Paid returns: what the fee actually changes, and why the same policy behaves differently in every market

2026-08-18

A return fee does not reduce returns. It moves who pays for them, and it changes what your policy signals.

• Ecommerce
• Customer Expereince
• Returns
• Retail

For around 15 years, the standard belief in the field of online retail had been that free returns were the cost you had to pay if you wanted in. People accepted this because otherwise the sale would go to someone else. This belief has now been broken in the UK, in a quiet manner with little debate, and the evidence shows that it has only broken in one direction.

What has not happened alongside it is much clear thinking about what a return fee actually does. Most of the conversations I have been in treat the fee as a lever on return rate. It is not. It is a lever on who absorbs the cost of a return and on what the policy signals about the retailer, and those two things behave very differently from each other and very differently across markets.

The UK has already crossed over

Return Economics 2026, published in February by Ingrid with Harper Concierge and Limesharp, benchmarked the returns policies of the UK's top 100 fashion retailers against the same set in 2023. Adoption of paid returns went from 23% to 35%. The movement upmarket is sharper than the headline: affordable luxury went from 4% to 20%, and luxury from 0 to 12%.

What is more important than the adoption rate is the fact that there have been no reversals. None of the brands which introduced a fee between 2023 and 2026 has taken it back. The fees have also increased during that period, generally in line with inflation. When a policy change has withstood three years of competitive pressure from among a hundred retailers without any single instance of retreat, this provides reasonable evidence that the feared collapse in conversions did not occur to the extent that people had anticipated.

I want to be careful about what that evidence does not show. Absence of reversal tells you the trade was tolerable on balance. It does not tell you the trade was free, and it does not tell you it would be tolerable for your particular basket, category and customer mix. A retailer with a 28 pound average order value and a thirty percent return rate is in a completely different position from one selling 400 pound coats at a fourteen percent return rate.

Why a fee does not reduce returns much

More than 70% of fashion returns come from fit uncertainty. That single fact should govern the whole discussion. Fit uncertainty is generated at the moment of choice, on the product page, where the customer cannot resolve the question of whether this garment will fit this body. Bracketing, ordering two or three sizes with the intention of keeping one, is not a moral failing on the shopper's part. It is a rational response to an information gap the retailer created and did not close.

A charge applied after the fact does nothing to that gap. It arrives at the wrong point in the sequence, after the decision that generated the return has already been made. What it can do is suppress the bracketing behaviour a little, because a customer who knows they will pay 1.99 pounds per parcel has a reason to order one size rather than three. In practice, that effect exists but is modest, and it competes with a second effect that runs against it: the same customer may simply order less, or order from someone else.

The academic view holds that leniency should be considered as consisting of five separate factors rather than one aspect. This approach stems from the Janakiraman, Syrdal and Freling meta-analysis published in the Journal of Retailing, where return policy leniency was described in terms of time, money, effort, scope and exchange, and it was found that leniency increases purchases more than it increases returns, different dimensions having the effect of driving each of these outcomes. Since that paper is now more than ten years old, I looked into whether more recent research still supports it.

It does so, with some useful improvements. In 2025, Duong and his colleagues carried out a study combining machine learning with logistic regression on a large dataset consisting of Walmart's return policies and reviews. The degree of monetary leniency, along with that regarding effort and the scope, has an effect on both the likelihood of returning a product and on customer satisfaction. Time affects satisfaction but not the likelihood of returning a product. Exchange, on the other hand, works in the opposite direction. Their practical conclusion was that customers are willing to accept reasonable limitations concerning time, effort and exchange, and that a moderate level of leniency achieves a better balance between the number of returns and customer satisfaction than does maximum generosity.

Read that against the UK behavioural data and something useful falls out. Shoppers here start a return after 12.4 days on average, comfortably inside a 14-day window. A 30-day policy is therefore giving away three weeks of stock-in-limbo, markdown exposure and working capital to buy almost nothing in customer satisfaction, because the customers who were going to return have already done it. That is the restriction I would reach for first, and it is not the one most retailers reach for.

What the fee does change

There are two points, and both of them are commercial, not behavioural.

