone doing this for the first time finds it higher than they guessed.

Time-stamp the cycle at every stage: request to label, label to receipt, receipt to inspection, inspection to resolution. The nine days will be concentrated in one or two places, and rarely the ones people assume.

Write your disposition thresholds down. Not as guidance. As rules with numbers in them, so the call stops depending on who’s working that shift.

Then go shopping for a partner. A returns proposal is impossible to judge until you know what your returns already cost, and that’s a conversation worth walking into with your own numbers.

Sources: National Retail Federation and Happy Returns, 2025 Retail Returns Landscape (October 2025); McKinsey & Company, From cost center to competitive advantage: Modernizing reverse logistics with AI (February 2026); Loop 2026 Global Ecommerce Retention Report; Appriss Retail with Deloitte; Shorr Packaging consumer returns survey (November 2025); Corso returns processing benchmarks (2026); WarehousingCosts.com (August 2026); Guide, Souza, Van Wassenhove and Blackburn, Time Value of Commercial Product Returns, Management Science; US Customs and Border Protection, HTS subheading 9801.