How Refunds Affect Shopify Profit
A refund removes the revenue but not the costs. Outbound shipping is already paid, payment fees are frequently not returned, packaging is consumed, and the item comes back needing inspection and often a markdown. A fully refunded order typically costs more than an order that was never placed at all.
Written by Atul Tirkey · Co-founder, NetNet
Updated September 6, 2026 · 4 min read
The intuitive model of a refund is that it undoes the sale: money out, item back, net zero. Almost none of that is true.
Revenue reverses cleanly. Costs do not. The parcel has been shipped, the packaging is used, the fee has been taken, and the item is now somewhere in the postal system on its way back to be inspected. A returned order is not a cancelled order — it is an order that cost you everything a normal order costs, plus a second set of logistics, and produced no revenue.
What a refund does and does not reverse
Reversed: the revenue, and in most cases the cost of goods, since a resellable item returns to stock.
Not reversed: the outbound shipping label. It was purchased, the carrier moved the parcel, and no part of that comes back.
Not reversed: packaging. The box, filler and insert are consumed whether the customer keeps the item or not.
Usually not reversed: payment processing fees. Practice varies by gateway and market — some return the percentage, some keep the fixed component, some keep everything. Worth checking your own payout records rather than assuming.
Newly added: return freight, inspection labour, restocking, and any markdown required to sell a no-longer-new item.
The asymmetry is the whole story. Revenue is one line and reverses in full. Costs are seven lines and reverse partially at best.
Why partial refunds are proportionally worse
A goodwill discount after delivery — twenty percent off because something arrived scuffed — feels cheaper than a full return. Per dollar of revenue given back, it is worse.
The order value drops by twenty percent, but every cost stays exactly where it was. Nothing ships back, so nothing is recovered. Against a forty percent contribution margin, giving away twenty percent of order value removes half the profit on that order.
The same arithmetic applies to post-purchase discount codes and to shipping refunded as an apology. These are usually authorised by support staff working from a policy that names a percentage of order value, without anyone having established what percentage of margin that represents. Twenty percent of revenue and fifty percent of profit are the same decision described two ways, and only one of the descriptions makes it feel expensive.
Where the cost belongs
A refund issued in April against a February order can be recorded in either month, and the choice determines whether product-level profitability means anything.
Booking it to April is easier. The books balance and no closed period reopens. But the cost is now attached to the wrong month, the wrong product and the wrong channel — so the product that generates a high return rate looks identical to one that never comes back, and the campaign that acquired returning-prone customers looks as efficient as the one that did not.
Attributing back to the original order is the version that survives contact with a real question. It means February’s numbers change after February has closed, which is uncomfortable and correct. Every meaningful analysis of returns depends on the link between the refund and the sale holding.
Reading the return rate properly
Return rate as a percentage of orders is the standard metric and the least useful cut of it.
Three cuts are worth more:
By product. Return rates concentrate. Apparel sizing, anything with fit or colour expectations, anything fragile. A store-level rate of eight percent frequently hides one product at thirty.
By reason code. Damaged in transit points at packaging. Not as described points at product photography or copy. Wrong size points at a sizing guide. Each has a different and inexpensive fix.
By acquisition channel. Traffic bought on aggressive discounting returns at higher rates than traffic that arrived through search. If channel profitability is calculated without return costs attributed, the worst channel can appear to be the best.
Return policy as a pricing decision
Extending a return window is usually discussed as a customer experience decision. It is also a pricing one, and the two get separated far too often.
A longer window raises conversion — the reassurance is real and measurable. It also raises the return rate, lengthens the period before an order can be considered final, and increases the share of returned items that come back out of season or no longer sellable as new. A ninety-day window on apparel means accepting items that return after the range they belong to has been marked down.
None of that argues for a short policy. It argues for pricing the policy. If a generous window lifts conversion by a point and return rate by three, the arithmetic is knowable: compare the additional contribution margin from the extra orders against the additional cost of the extra returns, using the per-return figure from the example above rather than the refunded amount.
Stores that run this comparison sometimes extend the window further, having discovered it pays for itself. The failure mode is not choosing the wrong policy — it is never converting the policy into a number at all.