The first factor is the contribution per order. Reverse logistics on a parcel delivered to the UK amounts to between 10 and 25 pounds before the item is taken into account, and in the case of fashion products, when the discount on the returned item is factored in, the total cost can amount to a major portion of the original price of the item. A fee of two pounds is therefore not enough. It only covers a small part, and this small amount is most significant precisely in those cases where the ratio of return cost to order value is the worst, that is to say when the average order value is low, but the return rate is high. With a 30 pound order and a 30% return rate, recovering two pounds on each return does make a real difference to the contribution; but with a 300 pound order and a 12% return rate, the amount recovered is merely a rounding figure, and you have had to invest brand equity in order to achieve it.

The other factor is self-selection, and it is in this area that research into fairness becomes important. A study carried out in 2025 in the Journal of Retailing and Consumer Services concerning return fees showed that both the level of the fee and the way it is designed have an impact on perceived fairness and on the likelihood of a purchase, with more negative effects when the fees are higher and in the case of fixed fees. Perceived unfairness is not a minor issue; it is the factor that causes a fee to cease functioning as a cost-recovery tool and begin to become a customer acquisition problem, since fairness judgments are associated with the brand rather than with the transaction.

That is why the types of structure which work best with the UK data are conditional rather than universal. Boohoo charges a fixed fee but excludes Premier members in the case of one return per order. New Balance regards free returns as a benefit of membership rather than as a standard feature. ASOS never imposes a general fee at all; instead, it looks at return rates over a twelve-month moving period and only imposes charges on customers who are consistently returning about 70% of what they buy. All of these approaches target the cost at the group which is causing it while keeping the benefit intact for everybody else.

The legal floor is different in every market you sell into

This is the section that gets skipped, and it is the one that costs money. A returns policy is a commercial document sitting on top of a statutory floor, and the floor moves.

In the European Union, Directive 2011/83/EU provides consumers with a right to withdraw from distance sales within 14 days. The clause that has commercial importance is Article 14(1): the customer is responsible for the cost of returning the goods unless the trader has agreed to cover that cost or has not informed the customer that such a cost would be their responsibility. Therefore, free returns in the EU are essentially a commercial decision rather than a legal obligation, and it is the obligation to make the necessary disclosure that applies. Member states are also allowed to extend the withdrawal period beyond 14 days, which is the reason why the model instructions in the directive include a 30-day option.

The UK position is materially the same and comes from the same origin. The Consumer Contracts (Information, Cancellation and Additional Charges) Regulations 2013 give 14 days to cancel from receipt and a further 14 to send goods back. Regulation 35(5) puts the direct cost of return on the consumer unless the trader agreed otherwise or failed to give the required pre-contract information. Note what that means for a UK retailer charging a fee: you are not adding a charge, you are declining to waive one you were never obliged to absorb.

The United States has no federal right to change one's mind and get a refund. The cooling-off rule introduced by the FTC applies to door-to-door and off-premises sales where the amount is over 25$, not to online orders, and the Mail, Internet, or Telephone Order Rule is concerned with the timing of shipment and not with returns. In its place, there is a series of state-level disclosure rules. California requires its refund policy to be clearly displayed and regards the failure to do so as entitling the customer to a full refund within 30 days. New York's General Business Law section 218-a works in the same way. A number of other states have adopted this approach, but with different default provisions. The result in practice is that in the United States your policy is your own policy, and the legal risk lies in how clearly and visibly you display it rather than in the content of the policy.

China has a different system. According to Article 25 of the Law on the Protection of Consumer Rights and Interests, consumers have the right to return online purchases within 7 days without giving a reason, although there is a specifically listed set of exclusions which include specially made goods, perishable items, downloaded digital content and delivered periodicals, as well as another exclusion for goods to which the consumer has at the time of purchase accepted as unsuitable for return. The items must be returned in good condition, a refund must be issued within 7 days of the receipt of the goods, and the cost of returning them remains the consumer's responsibility unless the seller has agreed otherwise. The 2020 interim measures then placed the duty to enforce this rule on the platforms, which had to require sellers to honour the right and provide the necessary functionality for it.

Japan once more takes a different approach, thereby fooling almost all foreign companies that are entering the market. The Act on Specified Commercial Transactions does not provide for a cooling-off period in the case of mail order, this including online sales; instead, Article 15-3 sets a default rule: if you do not clearly state your return conditions, the customer has the right to return the item within 8 days of delivery at their own expense. If you do make your return conditions clear, then those conditions will apply. In Japan, therefore, the return policy is not something you have to work around but rather a document which determines your legal position, and the penalty for not drawing up such a policy is being bound by one that you did not select.

