How to Measure the Cost of Returns in Fashion Ecommerce

Every example and amount on this page is illustrative. It shows the method and is not a benchmark, a quote or advice.

What Return Cost Actually Measures

The cost of returns in fashion ecommerce is the economic impact of an order being returned instead of completed and kept. It is broader than the return label or warehouse handling fee, but narrower than every cost of running the store. The starting point is realized revenue: a returned or cancelled order should not be treated as completed sales revenue. The return then creates costs or value loss that would not exist if the customer had kept the item.

Visible processing costs include the return label or carrier charge, warehouse receiving, inspection, repackaging, restocking, customer support and any payment or refund fee the merchant cannot recover. If outbound shipping was paid by the merchant and cannot be recovered after the refund, that unrecovered amount belongs here too. These costs are tied to the returned order, rather than allocated general ecommerce overhead.

The less visible component is lost merchandise value. A returned garment may come back sellable, but only after its original selling opportunity has been lost. It may later sell at a markdown, become damaged or unsellable, or remain in stock long enough to become dead stock. The economic cost is the reduction in value that the return creates, including any inventory value written off or tied up because of it. A returned item that can be resold at full price has a different cost from one that can only be sold at 40% off.

The Return Cost Formula, in Plain English

Use two measures, because a return creates both an operational expense and a lost-profit effect. Keep the denominator consistent: use returned orders when measuring order economics, or returned units when products in one order can be returned separately.

The basic formula is: cost per return = gross return expense ÷ number of returns. Gross return expense includes costs triggered by the return, such as the return label, reverse shipping, inspection, repackaging, return handling, non-refundable payment fees, and any markdown or write-off required because the item cannot be sold as new. Subtract any recovery you actually receive, such as a resale value or supplier credit, if you want a net operational cost rather than a gross cost.

A useful rate is: gross return cost rate = gross return expense ÷ realized revenue × 100. Use realized revenue, not booked order value: refunded and cancelled orders should not inflate the denominator. This makes the rate comparable with other profitability measures. A separate view can show contribution margin lost to returns ÷ realized revenue × 100. That is the broader economic rate.

Contribution margin lost is not the same as the return bill. Start with the contribution margin the order would have produced if the customer had kept it. Subtract the contribution margin actually retained after the refund, return-related costs, recovered product value, and any remaining inventory loss. The gap is the margin lost. An item returned in resalable condition may create a small margin loss plus handling costs; an item that becomes dead stock can destroy most or all of its original contribution.

• Gross return expense: the direct operational cost of processing the return.
• Contribution margin lost: the profit contribution that disappeared because the order was not kept.
• Return cost rate: report both where possible—gross expense as a percentage of realized revenue, and contribution margin lost as a percentage of realized revenue.

A Worked Example From Order to Refund

Consider a Shopify fashion merchant selling a jacket for $120. The merchant’s product cost is $36, and the original payment-processing fee is $3.60. The customer returns it, so the merchant’s realized revenue from this order is $0, not $120. Assume the $3.60 payment fee is not refunded, and the merchant pays $8 for reverse shipping, $4 for warehouse handling, and $2 to inspect, re-tag and restock the jacket.

The jacket is not necessarily a total loss. Inspection confirms it can be sold again, but only at a $84 markdown price. That $36 reduction from the original $120 price is the resale-value loss created by the return. The product’s $36 cost is recovered as inventory value when it returns to stock; it should not also be counted as a new return expense. If the jacket later sells for $84, the merchant recovers revenue, but not the original selling economics.

Here is the arithmetic for the return event, before any later resale payment fee or fulfilment cost:

• Refunded revenue: −$120
• Nonrefunded payment fee: −$3.60
• Reverse shipping: −$8
• Handling: −$4
• Restocking: −$2
• Resale markdown: −($120 − $84) = −$36
• Recovered product value: +$36 inventory value
• Net economic shortfall versus a kept order: $137.60 before the later resale, or $101.60 after recognizing the recovered $36 product value

The Inputs That Commonly Distort the Result

A unit count can overstate or understate the cost of returns because the operational event is usually the order. If one customer returns two dresses from a three-item order, that is two returned units but one return shipment, one support case and potentially one refund transaction. Conversely, three separate one-item orders create three handling events even if the unit total is identical. Report both unit return rate and returned-order cost, but do not divide order-level expenses by returned units without stating the assumption. This is why ecommerce return rate should not be treated as a complete cost measure.

Leaving out outbound shipping is another common distortion. A $100 order may have generated $8 of outbound freight before the customer returned it. If the merchant also pays a $7 return label and $3 of handling, the relevant logistics expense is $18, not $7. Whether the customer paid shipping does not automatically make the merchant’s original delivery cost disappear. Use the actual shipping economics for the order, including any subsidy or free-shipping threshold.

