Returns as a retention system: the policy is part of the price, and the portal is part of the loyalty program.
Most brands run returns as a cost center: a warehouse queue, a label vendor and a policy page nobody has reread since launch. The research says that’s backward. How you handle a return has a large say in whether that customer buys again.
The first number says returns are not an edge case: nearly one dollar in six comes back, and about one in five online Reported. The second says what a clumsy return does to the customer who sent it. Customers who paid for their own return spent 75% to 100% less over the next two years; those whose return was free spent 158% to 457% of what they had before Published. The gap is too large to be all policy, as chapter 3 explains, but the direction matches the rest of the research: a good return keeps a customer.
So this guide follows the return as a trip the customer takes with you, from the promise on the product page to the next order. Each stop can be measured, and each is cheap to fix compared with buying a new customer.
A return is the only time a customer tells you, in writing, what went wrong. Most brands pay to hear it and then don’t read it.
It builds on The Second Order for cohorts and The Whole Machine for contribution margin, and links to them rather than repeating them.
Start with The Returns Audit, or with the one-page map just below. Or follow a path:
Three calculators and a scored audit run in the page. Nothing you type leaves your browser.
Examples that open with Say or Picture use made-up round numbers. Every source is listed in Appendix C.
What this guide argues, and what would prove each claim wrong.
A position says what would prove it wrong. Test each on your own store.
Eight stops on the trip a return takes. Each has a number that tells you how it’s going, and a way it quietly goes wrong.
Every return passes the same eight stops. Most brands measure the return rate and the cost, and nothing in between.
The whole book is above and always will be. These are the same chapters addressed individually, for linking to one idea rather than to ninety.
| Stop | The question | The number that tells you | Where it leaks |
|---|---|---|---|
| 1. The promise | Did they know the terms before they paid? | Tickets asking about returns, per 100 orders | Terms that differ between product page, policy page and portal |
| 2. The request | Can they start a return in under a minute? | Portal start-to-finish rate; return tickets per 100 returns | An order lookup that needs a number nobody kept |
| 3. The choice | Were they offered something better than money back? | Share of returns ending in exchange, credit or refund | A hidden refund button, or a bonus paid to people who’d have exchanged anyway |
| 4. The trip | Is sending it back easy? | Days from request to first carrier scan | Labels that need a printer; drop-off points far away |
| 5. The money | When does the refund land? | Days from first scan to refund | Refunds waiting in the warehouse queue |
| 6. The goods | What happens to what came back? | Share back on sale within 14 days; recovery rate | Returns sitting in a corner losing season |
| 7. The reason | What did it teach you? | Top three reasons by product, and “other” as a share | Vague reason codes nobody reads |
| 8. The next order | Did they come back? | Twelve-month repeat rate by how the return ended | Nobody measures it, so every policy is set on cost alone |
The stops interact. A fee at stop 3 lowers cost at stop 6 and can quietly empty stop 8. Strict inspection protects stop 6 and stretches stop 5 by a week. Fixing one stop in isolation often moves the cost rather than removing it.
Twelve checks on whether your returns keep customers or lose them. About forty-five minutes with your returns app, your order data and a phone.
The audit isn’t about your return rate. A low rate can mean a great product, or a policy so hostile that unhappy customers keep the item and never come back. It asks what each return costs, what it does to the next order, and whether the trip is one a customer would take twice.
Open your returns app, twelve months of orders and refunds, and your policy page. Score each check 0 to 2: 0 if it failed or nobody can answer it, 1 if partly true, 2 if clean. “Our vendor handles that” isn’t an answer.
A return rate on its own tells you almost nothing. What happened to the customer afterward tells you almost everything.
