eCommerce
The real cost of a return: the per-SKU profit math
A single non-resalable $60 return costs about $26 in hard cash beyond the reversed sale: the $18 COGS write-off plus the $8 outbound label you already paid. Add return shipping and reprocessing and it runs closer to $37. A 15% return rate on 1,000 units is 150 returns, about $3,900 of hard cost on that one SKU, invisible on your P&L.
Key Takeaways
- A return is not a reversed sale. It is a stack of cash you already spent. The COGS write-off, the outbound label, the inbound label, and the reprocessing labor are all gone whether or not the unit ever resells. Optoro data pegs the processing cost alone at roughly 66% of the product's price.
- On a $60 item, one non-resalable return costs about $26 in hard cost beyond the lost sale. That is the $18 COGS write-off plus the $8 outbound label. Count return shipping and reprocessing and it climbs to about $37.
- Most returns do not go cleanly back to full-price stock. Only about 48% of returned items are resold at full price, so the 100% COGS write-off is the base case for a lot of SKUs, not the exception.
- A blended return rate is an average that hides the bleeding SKU. Apparel runs 20-40% online while beauty runs 4-12%, so one company-wide number buries a margin-destroying product inside a healthy one.
- Rank every SKU by return-adjusted contribution, not gross sales. A high-return product can post strong gross contribution and still fall near zero once its return rate and per-return cost are subtracted. That ranking is the reprice, restrict, or cut decision.
When a DTC founder tells me their return rate, it is almost always one number. "We run about 15%." They say it the way you would quote a shipping cost, a single blended figure treated as a marketing or customer-experience metric. The problem is that a return rate is not a marketing metric. It is a per-SKU margin event, and the blended number is an average that quietly buries your worst-performing product inside your best one. This post gives you the exact per-SKU cost-of-a-return formula, a worked $60 example, and the ranking method, return-adjusted contribution, that surfaces the SKUs destroying margin despite strong gross sales. It is scoped to US DTC brands.
A return is not a reversed sale. It is a stack of costs you already spent.
Start with the reframe, because everything else follows from it. When a customer returns a product, the visible half of what happens is obvious: you reverse the sale and refund the money. That is the part founders picture. The invisible half is the cash you already spent to make and ship that order, cash that does not come back with the box.
The scale of this is not small. US retail returns hit $849.9 billion in 2025, and the online return rate ran 19.3% of ecommerce sales, according to the National Retail Federation. Roughly one in five online orders comes back. So a "15% return rate" is not unusual. What is unusual is how few brands have modeled what those returns actually cost.
The single most useful number I can give you comes from reverse-logistics firm Optoro, cited by Coresight Research: it typically costs a company about 66% of a product's price just to process a return. Read that again. Before you even count the reversed sale, you spend roughly two-thirds of the price handling the thing coming back. A return is not a sale worth zero. It is a sale worth negative.
When I dig into a brand's return pile with them, the recurring shock is how much of it never goes back on the shelf. One brand we worked through had 371 returned units in a year, and nearly two-thirds of them never resold. That is not a rounding error on the P&L. That is a warehouse full of cash they had already spent and would never recover.
The per-SKU cost-of-a-return formula
Here is the formula that turns "we run about 15%" into a number you can act on.
Cost of a return = COGS write-off + outbound shipping (already paid) + inbound return shipping + reprocessing labor + refund processing, minus salvage value recovered
Most of those line items are fixed the moment the order shipped. The outbound label is paid. The pick-and-pack labor is spent. The one variable that swings the whole calculation is the COGS write-off, and it is decided by disposition, meaning what condition the unit comes back in and whether you can sell it again.
There are two cases. A non-resalable return, one that comes back damaged, opened, used, out of season, or hygiene-restricted, is a 100% COGS write-off. The unit gets liquidated for pennies or scrapped. A restockable return, one that comes back like-new and loops quickly back into primary inventory, only costs you the reprocessing, roughly 15-25% of COGS in inspect, repackage, and relabel labor, because you recover the unit itself.