Refunds and exchanges should not be placed in one bucket. A $120 refund removes the sale, while a $120 exchange may retain revenue but still create a return label, inspection work and another outbound shipment. An exchange can also produce a partial refund or a price difference. Track the original order, returned item, replacement item and cash refunded separately; otherwise a high exchange volume can look like lost revenue when its main effect is operational cost—or hide a real loss when the replacement never ships.

A single average return cost also hides product differences. Suppose jackets generate $14 of average return expense and dresses generate $31 because dresses are returned more often and require more markdowns. Applying $22.50 to every return misstates both products and can lead to the wrong pricing or assortment decision. Calculate at product or collection level where order volume allows, alongside the broader view in product profitability.

Finally, returned inventory is not automatically worthless. A clean item restocked at its original selling price may preserve most of its product value; a worn item sold at a 40% markdown, or written off entirely, creates a different economic loss. Record recovered inventory at its realistic recoverable value, not automatically at zero and not automatically at original COGS. Inventory condition, markdowns and dead stock belong in the calculation, alongside the COGS treatment.

What a Return Cost Tool Should Show

A useful return-cost tool should start with complete Shopify records, not a refund total. It needs order and line-item IDs, product and variant, collection, quantity, selling price, discounts, taxes, shipping charged, payment fees, order status, delivery date, return date, refund amount, and refund status. It should also retain product cost and inventory fields so the calculation can distinguish COGS from return handling expenses. This follows the separation explained in Ecommerce COGS: What Belongs in Cost of Goods and What Does Not.

The tool should connect each return to its original order and product line, then show the result by product, collection, customer, sales channel, acquisition source, and period. Useful views include return rate by units, refund value against revenue, gross profit, net profit, contribution after advertising, remaining stock value, and dead stock. Customer views should show whether returned orders came from first-time or repeat buyers. Advertising data should sit alongside these results, with MER and attributed ROAS kept separate rather than treated as profit; see ROAS vs Profit: Why a High ROAS Can Still Lose Money.

The audit trail matters as much as the headline number. Each result should expose the source order IDs, included and excluded statuses, event dates, currency, refund and shipping treatment, product-cost assumption, ad-spend allocation, and calculation period. A user should be able to move from a collection total to the affected SKUs and orders, and reproduce the arithmetic without relying on a hidden adjustment.

For ORVX, Shopify provides the connected orders, products, and customers. It calculates realized revenue from delivered orders only, never counting returned or cancelled orders as revenue, and keeps each merchant's trading currency separate. It then applies the product cost model to report profitability and return metrics at product, collection, and customer level.

• Minimum comparison dimensions: product and variant, collection, customer, channel, acquisition source, order status, return status, and reporting period.
• Minimum reconciliation fields: order ID, line-item ID, original and refund dates, quantities, refund value, shipping treatment, product cost, advertising allocation, currency, and calculation version.

How to Act on the Number

Use return cost as a decision threshold, not a single store-wide score. At SKU level, compare return cost with contribution before returns. If a $60 product generates $24 contribution but costs $15 per returned order, returns consume 62.5% of that contribution. Flag it for fit, sizing, fabric, or quality review rather than immediately restricting returns. Repeat the comparison by return reason: “too small” points to size guidance, while “not as pictured” points to photography or merchandising.

Segment the result by customer behaviour and acquisition channel. A first-order customer acquired through a paid campaign may look unprofitable after one return, while a repeat customer may remain profitable across several kept orders. Calculate both immediate order contribution and customer-level profitability before changing eligibility or messaging. Compare channels using contribution after advertising, not ROAS alone; a channel can produce efficient ad revenue while returns erase its margin. See ROAS vs Profit for that distinction.

Use controlled policy tests rather than blanket restrictions. Test clearer size charts, garment measurements, model details, exchange incentives, or different return windows against a control group, then compare return cost, kept-order contribution, repeat behaviour, and customer complaints. Set a stop or fix threshold when the policy reduces return expense but also reduces profitable repeat purchasing. Profitability should guide the intervention, not discourage customers who return occasionally but remain valuable. Use product profitability to connect the SKU decision to the full margin picture.

• SKU threshold: investigate when return cost consumes a defined share of contribution, such as 50%, rather than when return rate alone looks high.
• Reason threshold: prioritise fixes when one reason explains a large share of a product’s refunds.
• Customer threshold: avoid penalising a segment until its total customer contribution, including repeat orders, is negative.

A return rate only tells you how often products come back; return cost tells you what those returns remove from profit. Calculate it from the complete order economics, validate the inputs against actual refunds and inventory outcomes, then use the result to target the products, policies, and operational steps creating the largest avoidable loss.