Score as you go; your band appears when all twelve are in.
| Score | What it means | Read next |
|---|---|---|
| 20–24 | Your returns already work as retention. Your job now is to prove it with holdouts and push the exchange share up. | Exchange First, then The Returns Scorecard |
| 14–19 | The flow is decent but you can’t yet see what it does to the next order. Measure that first, then fix the zeros. | Repeat Rate by How the Return Ended, then the chapter linked from your lowest check |
| 8–13 | Returns are run as a cost center. Customers who return are probably leaving at a rate you haven’t measured. | Part one, starting at What Lenient Policies Do |
| 0–7 | Start with the basics: one set of terms, a portal that works on a phone, and refunds that don’t wait on the warehouse. | The Portal Is a Flow, then The First Thirty Days |
If you sell consumables that rarely come back, score checks 6 and 9 on how you handle damaged or wrong deliveries instead.
A shopper who can’t touch the product buys insurance along with it. The return policy is that insurance, and it has a measurable value.
Online, every purchase is a small bet. The customer can’t try the shoes on or hold the sweater to the light. The return policy tells them what the bet costs if they lose, which makes it part of the price.
Eric Anderson, Karsten Hansen and Duncan Simester put a number on it. Using a catalog retailer’s purchase and return data, they modeled the right to return as an option the customer holds, like insurance. For women’s footwear, the option was worth more than $15 per purchase on average, and having it raised purchase rates by more than 50% Published. The value varied a lot by customer and by category, which is the useful part: the option is worth most where the customer is least sure, on a first order, in a category with sizes, colors or textures that photographs don’t settle.
A return policy is insurance the customer buys with every order. Price it like insurance, and write it so they can read it.
“Lenient” isn’t one setting. Narayan Janakiraman, Holly Syrdal and Ryan Freling reviewed the research on return policies and found that it describes leniency along five dimensions. Their meta-analysis of 21 papers then asked what each dimension does to purchases and to returns Published.
| Dimension | What it means | What the meta-analysis found |
|---|---|---|
| Money | How much of the price comes back: full refund, fee deducted, restocking charge | More lenient raises purchases |
| Effort | How hard it is: printer, box, drop-off distance, forms, receipts | More lenient raises purchases |
| Scope | What can come back: sale items, opened items, worn items | More lenient raises returns |
| Time | How long the window is | Longer windows reduced returns |
| Exchange | Whether you can swap rather than refund | Exchange options reduced returns |
PublishedJanakiraman, Syrdal and Freling, Journal of Retailing, 2016. Overall, leniency increased purchases more than it increased returns.
Two findings deserve attention. Time works the opposite way from what most operators assume: longer windows went with fewer returns, possibly because of the endowment effect (the longer you own a thing, the more it feels like yours). And the dimensions that sell (money and effort) differ from the one that drives returns (scope). Be generous where it buys orders and firm where it only buys returns. One caution: many of the studies are scenario experiments, in which people read a policy and say what they’d do. Treat the table as a map of where to test, not a result.
Four lines of research point the same way: a good return experience makes the next purchase more likely. None of them says returns are free.
The usual view is that a customer who returns a lot is a cost to be managed. The research says that’s incomplete, and sometimes backward: returners are often your most engaged buyers.
J. Andrew Petersen and V. Kumar studied individual customers’ buying and returning over several years. Their 2009 paper in the Journal of Marketing concluded that returns are “inevitable but by no means evil”: they are part of the buying relationship, and up to a threshold, allowing them raises profit Published. Their 2010 article for MIT Sloan Management Review makes the same case for managers: a lenient policy, managed well, can earn more than a strict one Published.
Read it as a curve with a peak. Choke returns off and you lose the purchases the safety net made possible; let them run and the cost overwhelms the benefit. The peak is not near zero. Their 2015 paper in the Journal of Marketing Research reports a six-month field experiment with 26,000 customers of an online retailer, in which accounting for the risk-reducing effect of returns raised both short- and long-term profit Published.
Amanda Bower and James Maxham followed customers of two online retailers over 49 months, combining two surveys with actual spending. Customers who paid for their return shipping cut their spending with the retailer by 75% to 100% over the next two years. Customers whose return was free spent 158% to 457% of their pre-return level Published. The retailers charged for return shipping when they judged the customer to be at fault.