The trap is assuming most returns are restockable. They are not. Only about 48% of returned items are resold at full price, per ShipNetwork's 2026 data. For a lot of SKUs, the non-resalable, full-write-off case is the base case, not the exception. On the clean end, a resalable return is close to a wash: if something costs you $40 to make and it comes back sellable, you reverse the sale, put the $40 back into inventory, and it is roughly a net-zero COGS adjustment. A non-resalable return has no such recovery. The $40 is simply gone.
Returns are quietly eating your margin. See by how much.
Get our Real Cost of Returns calculator: plug in your numbers, see the true hit per return.
Check your inbox. We'll send the Real Cost of Returns calculator shortly.
The worked example: what a $60 return really costs
Take a concrete SKU. A $60 product, $18 COGS, $8 outbound shipping already paid. It comes back non-resalable. Here is the hard cash cost, before we even talk about the refund.
| Cost component | Amount | Note |
|---|---|---|
| COGS write-off (non-resalable) | $18 | 100%, unit cannot be resold |
| Outbound shipping (already paid) | $8 | sunk the moment it shipped |
| Inbound return shipping | ~$8 | reverse label on brand-paid free returns |
| Reprocessing / inspection | ~$3 | receive, inspect, dispose |
| Hard cost of the return | ~$37 | before the refund |
| Refund issued to customer | $60 | revenue reversed, shown separately |
There are two honest ways to headline this. The tight, defensible number is $26: the $18 COGS write-off plus the $8 outbound label, the irrecoverable cash you already spent. Load in the reverse shipping and reprocessing and it is closer to $37. Either way, the framing that matters for a founder is this: you lose the $60 sale, and you are out another $26 to $37 in cash on top of it. The refund is the part everyone sees. The $26-to-$37 is the part nobody books to a line.
Now scale it to one SKU. At a 15% return rate on 1,000 units, that is 150 returns. At the conservative $26 of hard cost each, that is about $3,900 of cost that never appears as a line item on your blended P&L, on this one product. It hides inside "returns and allowances" and a gross margin that reads a point or two lighter than you expected. If most of those returns are non-resalable and you load the full reverse stack at $37, it is closer to $5,550. The $3,900 is the number I would plan against per SKU, because it is the one you can defend to the penny, and then it compounds: run that same math across every SKU in a catalog and the invisible cost climbs into the tens of thousands fast.
The reason this matters so much is the contribution math underneath it. The typical ecommerce company has to bring in four to five dollars of revenue to cover every dollar of fixed cost, and we typically see operators hold a contribution-margin floor around 20% so a sale is worth making at all. A return that eats 66% of the price does not just zero out that order. It drags the SKU below the floor and starts pulling down the ones that stuck.
Why your blended return rate is lying to you
Here is where the single number does the most damage. Return rates are not uniform across a catalog. They are wildly category-skewed.
| Category | Online return rate (range) | Notes |
|---|---|---|
| Apparel & fashion | 20-40% | Highest, sizing and fit driven |
| Footwear | 17-35% | Fit plus multi-size ordering |
| Home goods & furniture | 15-23% | Lower frequency, high per-unit cost |
| Electronics | 8-11% | Low frequency, $30-$65 to process |
| Beauty & personal care | 4-12% | Among the lowest, hygiene-restricted |
Online apparel runs 20-40% while beauty runs 4-12%. Coresight found apparel specifically returning at 24.4% online, nearly eight points above the overall online rate, and estimated the category spends about $25.1 billion a year just processing those returns. If your catalog spans both an apparel line and a beauty line, a blended 15% is the mathematical average of a product bleeding 30% and one barely returning at 6%. It describes neither. It is the temperature of a patient with one foot in ice and one in boiling water.
For planning, the DTC brands we work with tend to sit in tighter bands than the broad ecommerce averages, because they are single-channel: apparel roughly 15-25%, beauty 8-12%, home goods 10-18%. Those bands are drawn from our anonymized client panel and are a planning benchmark, not a published figure, so treat them as a starting estimate you replace with your own SKU data as soon as you have it.
