In short
- A return is recorded as a refund, not as a cost. So it appears in no expense dashboard, and nobody arbitrates it.
- The e-commerce return rate is 19.3%, not to be confused with the 15% that describes the whole of retail, physical shops included. In fashion, the only European players publishing a firm figure sit at around 45 to 50%.
- Processing a return costs less than everyone repeats: the best European measurement available, covering 411 retailers for the year 2021, gives 2.85 euros per item and 6.95 euros per parcel, transport and handling included, before the loss of value on the product.
- The right indicator is not the return rate but the margin on the order once returns are deducted. A channel at 30% returns can still be the most profitable, and one at 8% can be destroying value.
Here is why this line escapes everyone. When a 90 euro order comes back, the shop records a 90 euro refund. Revenue drops by that much, and the affair looks neutral. It is not: the outbound shipment was paid for, the return is often free for the customer and therefore paid by you, the product was handled twice, inspected, sometimes repackaged, and it sometimes comes back unsellable at full price.
That expense has no line of its own. It is spread across transport, logistics and markdown, three cost centres nobody reconciles with the order that caused them.
The calculation almost nobody does
It requires no tooling, only a decision about what to count.
| Item | What belongs in it | Order of magnitude |
|---|---|---|
| Outbound shipping | Lost, the order no longer exists | Your real negotiated rate |
| Return transport | Yours as soon as the return is advertised as free | No primary source establishes that it costs more than the outbound leg |
| Handling and inspection | Receipt, checking, restocking or scrapping | From a few euros to over ten depending on the process |
| Loss of value | Damaged item, out of season on return, or repackaged | The heaviest item in fashion, and the most ignored |
| Payment fees | Rarely refunded in full when you issue a refund | Small per unit, significant at volume |
The mistake that costs the most Steering this by the return rate. A falling rate can perfectly well accompany a falling margin, if the fall was obtained by tightening conditions to the point of losing sales. The indicator to track is margin per order once returns are deducted, not the percentage of parcels that come back.
What market data says
The e-commerce return rate is 19.3%. The 15% quoted everywhere describes the whole of retail, physical shops included, and does not apply to an online shop. Variation by sector remains wide: 5 to 8% in food and cosmetics, and in fashion, the only European players publishing a firm figure sit at around 45 to 50%. As for the processing cost, this is where almost every source gives way: the best European measurement available, 411 retailers for the year 2021, gives 2.85 euros per item and 6.95 euros per parcel, transport and handling included, before the loss of value on the product. The same researchers write that the 10 to 15 euros put forward by earlier studies, their own included, overstated the real cost.
Two cautions belong here. A survey of 146 German-speaking merchants in the summer of 2024 does not produce an average but a spread of self-reported estimates: 18% put the cost below 5 euros, 30% between 5 and 10 euros, 26% between 10 and 20 euros, and 18% say they do not know what it costs them, which is the soundest data point of the lot. And no primary source establishes that return transport costs more than the outbound leg, let alone a multiple of it: the one open measurement puts checking and quality control, at 67%, just ahead of transport, at 64%. Those two items dominate, and they weigh much the same.
These orders of magnitude serve one purpose: calibrating the effort. At a thousand monthly orders and a 20% rate, this line runs into thousands of euros a month once the loss of value is included. It deserves the same seriousness as an advertising line of the same size, which gets reviewed every week.
The three levers, in order of return
- Cut the disappointment returns. The customer expected something else. They are handled upstream, on the product page: real dimensions, a size guide based on your own returns rather than the supplier’s, photos on several body types, fabric described honestly. It is the highest-return lever because it avoids the cost rather than reducing it.
- Speed up the return to sale. An item that comes back in three days and is restocked the next day keeps its value. The same item processed over three weeks comes back out of season. The speed of the loop is often worth more than making it free.
- Steer by product, not globally. The return rate is an average that hides everything. Pull the ranking of your twenty most-returned references: you will almost always find two or three products carrying a disproportionate share, with an identifiable and fixable cause.