So the customers who paid weren’t a random group; they were the ones the retailer blamed, and some of the gap is who they were. What survives is the direction, and the mechanism the authors point to: regret over paying for a return was the strongest predictor of whether a customer bought again Published.
Stanley Griffis and colleagues used an online retailer’s order and return records and found that customers who had a return went on to buy more often, bought more items per order and bought higher-value items. The faster their refund was processed, the larger the increase Published. That’s an association, not an experiment, but it’s consistent with the others, and it’s the reason chapter 8 exists.
People who run returns for large retailers say the same thing in plainer words. Amena Ali, chief executive of Optoro, a returns technology company, told the Associated Press in 2024 that “your most profitable customers tend to be high returners” Reported.
Customers who return are often the ones who buy the most. A policy built to punish returns punishes them first.
A refund, an exchange and store credit look alike in the portal. On the ledger they are three different events, and an exchange is usually worth far more than it costs.
Most brands know their return rate. Few know what one return costs, or what it’s worth to turn a refund into an exchange. That second number decides how much bonus credit you can afford.
Every return carries the same basic costs, whatever the outcome: the label, receiving and inspection, repackaging, and a write-off on the share that can’t be sold again at full price. The Wall Street Journal reported in 2021 that processing an online return can cost $10 to $20, excluding freight Reported. Ask your 3PL for your own number, and finance for the share written off.
| Outcome | What happens to the money | What happens to the goods | Extra cost |
|---|---|---|---|
| Refund | The sale is gone | Back on sale, or written off | None beyond the return |
| Exchange | The sale stays | One item back, another out | A second shipment, and a chance the new item comes back too |
| Store credit | Stays with you until spent | Back on sale, or written off | The goods and shipping when the credit is spent; any bonus |
Store credit needs care. Unspent credit is a liability, not income, and state rules on unused credit vary. Don’t plan on breakage (credit never spent) as profit, and have counsel and your accountant set how credit is recorded and whether it expires.
Say a brand ships 10,000 orders a month at an $80 average item price, with a 65% product margin, so each item costs $28. Fifteen percent of orders come back: 1,500 returns a month. Bringing one back and processing it costs $12, and 20% of returned items can’t be sold again at full price.
If 20% of returns end in exchange today and the bonus lifts that to 30%, the brand gains about $3,500 a month, after paying the bonus to everyone who would have exchanged anyway Derived. The bonus pays for itself if it lifts the exchange share to about 22%. That’s the tool’s default below.
With the defaults, a refund costs $17.60 beyond the lost sale, an exchange is worth $32.56 more than a refund, and a $10 bonus pays if it lifts the exchange share from 20% to about 22.1%. At 30%, it gains about $2,330 per 1,000 returns. Raise the second-return rate and watch the value erode: an exchange should be for the right size, not a random swap.
Split your customers by the outcome of their first return, watch what they do next, and judge every policy change by that, with a holdout to keep you honest.
Return cost shows up on this month’s P&L. The benefit of a good return shows up months later, as an order from someone who might have left. Measure only the first, and every policy decision drifts toward cutting cost and quietly losing customers.
Take every customer who made a first purchase in a given quarter, and split them by what happened to their first return: none, refunded, exchanged, or store credit (add returnless refunds if you use them). Track each group’s repeat rate at 90 days and twelve months, and their contribution. Appendix A has the query; The Second Order covers cohorts in depth.
Expect the exchange group to look best, and don’t read too much into it: customers who exchange were already more committed. The split is a map of where the value sits, not proof of what caused it.
The split tells you where to look. Only a holdout tells you what a policy did.
Almost every change in this guide can be randomized by customer: bonus credit on or off, refund at scan or at receipt, fee or no fee. Keep a random 10% to 50% of customers on the current policy and compare repeat purchase and contribution per customer. The Honest Test covers reading a test without fooling yourself.
