Returns are the cost that arrives wearing the face of good service. Each one looks like a refund line, feels like a courtesy, and quietly takes back more than the sale ever gave: the margin, the parcel that already shipped, the label coming back, the minutes at the returns bench, and sometimes the unit itself, no longer sellable as new. A store with strong products and a strong return habit can out-sell its own profitability for years without the dashboard ever saying so. This post treats returns as the margin problem they are: how to price the real damage, where returns concentrate so the fix can concentrate too, which levers actually reduce them without punishing the customers you want to keep, and how to keep the lag between a sale and its return from corrupting the numbers you steer by.
- The damage is always bigger than the refund line: margin handed back, outbound fulfilment stranded, return postage and handling added, and a slice of units written down.
- Returns concentrate hard: a few products, usually sized or fit-dependent ones, carry return behaviour at multiples of the store average, so the fix can be surgical.
- The levers that work are preventive, not punitive: sizing truth, expectation-true photos, fit feedback and delivery speed beat fees and bans on every axis that matters.
- Returns lag sales by weeks, so book every refund to the day it was issued and read young months as immature, or a strong month will flatter itself and bill its successor.
Price the damage before choosing the medicine
The refund is the visible quarter of the iceberg. Underneath: the contribution margin the order had earned, now reversed; the outbound parcel and payment fee, spent and never coming back; the return label and the handling minutes, newly added; and, on a share of returns, a unit that comes back unsellable at full price. Itemised properly, the true cost of a return exceeds the refunded amount, and the full anatomy, with the example store's own figures, is walked through in what a return really costs. The reason to price it before acting is proportion: a store that thinks a return costs 'the refund' will under-invest in prevention by half, and a store that prices the whole iceberg suddenly finds that a better size guide has one of the best returns on investment in the building.
Where returns concentrate
Store-level return rates are almost useless for action, because returns are never evenly spread. They concentrate by product: sized and fit-dependent items return at multiples of the catalogue average, and one problematic bestseller can account for a third of the returns bench by itself. They concentrate by behaviour: bracketing, the same item ordered in two sizes with one always coming back, is invisible in order data and obvious in return data. They concentrate by customer age: first orders return more than repeat orders, because fit and expectations are unproven. And they concentrate by acquisition path: discount-driven and impulse traffic returns more than search-driven demand. Cut the return list by product and the fix stops being a policy question and becomes a product question, the same per-SKU discipline as which products actually make money, where the ranking corrects each product's wins by its take-backs.
What normal looks like, and when to worry
Category decides the baseline: consumables and accessories live in single digits, fashion lives far higher, and footwear higher again, so a store should benchmark against its own category and its own history rather than a universal number; what counts as a normal return rate covers the reference points. The actionable signals are relative: a product returning at twice its category's norm, a rate trending up while the catalogue is stable, or a gap opening between two size curves of the same product. Absolute rates start conversations; relative anomalies name the fix.
| Pattern | What it means | The lever |
|---|---|---|
| One product far above its category norm | a fit or expectation problem on that product | size guidance, truer photos, or a spec fix |
| Two sizes bought, one always back | bracketing: the size chart is not trusted | fit data on the page, size-advice prompts |
| First orders return, repeats do not | expectations set wrong before the first box | honest imagery and copy at the ad and page level |
| Returns spike after long delivery times | the need passed before the parcel arrived | delivery speed and honest delivery promises |
| Discount-window orders return more | impulse purchases cooling off | target promotions at proven intent, not cold reach |
The levers, preventive before punitive
Prevention works because most returns are decided before the parcel ships: the size chart that did not match the garment, the photo that flattered the colour, the delivery estimate that slipped past the occasion. Sizing truth, expectation-true imagery, visible fit feedback from buyers, and faster dispatch each remove a reason to return, cost nothing per order once built, and improve conversion at the same time, which no punitive lever can claim. The punitive end, charging for return postage where your market's consumer rules allow terms to say so, or restricting serial-return accounts, does reduce rates, and it also taxes every honest customer and hands a talking point to competitors, so it belongs at the end of the list, applied narrowly, after the preventive levers have taken the cheap wins. A store that starts with fees is treating its best customers for a disease a few products have.
Survive the lag: recognition and reserves
Between the sale and its return sit two to four weeks, and that lag corrupts unmanaged numbers twice. Backwards: a refund netted against its original order rewrites a month you already closed, which is why which day a refund belongs to has a firm answer, the day it was issued. Forwards: this month's return rate is immature, because its sales have not finished returning yet, so a young month always looks better than it will end up, and comparing it against a matured one flatters the present. Read rates at matched maturity, hold a cash reserve sized on your strongest recent month's expected returns, and the lag becomes a known property of the business instead of a monthly ambush.
The two-week plan
- Price one average return end to end, refund, stranded outbound costs, return postage, handling, write-down share, so every later decision uses the real number.
- Rank the last quarter's returns by product and by size; the concentration will usually hand you two or three specific fixes.
- Fix the worst product's page first: measured size chart, colour-true photos, and the fit feedback your buyers have already left.
- Book refunds to their issue day from now on, and re-read the last three months that way once; the months will move, and the moved version is the true one.
- Set the reserve, then re-measure in a quarter at matched maturity, product by product, because the store average will hide exactly the progress you made.
Three questions that always come next
Should returns change which products I advertise? Directly. A product's advertised margin is its corrected margin, wins minus take-backs and their costs, and a high-return product can sit below the threshold where an ad euro pays for itself while its low-return neighbour clears it comfortably. If a product's corrected contribution cannot fund its acquisition, pause its ads before delisting it, the same floor arithmetic as ads eating your profit, applied one SKU down.
Are free returns worth it? They convert, measurably, and they cost, measurably, and the honest answer is that both numbers are yours to compare rather than a policy debate to win. Price the conversion lift free returns buy on your traffic against the iceberg cost of the returns they invite, per category if the store spans several. Many stores land on a split: free where returns are cheap and rates low, customer-paid where a category's return economics cannot carry it, stated plainly either way.
When is a high rate acceptable? When it is priced in and still profitable. A fashion store running far above a consumables store's rate is not failing, it is operating its category, and the only question that matters is whether the corrected unit economics, margin after returns at maturity, still clear the store's costs. Returns are a cost of honesty in categories where fit is a gamble; the sin is never the rate itself, it is steering by numbers that pretend the rate is zero.
From courtesy to line item
The stores that beat their returns problem all make the same first move: they stop treating returns as service weather and start treating them as a cost line with structure, concentration and levers. Priced fully, located precisely, reduced preventively, booked to the right day and read at maturity, returns shrink from a margin killer to a managed cost, and the statement finally says so, which is the point: nouz books every refund to its issue day and carries the return costs beside it, so the correction happens in the numbers you read every morning rather than in a quarterly unpleasant surprise.