The reason this stays hidden is accounting mechanics. Returns get netted into a blended revenue line, and net revenue comes out the same or slightly below your gross DTC sales. Founders ask me all the time why a single bad return day, say a $10,000 reversal, makes the day look terrible, and the answer is that the P&L never separates the per-SKU damage in the first place. It all lands in one averaged line.
Rank your SKUs by return-adjusted contribution
This is the payoff. The metric that surfaces the margin killer is return-adjusted contribution per unit, and the formula is:
Return-adjusted CM per unit = gross CM per unit times (1 minus the return rate), minus the cost of a return times the return rate
In plain terms: for every 100 units you sell of a SKU, the ones that stick earn gross contribution, and the ones that come back cost you the full return-cost stack. Rank every SKU by this number and the products that look fine on gross sales but sit near or below zero jump out.
Watch what it does to four SKUs that all look identical on gross contribution. Each is a $60 product with $18 COGS and $34 gross contribution per unit before returns. The only thing that differs is the return rate, and I am using a flat $26 hard cost per return so the math is clean and consistent.
| SKU | Return rate | Gross CM per unit | Return-adjusted CM per unit |
|---|---|---|---|
| SKU A (apparel, high-return) | 30% | $34.00 | $16.00 |
| SKU B (apparel, mid-return) | 18% | $34.00 | $23.20 |
| SKU C (home goods) | 12% | $34.00 | $26.80 |
| SKU D (beauty, low-return) | 8% | $34.00 | $29.20 |
All four post the same $34 gross contribution. On a gross-sales report they are indistinguishable, and the high-volume apparel SKU probably looks like a hero. But SKU A keeps less than half its contribution once returns are priced in, $16 against $34, while the beauty SKU barely moves, from $34 to $29.20. Push SKU A's hard cost up to the full $37 stack and it slides toward zero. Same gross margin, wildly different real economics.
This is the decision the whole exercise drives toward. When we run this for a brand, the conversation shifts fast. On one call the merchandiser looked at 100 styles and realized half were not really contributing, and that the catalog could shed a few hundred thousand dollars of inventory without losing meaningful margin. Return-adjusted contribution is how you decide which half. It is not a reason to cut blindly. It is the ranking that tells you where to reprice, where to tighten sizing and product descriptions, where to add a restocking fee, and where to simply stop reordering.
A 15% return rate is not a 15% problem. It is a per-SKU margin event that a blended number is built to hide. Model what one return actually costs, rank your catalog by return-adjusted contribution, and act on the bottom of that list. The SKU that looks like a winner on gross sales is often the one quietly funding its own losses.
What to do this week
You do not need a data warehouse to start. You need disposition data and one spreadsheet.
First, pull the return rate for your top 20 SKUs by volume, not the blended company number. Second, get the disposition split for each: what share of that SKU's returns come back resalable versus non-resalable. Your 3PL or returns platform has this, and if it does not, a week of manually tagging received returns will get you a usable estimate. Third, compute the cost of a return for each SKU using the formula above, then compute return-adjusted contribution per unit. Rank the list.
Act on the bottom. For the SKUs sitting near or below your contribution floor, the levers are reprice to rebuild margin, tighten the sizing chart and product photography to cut the return rate at the source, add a restocking fee, or cut the SKU. One caution on restocking fees and paid returns from the brands we have watched try it: it works, it does bring returns down, but it can also cost you real demand, especially in women's apparel where free returns are close to table stakes. Treat a paid-return policy as one lever with a demand cost attached, not a free win. The point is not to punish returns. It is to know, per SKU, what every unit that sticks has to earn back, and to stop reordering the products that cannot clear the bar.
Related reading. For the metric itself, see what ecommerce return rate is, and for what a returned unit really costs, see the true cost of apparel returns. For how we find the SKUs where returns quietly erase margin, see our fractional CFO work.
Related reading. For why the same SKU can be a top seller and your worst margin line, see our bestseller-SKU profitability breakdown.
Sources and methodology
US returns totals and the online return rate come from the National Retail Federation. The $849.9 billion in 2025 US retail returns and the 19.3% online return rate are from the NRF's 2025 returns data. See the NRF press release.