Repeat rate is slow and noisy, so size the test first. To see twelve-month repeat move from 30% to 33%, you need roughly 3,700 customers per group Derived, from the rule of thumb 16 × p(1 − p) ÷ d² (80% power, 5% significance). Read a signal at 90 days; decide at twelve months.
A more generous policy has a cost you can count: labels you now pay for, a few more returns. The tool below turns that cost into the repeat-rate lift it has to buy. For a fee you’re considering, enter its expected income as the cost of not charging it: the answer is how much repeat purchase the fee can lose before it loses money.
With the defaults, free labels cost $49,500 a year, which repeat purchase repays if the twelve-month repeat rate rises 3.3 points, from 28% to 31.3%. If only the 20% of customers who return something respond, their repeat rate must rise 16.5 points. A holdout needs about 3,000 customers in each group to see the overall lift. Whether that’s plausible is exactly what Bower and Maxham’s numbers suggest and don’t prove.
Treat the return portal the way you treat checkout: a sequence of steps, each with a drop-off rate, ending in an outcome you can raise.
Brands spend months on checkout and install a returns app in an afternoon. Yet the portal is where a disappointed customer decides whether you’re worth another try. It has steps, a conversion rate (to exchange) and abandonment: the customer who gives up and emails support, or calls their bank.
Then the messages: at first scan, when the refund or credit is issued, and when an exchange ships. They’re transactional, with no promotions stacked on top; anything promotional goes only to customers who agreed to marketing.
Effort is one of the two dimensions that raised purchases in the meta-analysis in chapter 2. In the 2024 NRF and Happy Returns survey, 84% of consumers said they were more likely to shop with a retailer offering returns with no box, no label and an immediate refund Reported. Happy Returns, a UPS company, runs box-free drop-off points, so that’s vendor data, and stated preferences run ahead of behavior. The direction is still right.
The customer who can’t finish your return portal doesn’t disappear. They email support, post a review, or call their bank.
A damaged, wrong or missing item isn’t a return; it’s your mistake. Send the replacement, ask for a photo if the carrier claim needs one, and don’t make the customer ship anything back.
Starts to completions in the portal; return tickets per 100 returns (every “how do I return” email is a portal failure); days from request to first scan; and the outcome mix, weekly.
Order the choices so the right size comes before the money back, pay a bonus only where it changes the answer, and never hide the refund.
Most returns in sized categories aren’t a rejection of the brand; it didn’t fit, or the color was off. An exchange keeps the sale and the customer, and in chapter 4’s example was worth about $33 more than a refund. The portal’s job is to make it the easiest right answer.
In the meta-analysis in chapter 2, exchange leniency reduced returns Published. A 2026 study of a fast-fashion retailer found that customers who visited a store to collect or return an online order were more likely to exchange; follow-up analyses suggest seeing the product reduced uncertainty Published. Without stores, borrow the mechanism: at the moment of exchange, answer the fit question, for example with what customers who returned for “runs small” chose instead.
An exchange that ships only when the old item reaches the warehouse can take two weeks, and by then the customer has bought elsewhere. Ship at the first carrier scan, or at the request for trusted customers, using the trust tiers in chapter 8.
Loop, a returns platform for Shopify stores, reports that across more than 4,000 merchants in the year to October 2025, 73.6% offered exchanges and 49.2% let customers shop with their credit in the portal; of those, 51.7% added a bonus, averaging $11.28 Reported. That’s vendor data: it tells you what’s common, not what works.
What works depends on how many customers the bonus moves, since it’s paid to everyone who exchanges, including those who would have anyway. The chapter 4 tool gives the break-even share. Keep it a flat amount, spent at once in the portal, not on deeply discounted items, and tested against a holdout.
A bonus is paid to everyone who exchanges. It only earns on the ones it changed.
An exchange that comes back is a refund with extra shipping. If “too small” exchanges come back as “too big”, your size guidance is the problem; that’s a fix for chapter 10.
To the customer, the money is theirs the moment the parcel leaves their hands. Pay it then for most customers, and wait for inspection only where the risk is real.