The Gulf is where I would be most careful, because the sources genuinely conflict. Saudi Arabia is the clearer case: the E-Commerce Law and its implementing regulations give the consumer the right to terminate and obtain a refund within 7 days of receipt where the product has not been used, with exclusions for custom-made goods, newspapers and similar, alongside a separate right to cancel if delivery exceeds 15 days. For the UAE, some legal commentaries read Cabinet Decision No. 66 of 2023, the executive regulations to the consumer protection law, as giving a 14-day return right on online purchases from UAE licensed sellers where goods are unused and in original packaging. Others, writing as recently as this month, state plainly that UAE law gives no automatic change-of-mind refund right and that the statutory protection covers defective goods. I have not been able to resolve that from primary sources, so I would treat UAE returns as a question for local counsel rather than something to settle from an article, mine included.

What breaks when you port one market's norm into another

Four modes of failure, all of which I have seen occur.

The first of these is having to pay for generosity that had not been requested. When a brand from the UK or the US enters the EU, it offers free returns since that is what customers in the home market have come to expect. By doing this, it effectively gives up a cost which Article 14(1) had already imposed on the customer in every market it enters, permanently, since withdrawing the right to free returns later appears to customers as a negative move, whereas not having introduced them in the first place does not. The money involved is real and increases as sales volume rises.

The other case is the reverse situation, and this is a compliance issue, not a margin one. If a US company applies its domestic policy to Europe, where that policy is basically down to what the retailer has decided, for example, 30 days' return period, store credit only, a restocking fee on opened items, then in the EU and in the UK that policy is subject to the statutory right to withdraw within 14 days with a cash refund, and the retailer's terms cannot override it. In fact, if information is missing before the contract is entered into regarding who is responsible for paying the return postage, the trader is then obliged to pay the cost by operation of law.

The third instance is about treating a policy as a promise in a market that sees it as an instruction. In Japan, the return terms as officially published make up the legal agreement. If a brand copies over a general English policy or, even worse, fails to include any information on the Japanese page, it does not get a lenient interpretation; rather, it has to comply with the 8-day default rule at its own cost, a rule that is more stringent than what was likely intended and one which imposes an obligation on the company that it did not foresee.

The fourth approach involves offering free return shipping in a market in which the law requires the customer to pay for it, then treating the resulting return rate as the standard for that category. In China, the 7-day right includes the consumer having to pay for the postage. If a foreign brand takes on that cost, it is not following local practice; it is in effect subsidising against local practice, and the return rate it records is not the actual market rate but the rate caused by its own subsidy. Therefore, any later decision to withdraw the subsidy is being based on a baseline that was never genuine.

How I would actually decide

Not by means of a rule, but by means of three numbers and one sequence.

The ratio in question is the amount obtained when the total cost of returning a product is divided by the average order value, within each category, not averaged across categories. This is calculated by taking the sum of the reverse logistics cost, the processing cost, and the markdown on the returned item and dividing it by the average order value. If this ratio is high, then the fee is justified; but if it is low, the fee becomes a cost to the brand, even though there is a rounding advantage.

The second is the distribution of return behaviour across the customer base, not the average. If a small share of customers generates a disproportionate share of returns, and it usually does, then a blanket fee charges the whole base to solve a problem concentrated in a fraction of it. That is the arithmetic that produced the ASOS design, and it is generally the better answer.

The third point relates to the return window in relation to the actual time of return. In this situation, if your customers initiate returns after 12.4 days and your window is 30 days, then you are retaining a liability for 18 days during which you gain very little, and the research indicates that time is the aspect of leniency which customers care about the least.

The order of the items is just as important as the numbers themselves. You should address the fit gap before assessing the consequences of it. The size advice has been developed from the actual reason codes for returned items, fit feedback drawn from reviews at the size level, imagery that displays scale on real people, and a return process that is simpler than the refund process. These are the measures that act at the stage when a return is created; the fee, on the other hand, applies at the stage when the return has already occurred, and a business that opts for the fee first is effectively charging the customer for its own unresolved information issue.

What I would report instead of return rate

Return rate is a weak management number because it is mostly determined by category mix. Contribution per order after returns is the number that moves with your decisions. Alongside it, exchange rate as a share of returns, because an exchange retains the revenue and the customer while a refund retains neither, and the cost per return processed, because that is the figure a policy change is supposed to move.

Then, and only then, the fee question becomes answerable, because you can see what you are buying and what you are spending to buy it.