The Maze: A refund reverses a sale. It does not rewind the spending that produced it. In an illustrative €40 Amazon order, a sellable return leaves €10.70 of unrecovered costs; an unsellable one leaves €22.70. The example comes from Claudiu Clement, co-founder of Clarisix, an Amazon profitability analytics tool. These are assumed unit economics, not average seller losses. The useful distinction is between returning customer money, recovering inventory and absorbing expenses that have already happened.
Getting the product back does not get the selling costs back. The €10.70 example combines €1.20 of refund administration, €5.50 of fulfilment and €4 of advertising. Those amounts add up even when the returned unit can be sold again. Fulfilment by Amazon, or FBA, means Amazon handles storage, picking and delivery for a fee. Advertising bought the customer visit; a return cannot unbuy that visit. A refund rate therefore says how often revenue reverses, but leaves unanswered how much spending remains attached to each reversed order. The distinction matters even for stock that arrives back in perfect condition.
Inventory condition more than doubles the assumed loss. An unsellable unit adds €12 of landed cost of goods, meaning the product cost including getting it ready for sale. That raises the example from €10.70 to €22.70. The €12 difference comes entirely from the inventory outcome; the other three components stay unchanged. A single blended return metric hides this split. A business with many recoverable returns faces different economics from one receiving damaged stock. Recovery proceeds and reimbursements would change the result, so “unsellable” should trigger a recovery review rather than an automatic assumption that every euro is permanently gone.
A small refund rate can become a cost across every sale. At an assumed 5% refund rate, €10.70 multiplied by 5% is €0.535 per gross unit sold; €22.70 multiplied by 5% is €1.135. Rounded to cents, that is €0.54 or €1.14 before extra return costs or recoveries. These are endpoint scenarios: every return sellable versus every return unsellable. An actual seller sits between them only if the same input costs apply. The exercise turns an operational percentage into a pricing and margin question. It also shows why the condition mix deserves attention alongside the frequency of refunds.
Separate the analysis without counting the same expense twice. The customer receives €40 back, but the €10.70 and €22.70 amounts describe unrecovered spending, not an extra €40 refund charge. Fulfilment and advertising were already incurred. If the profit statement already includes them, adding them again as new return expenses would exaggerate the loss. Amazon UK's pricing page supports the administration-fee mechanism: the platform returns the referral fee less a capped administration charge. It does not validate Clement's euro inputs as universal rates. Category, fulfilment and market details still matter; removal, disposal, tax and reimbursements are outside this simplified example.
Why it matters: Returns deserve two questions: how often do they happen, and what remains unrecovered when they do? Split the answers by product and inventory outcome. Keep revenue reversal, retained fees, acquisition spending and stock recovery visible, then reconcile them to costs already booked. A lower refund rate helps. Turning unsellable stock into recoverable stock can also change the economics materially. The practical target is the contribution left after returns, rather than a reassuring percentage on its own.