The processing-cost and apparel figures come from Coresight Research and Optoro. The "about 66% of a product's price to process a return" figure is attributed to Optoro, and the 24.4% online apparel return rate plus the roughly $25.1 billion annual apparel processing cost are Coresight's, over the 12 months ended March 2023. See Coresight Research: The True Cost of Apparel Returns.
Resale rates, per-item processing costs, and category ranges come from ShipNetwork and Richpanel. The "only about 48% of returned items are resold at full price" figure and the $10-$20 per-item processing cost ($30-$65 for electronics) are from ShipNetwork. The category return-rate ranges are corroborated by Richpanel, both derived from NRF, Statista, and Appriss Retail data. Category ranges differ by source and by whether they measure share of purchases returned or merchant-side financial return rate, so they are presented as ranges.
The per-SKU cost stack and DTC panel bands are planning models, not published data. The $60 worked example, the return-cost stack, and the return-adjusted contribution ranking are illustrative CFO models. The tighter DTC category bands (apparel 15-25%, beauty 8-12%, home goods 10-18%) are drawn from Eightx's anonymized DTC client panel and are a planning benchmark, not a published figure; they sit below the broad ecommerce averages because they reflect the DTC-only channel. Eightx's published benchmark for overall ecommerce is available at Average Ecommerce Return Rate. This post is general information, not accounting advice; confirm your own numbers against your returns and inventory data.
Frequently asked questions
how much does a return actually cost my brand?
More than the reversed sale. On a $60 product with $18 COGS and $8 outbound shipping, one non-resalable return costs about $26 in hard cash beyond the lost revenue, and closer to $37 once you add inbound shipping and reprocessing. Industry data from Optoro puts the processing cost alone at roughly 66% of the product's price.
how do i calculate the cost of a return per sku?
Add the COGS write-off, the outbound shipping you already paid, the inbound return label, and the reprocessing labor, then subtract any salvage value if the unit resells. The COGS write-off is the swing variable: a non-resalable unit is a 100% write-off, while a like-new restockable unit is only the reprocessing cost, roughly 15-25% of COGS.
is a 15% return rate bad?
A 15% blended rate is roughly normal for DTC, but the rate is not the point. The cost is. A blended number averages a high-return apparel SKU against a low-return one and hides both. What matters is the return rate on each SKU and the hard cost per return, because that is what decides whether a product still makes money.
what return rate is normal for apparel vs beauty?
They are worlds apart. Online apparel runs roughly 20-40% and footwear 17-35%, driven by fit and multi-size ordering, while beauty and personal care run about 4-12% because hygiene rules limit returns. That spread is exactly why a single blended rate misleads you.
what percent of returned items can be resold at full price?
Only about 48%, per ShipNetwork's 2026 data. That means the majority of returns trigger a partial or full COGS write-off rather than going cleanly back to full-price stock. It is why the 100% write-off case is the base case for many SKUs, not the exception.
what is return-adjusted contribution margin?
It is the contribution a SKU actually earns once returns are priced in. For every 100 units you sell, the ones that stick earn gross contribution and the ones that come back cost you the full return-cost stack. The formula is gross CM per unit times (1 minus the return rate) minus the cost of a return times the return rate. Rank your catalog by this number and the margin killers surface.
does a high-gross-margin product still lose money if it gets returned a lot?
Yes. A product can post a strong gross contribution per unit and still fall to near zero on return-adjusted contribution once a high return rate and the per-return cost stack are subtracted. Gross margin tells you what a sale is worth if it sticks. Return-adjusted contribution tells you what it is worth on average, which is the number you should reorder against.
why does my p&l not show what returns are costing me?
Because returns get netted into a blended revenue line and a slightly lower gross margin, so the per-SKU damage disappears into an average. The COGS write-offs, reverse shipping, and reprocessing labor scatter across several expense lines. Nothing on the P&L says "this SKU lost $9,000 to returns this quarter," which is why you have to model it at the SKU level yourself.