A refund that waits for the parcel to reach the warehouse, sit in a queue and pass inspection can take two or three weeks. The customer spends that time checking their bank balance and thinking about you. It’s the last impression of the trip.
| Tier | Who | When the money moves |
|---|---|---|
| Instant | Customers with a long, clean history and a low return rate; low-value items | When the return is requested |
| At first scan | Most customers | When the carrier or drop-off point first scans the parcel |
| At inspection | Customers above your serial-returner threshold, high-value items, categories with known fraud | When the warehouse receives and checks it, within a stated number of days |
Pay at the scan for most, wait for the warehouse for a few. Decide which is which by rule.
The risk of paying before inspection is the empty box, or a different item sent back. In the 2025 NRF and Happy Returns survey, 65% of retailers that track return fraud saw more empty-box returns Reported. Box-free drop-off, where a person scans the item itself, removes the empty box; weight at first scan catches obvious mismatches; and the tiers keep scan refunds away from the few customers where fraud concentrates (chapter 11).
Under the EU Consumer Rights Directive, a trader must refund within 14 days of being told the customer is withdrawing, and may hold the refund only until it has received the goods back or the customer has supplied evidence of having sent them, whichever comes first Published. A carrier’s tracking scan is the kind of evidence that can end your right to wait. Have counsel confirm how that applies in the countries you ship to.
Some returns cost more to bring back than they’re worth. Some come back and lose value every week they wait. And some customers want to bring things back that they bought years ago.
What happens after the customer lets go of a return is invisible to them and expensive for you. Three decisions matter: whether to bring it back at all, how fast to get it back on sale, and whether to invite old products back as trade-ins.
The Wall Street Journal reported in January 2021 that Amazon, Walmart and Target were telling some customers to keep items they wanted to return, because for cheap or bulky items it was often cheaper to refund the price than to ship the item back. Walmart said it decided using the customer’s purchase history, the value of the product and the cost of processing the return Reported. By 2024, the Associated Press found returnless refunds used widely and quietly, described by people in the industry as an unofficial, discreet loyalty benefit Reported.
The rule is arithmetic: bring an item back only if what you recover is worth more than the trip. Say an item sells for $18, costs $6 to make, and 80% of returns can be resold. You recover about $4.80 of goods. If the label and handling cost $11, every return of that item loses $6.20 more than letting the customer keep it.
A returned item loses value every week it waits: its season ends, its color gets marked down. Measure days from receipt to available for sale. Sort returns at receiving into back to stock, seconds sold at a discount, recycling, and bulk liquidation. Every stream but the first loses margin, so the fastest win is usually a shorter queue for the first.
Some brands invite products back years later. Patagonia’s Worn Wear program buys back eligible used Patagonia gear to give it a new home Reported. For durable goods with a resale market, such as outerwear and bags, trade-in credit works like store credit: it brings the customer back to shop. Run it as its own program with its own economics, not as part of the returns policy.
Every return comes with a free report on what went wrong. Most of the defects it finds are in the product page, not the product.
A factory that shipped the same defect thousands of times would stop the line. Online brands ship the same wrong size chart and misleading photo every week, with the reason sitting in a database nobody opens. The cheapest return is the one you prevent, and the reason code tells you how.
Petersen and Kumar note that only about 5% of products are returned because they’re defective Published. The rest are gaps between what the customer expected and what arrived. The “defect” is usually in the information, so the product page is the first place to look.
Two studies show how the gap gets made. Alec Minnema and colleagues found, at an electronics and furniture retailer, that reviews more glowing than a product’s long-term average raised purchases and returns, by setting expectations the product couldn’t meet Published. Edlira Shehu, Dominik Papies and Scott Neslin found that free-shipping promotions pushed customers toward riskier items and raised returns enough that, in their simulation, the promotions lost money Published. The First Offer covers shipping as price.