The number that matters
The figure worth tracking is not the return rate. It is contribution margin after returns, for the period.
Take the margin the shipped orders earned, subtract the full cost of the returns against them, and express what remains as a percentage of net sales. That single number captures both how often customers return and how expensive each return is, and it moves when either does.
A store with a twelve percent return rate on cheap, light, resellable items may be in better shape than one at six percent on heavy items that come back damaged. Only the margin-after-returns figure distinguishes them, and it is the one that determines what you can afford to pay for the next customer.
Costing a refund properly
Work through what stays spent after the customer is made whole, rather than recording only the amount returned to them.
About 20 minutes the first time
- 01
Start from the original order's contribution margin
The order earned a margin when it shipped. A refund does not simply cancel that figure, because several of the costs behind it cannot be recovered.
- 02
Reverse the revenue, in full or in part
Take out what was actually returned to the customer, which for a partial refund is less than the order value but still removes margin at full rate.
- 03
Keep every cost that stays spent
Outbound shipping, packaging, pick-and-pack and in many cases the payment processing fee remain paid whatever the customer gets back.
- 04
Add the cost of the return leg
Return freight, inspection labour and restocking are new costs that would not exist had the order never been placed at all.
- 05
Write the item down if it cannot be sold as new
Opened, used or seasonal stock rarely resells at full price. The markdown is a real cost and belongs against the return that caused it.
- 06
Attribute the whole result back to the original order
Book it against the month the order shipped, not the month the refund was issued, or per-product profitability will be permanently wrong.
What one fully refunded order costs
An $86 order, refunded in full three weeks after delivery, with the item returned opened and resold later at a discount.
| Line | Relative size | Amount |
|---|---|---|
| Revenue retained after full refund | $0.00 | |
| Outbound shipping label Already paid, not recoverable | $7.90 | |
| Return shipping label | $8.40 | |
| Payment processing fee Frequently not returned on refund | $2.79 | |
| Packaging consumed | $1.05 | |
| Inspection and restocking labour | $2.50 | |
| Markdown to resell the opened item Margin lost reselling it as opened stock | $9.30 | |
| Net result of the returned order | −$31.94 |
The customer was made whole and the business is $31.94 worse off than if the order had never existed. At a 40% contribution margin, roughly one additional order must be sold and kept just to return to where the store started before this one arrived.
Where the numbers come from
Every figure above traces to a specific field in a specific system. These are the ones that matter, and where each one goes wrong.
| Figure | Source | Where it breaks |
|---|---|---|
| Refund amount and date | Shopify refund records against the original order | Refunds are dated when issued, so they land in a later period than the sale they reverse. |
| Return freight | Carrier invoice for the inbound leg | The return leg is often billed under a different service code than the outbound parcel. |
| Payment fee treatment | Gateway payout records for the refund transaction | Whether the original fee is returned varies by gateway and by market, so it cannot be assumed. |
What this does not tell you
- This costs a single return in isolation. It does not capture the customer-lifetime effect, where a well-handled return can produce a repeat buyer and a badly handled one ends the relationship entirely.
- Markdown severity is an estimate until the item actually resells, so any return cost calculated at the moment of refund carries an assumption that may prove optimistic.
Frequently asked questions
Do refunds reduce revenue or increase costs?
Both. The refund removes revenue, and the return adds costs that did not previously exist — inbound freight, inspection, restocking and usually a markdown. Recording only the revenue reversal understates what a return costs by a substantial margin.
Are payment processing fees returned when I refund an order?
Often not. Practice varies by gateway and by market, and some return the percentage but keep the fixed component. Check your own payout records rather than assuming, because at scale the difference is material.
Should a refund be booked to the month it was issued?
Not if you want product-level profitability to mean anything. Attribute it back to the month of the original order, so the cost sits against the sale and the product that caused it rather than against an unrelated later period.
How does a partial refund differ from a full one?
Proportionally it is worse. A twenty percent goodwill refund removes twenty percent of order value from a contribution margin that might be forty percent, so it consumes half the profit while none of the shipping or fulfilment cost comes back.
Keep reading — Refunds, discounts & fees
True profit on Shopify
Costs that land after a period closes.
Hidden costs reducing Shopify profit
The small lines that compound.