One-hour self-check: pull the last six months of returns by reference, and set the ranking against the stated reason. If the same product keeps coming back with the same reason, you do not have a returns problem, you have a product page problem, or a sourcing one.
The options, and what they are worth
| Option | What you get | Who it suits |
|---|---|---|
| Fix the pages of the most-returned products | The best return, because it removes the cost rather than moving it. Visible within weeks | Everyone, starting with twenty references |
| Smooth the returns loop | Improves both return to sale and satisfaction. Needs logistics organisation more than technology | Regular volumes, seasonal products |
| Charge for returns | Lowers the rate and the revenue. Works on high baskets and established brands, dangerous elsewhere | To test on a segment, never as a blanket switch |
| Tighten conditions and deadlines | Lowers the displayed rate, damages the relationship, and moves the problem to reviews and customer service | No case we recommend |
Tooling
Two distinct needs. Giving the customer a clear returns journey, which speeds up the loop and above all yields usable reasons. And anticipating: some orders carry the signs of a likely return from the moment they are placed.
On PrestaShop we deploy the modules published by Datafirefly Limited, our agency’s sister company: Return Portal for the returns journey and structured reason capture, and Returns Predictor to flag at-risk orders. One-off purchase · 12 months of updates.
Method beats tooling, and the return reason illustrates it well: a module is useless if the reasons on offer are “does not suit” and “other”. The labels decide what you will learn, and they are written before installation.
What this says more broadly
This line is invisible for a structural reason: a shop’s accounts are organised by nature of expense, not by order. Transport is one line, logistics another, markdown a third, and none of them traces back to the sale that caused them.
It is the same mechanic that makes acquisition cost misleading when you do not set it against real margin. ROAS is its most expensive illustration: it compares revenue to spend, and never subtracts the returns that the sale brought with it. Seeing the full cost of an order, returns included, is measurement work before it is logistics work, and it is the ground of our data and analytics practice. The link with the buying journey is direct: what gets decided on the product page gets paid for in the warehouse.
Sources
The figures quoted in this article point to the study that produced them, opened on 11 August 2026. We do not cite a source we have not read.
- National Retail Federation and Happy Returns, 2025 Retail Returns Landscape, October 2025, surveys run in summer 2025 among 2,006 US consumers and 358 e-commerce professionals at US merchants. Read it
- Asdecker, Felch and Karl, European Return-o-Meter, Forschungsgruppe Retourenmanagement, University of Bamberg, 2022, survey of 411 European retailers covering the year 2021. Read it
- EHI Retail Institute, Versand- und Retourenmanagement im E-Commerce 2024, September 2024, survey of 146 retailers in Germany, Austria and Switzerland. Read it
- ABOUT YOU Holding AG, IPO prospectus, 15 June 2021, return rate reported as a share of gross items for financial years 2018/19 to 2020/21. Read it
- Zalando SE, IPO prospectus, 2014, return rate reported by value. Read it
- Stripe, official refunds documentation, policy on non-return of processing fees. Read it
FAQ
What return rate is normal?
Around 19% in e-commerce, 5 to 8% in food and cosmetics, and 45 to 50% in fashion among the European players who publish a firm figure. The 15% often quoted describes the whole of retail, physical shops included. But your useful benchmark stays your own rate from the previous quarter, at comparable catalogue.
Should returns be paid for?
It is a commercial trade-off, not a best practice. The rate falls and so does revenue, and the balance depends on your average basket and the strength of your brand. Test on a segment before any generalisation.
How do I calculate the real cost at my company?
Add up the five lines in the table above over one month, and divide by that month’s number of returns. The figure is imperfect and sufficient to arbitrate, which is all it is asked to do.
Are returns a bad signal in themselves?
No. A generous returns policy raises conversion and basket size. The problem is not the return, it is the avoidable return: the one an honest product page would have prevented.
Can Dotsland help?
Yes. Calculating the full cost per order, ranking the most-returned references against stated reasons, fixing the pages responsible and setting up tracking. It is part of our data and analytics practice. Let’s talk, or start with the last six months’ ranking.