Vague codes (“didn’t like it”) can’t be acted on. A good list is specific, short and tailored by category. For apparel:
| Group | Reasons | Usual fix | Owner |
|---|---|---|---|
| Fit | Too small; too big; too long; too short; fit in the wrong place | Size chart, fit notes, model measurements, fit-specific reviews | Ecommerce |
| Look | Color differs from photo; looks different in person | Photos in daylight, color notes, video | Creative |
| Feel | Material or quality not as expected | Fabric weight and feel described plainly; close-up photos | Product, ecommerce |
| Fault | Damaged; defective; wrong item sent | Packaging, supplier quality, pick accuracy | Operations |
| Choice | Ordered more than one size; changed my mind; found it cheaper | Size guidance for bracketers; price consistency | Ecommerce |
| Late | Arrived too late | Delivery promise, carrier | Operations |
Keep “other” below 10% of reasons. If it’s higher, the list is missing something your customers keep typing.
Ordering several sizes to keep one, known as bracketing, means the customer didn’t trust your sizing. In the 2024 NRF and Happy Returns survey, 51% of Gen Z shoppers said they bracket Reported. Better size guidance reduces it more than any fee.
Revolve, the online fashion retailer, sells in a category where returns are a large share of sales. For the fourth quarter of 2024, management said the return rate fell by more than two points year over year, credited size and fit initiatives and AI, and said lower returns helped cut selling and distribution costs to 16.5% of net sales, down 129 basis points Reported. The work isn’t described in detail, but a public company told investors that return prevention showed up in the P&L.
The best return policy is a product page that tells the truth.
Compare each product with itself before and after, and remember that promotions and seasons move return rates too.
A small group of customers causes most of the cost. Find them by rule, set the line with your own margin, warn before you charge, and leave everyone else alone.
When returns get expensive, the tempting fix is a rule for everyone. That charges your best customers for the behavior of a few. The better fix is to find the few.
The industry numbers come from companies that sell fraud prevention, so treat them as direction. Appriss Retail estimates that of $706 billion in US returns in 2025, $100 billion was preventable loss from fraud and abuse, with abuse costing about six times as much as fraud Reported. Narvar, another returns vendor, says more than 90% of return fraud and abuse comes from less than 3% of shoppers, without publishing its method Reported. The shape is consistent: the cost concentrates in a small group.
Draw it with your own margin, not a competitor’s rule. A customer earns you margin on what they keep and costs you shipping on everything, plus handling and write-offs on what comes back. Above some return rate, every order they place loses money. The tool finds that rate, then asks what friction above it would do.
With the defaults, an order loses money once a customer sends back more than about 66% of orders. The 400 customers at 75% each cost $51.90 a year. A $5 fee on their returns, even if it loses 30% of their orders, turns them into $54.35 a year each: about $42,500 more across the group. Now set both return rates to 40%, a heavy but profitable customer the fee doesn’t change: the same friction costs about $14,150 a year. That’s the case against blanket fees in one number.
ASOS, the UK online fashion retailer, shows a targeted rule in public. From October 2024 it began deducting £3.95 from refunds for customers with a frequently high return rate who keep less than £40 of an order, while most customers kept free returns Reported. Its fair use policy sets the line at a return rate of 70% or more by value over twelve months, with at least three orders; guest orders always pay; and customers whose rate falls go back to free returns automatically Reported. In January 2026 it began showing each customer their own return rate in the app Reported.
The design has the parts that matter: a line set by data, a rule the customer can see, and a route back. UK law keeps faulty items out of it regardless (chapter 13).
Warn before you charge, charge before you ban, and ban only with evidence.
For wardrobing in occasionwear, a large tag that must be attached for a return to be accepted does more than any customer rule. Whatever you automate, have counsel check it against privacy law; automated decisions about individuals are regulated in the EU and UK.
Most large retailers now charge for some returns. A DTC brand that copies them is copying a company with stores. Test a fee against a holdout before it goes to everyone.
Between 2024 and 2026, return fees became normal among large US retailers. Before you follow, look at what they have that you probably don’t, and at what they said happened next.
