A return costs you four things and the refund covers only one of them. You hand back the margin, you pay for the return label and the handling, you never get back the outbound shipping and pick and pack of a parcel that earned nothing, and the payment provider keeps its fee. On the example store below, an average refund of €92,39 costs €58,25 of profit, which is why a 5% return rate takes considerably more than 5% out of your contribution margin.
The refund is the smallest part of it
Most return calculators compute the refund plus a flat handling estimate, and land at a number that feels reassuringly small. The reason it is wrong is that the refund gives back revenue, and revenue was never profit. What you actually lose is the margin that revenue was carrying, plus three costs that no refund touches.
Here is the arithmetic on the example store used throughout this site, an apparel catalogue doing about €10,6M a year, over fourteen months to 20 Aug 2026. It processed 7.907 returns and refunded €730.522,43, an average of €92,39 each.
| Part | Total over 14 months | Per return |
|---|---|---|
| Margin given back | €352.848,73 | €44,62 |
| Return processing: label and handling | €47.507,15 | €6,01 |
| Stranded outbound shipping and packing | €43.342,64 | €5,48 |
| Stranded payment fees | €16.919,46 | €2,14 |
| Total cost to profit | €460.617,98 | €58,25 |
€460.617,98 against €15.194.923,20 of gross revenue is €3,03 of profit lost for every €100 the store sold, from a headline return rate of 4,81%. The gap between those two percentages is the whole point of this page.
Stranded costs are the part everyone misses
The outbound parcel happened. The carrier moved it, the picker picked it, the box was consumed, and none of that is undone by the customer changing their mind. Crediting outbound shipping back on a return would be telling yourself the parcel never went, and it makes your fulfilment costs look better than your invoices.
Payment fees behave the same way. The provider charged its percentage and its fixed fee on the original transaction and it keeps them. If you want to see what that costs across a real payment mix, the Shopify payment fees calculator does the blended version.
Then there is the part that is not a cost at all but decides how much of the goods you get back: resale rate. A returned item that goes back on the shelf at full price returns its unit cost to inventory. One that is discounted, or written off, does not. In apparel a realistic resale rate is well below 100%, and the difference goes straight through CM1.
Which day each part belongs to
This is the detail that decides whether the number is readable at all. The refund, the cost of goods credited back and the return processing belong to the day the refund was issued. The outbound shipping and the payment fee stayed on the order date, where they were incurred, and never move.
Book returns back against the original order instead and last month keeps improving every time this month gets worse. A closed month never closes, every report you ran changes after you read it, and the team stops believing the numbers. There is one exception worth knowing: the cost of goods credited back is valued at the cost rule that was effective on the original order date, so the credit exactly reverses the original debit and a supplier price change between order and refund can never invent a gain.
Goodwill refunds and chargebacks are different
A goodwill refund or a cancellation, where no parcel comes back, carries no return processing cost. There is no label to buy and nothing to handle. Charging one anyway overstates your logistics costs by however many of these you issue, which in a store with good customer service is not a small number.
A chargeback is not a return either. It is a disputed payment, so nothing physical comes back: the amount belongs with returns, the chargeback fee belongs with transaction costs, and no return processing applies. Treating the two the same overstates logistics and understates fees at once.
What to do with the number
Read it per product rather than in total. One badly sized item can carry a whole category, and a total return rate that looks acceptable often hides two products that should be re-photographed, re-described or dropped. The definitional version of all this is in true cost of a return, and the measurement conventions are in ecommerce return rate.
Then run the same arithmetic on a free shipping threshold or a free returns policy before you launch it, because both are usually decided on their effect on conversion and paid for out of CM2. If you would rather do it in a spreadsheet, the free P&L templates include the cost columns this page uses.