In the NRF and Happy Returns survey of large retailers, 72% charged for at least some returns in 2025, up from 66% in 2024 Reported. Loop says 65.2% of its Shopify merchants charge on some return outcomes, averaging $9.04 Reported. Coverage of the 2025 holidays listed these mail-return fees:
| Retailer | Fee, as reported in December 2025 |
|---|---|
| TJ Maxx and Marshalls | $11.99 per mailed return |
| Macy’s | $9.99 per mailed return, waived for loyalty members |
| Dillard’s | $9.95 label fee |
| JCPenney | $8 per mailed return |
| J.Crew | $7.50 per mailed return |
| Abercrombie & Fitch | $7 on mailed returns |
| Urban Outfitters | $5 on most mailed returns |
| Zara | $4.95 on mailed returns |
| H&M | $3.99 on mailed returns |
ReportedCBS News, December 26, 2025; TheStreet, December 21, 2025. Fees change; check each retailer’s current policy before citing it.
Note the word that repeats: mailed. Every retailer on the list has stores, and the fee falls on mailed returns, which steers customers to return in person, where they may buy something else. Amazon’s 2023 fee made the logic explicit: $1 for returning at a UPS Store when a Whole Foods, Amazon Fresh or Kohl’s drop-off was closer Reported. A DTC brand without stores that adds a mail fee isn’t steering anyone. It’s charging everyone.
In the same 2025 survey, of retailers that began charging, 47% saw more complaints, 37% said they lost customers, 34% saw average order value fall and 24% saw sales fall Reported. A year earlier, 54% said fees had cut return rates Reported. Both can be true. The question is whether a fee cuts returns more cheaply than it cuts customers, and Bower and Maxham’s findings in chapter 3 are the reason to worry.
For a century, L.L.Bean took back almost anything, at any age. In February 2018 it moved to a one-year limit with proof of purchase, while still covering manufacturing defects after that. The company said it had lost $250 million over five years on returned items that had to be destroyed, and that such returns had doubled over that period Reported. Its chairman said some customers had come to treat the guarantee as “a lifetime product replacement program, expecting refunds for heavily worn products used over many years” Reported. Within days a Chicago customer sued, seeking class-action status Reported.
The lessons carry to a small brand. The change targeted a specific abuse, the generous core survived (a year is still long), and the company explained itself with numbers. The lawsuit is the last lesson: a policy that has been part of the brand for years is part of the price customers think they paid, and taking it away feels like a price rise.
A fee cuts returns. The only question is whether it cuts them faster than it cuts customers.
Different customers seeing different terms is a pricing test, with legal and reputational risk. Disclose each customer’s terms before they buy, never change terms after purchase, and have counsel review the design. If a customer-level test isn’t possible, use a before-and-after with a comparison group the change didn’t touch, such as another country you ship to.
In the EU and UK, customers have a legal right to send most online purchases back. In the US, the law mostly asks that you do what you said. Either way, your policy can be more generous than the law, never less.
This chapter is a map, not advice, as the rules stood in September 2026. Have counsel review your policy for every country you sell to, and again whenever you change it.
The Consumer Rights Directive (2011/83/EU) gives consumers in the EU these rights for most goods bought online, with exceptions such as custom-made and perishable goods (Article 16) Published:
So in the EU, credit or an exchange instead of a refund must be the customer’s choice, and a return fee must be disclosed before purchase. And since 19 June 2026, Directive (EU) 2023/2673 requires a withdrawal function for contracts made through a website or app: a prominent “withdraw from contract here” function or an unambiguous equivalent, a confirmation step labeled “confirm withdrawal” or an equivalent, and an acknowledgment by email or another durable medium (new Article 11a) Published. Each member state applies it through its own law. Your portal can host it, but it must work as a withdrawal, not a funnel into exchanges.
The Consumer Contracts Regulations 2013 give a similar fourteen-day cancellation period from the day the goods arrive Published. Rights for faulty goods are separate and stronger; as Which? noted, ASOS’s return-rate fee can’t be applied to faulty or misdescribed items Reported.
The law is the floor. The policy you advertise is a promise on top of it, and the portal is where you keep it.
Ten numbers, reviewed monthly, that tell you whether returns are keeping customers or quietly losing them.
Most returns reports show the return rate and a cost. The scorecard below follows the trip from the promise to the next order, so a problem shows up where it starts.
| Number | How to count it | Good direction | Warning sign |
|---|---|---|---|
| 1. Return rate, by value | Refunded and exchanged value ÷ gross sales, by category | Stable or falling, for the right reasons | Falling after a fee, with repeat falling too |
| 2. Top-20 product return rates | Each product against its own last quarter | Falling after fixes | The same product on the list three months running |
| 3. “Other” as a reason | Share of returns with a vague or blank reason | Under 10% | Rising: the list is missing something |
| 4. Outcome mix | Exchange, credit and refund shares of returns | Exchange share rising | Refund share rising after a portal change |
| 5. Second-return rate on exchanges | Exchanges returned again ÷ exchanges | Low and falling | Rising: size guidance is off |
| 6. Days, request to first scan | Median | Short | Rising: sending back has become harder |
| 7. Days, first scan to refund | Median, and share refunded at scan | Same day for most | Any rise without a decision behind it |
| 8. Cost per return | Label, handling and write-off ÷ returns | Falling | Falling while metric 10 falls |
| 9. Return tickets per 100 returns | Support contacts about returns | Falling | Rising after any change to the portal or policy |
| 10. Repeat rate by outcome | 90-day and twelve-month repeat, split by how the first return ended | Returners close to non-returners | A widening gap between refund and exchange |
Quarterly, add the count of customers above your serial-returner line and any holdout results.
The fourth number is the one that moves fastest when you change the portal, so chart it. Picture a brand that reorders its outcome screen and adds a small bonus for exchanges:
In the chapter 4 example, that shift was worth about $3,500 a month after paying the bonus. The chart shows the change happened; only the holdout shows it paid.
Report the outcome mix next to the return rate, every month. One tells you the cost; the other tells you what you kept.
Week by week, from not knowing what a return costs to running returns as a retention system with a test underway.
Most of this guide can be done in a month without new software. The order matters: measure first, fix the trip second, deal with the few third, and test the expensive ideas last.
Measure the next order, fix the trip, charge only the few, and test anything that costs money.
At day thirty you should have a scorecard, one set of terms, a portal that works on a phone, refunds that move at the scan, a weekly defect review, a serial-returner rule and at least one holdout running. Most brands never find out what their returns policy does to the next order. You will.
What whoever owns returns needs on the first day.
Whoever owns returns, a new hire, an agency, or you on the Monday you decide returns are a retention problem, needs six things on day one.
The papers and books worth reading next, and what to take from each.
And the research: Petersen and Kumar (2009, 2015); Minnema and colleagues (2016) on reviews and returns; Shehu, Papies and Neslin (2020) on free shipping; Torkaman and colleagues (2026) on exchanges. Full references are in Appendix C.
Andrew Lauchner runs Growth Legend, embedding inside consumer brands to own lifecycle, email and SMS, and revenue operations. He is the author of The Second Order, on turning first-time buyers into second-time buyers, and The Whole Machine, on the fundamentals of DTC growth, along with a series of field guides for DTC operators at andrewlauchner.com.
As Senior Director of Growth and Retention Marketing at Gallery Furniture, he rebuilt the customer journey and the sales playbooks together. He has worked on growth and retention at Binance and 3Commas, and has been Head of Growth and Retention at Greatness Wins and at Nexus Agriscience.
“Andrew led retention, lifecycle, and email/SMS, but what separates him from most in this space is how deeply he understands the role retention plays in the overall growth engine.”
Akram Khan, Head of Marketing at Gallery Furniture, senior to Andrew but didn’t manage Andrew directly
Andrew answers every note from operators working on this, including those looking for someone to own it. Write to andrew@growthlegend.com or message him on LinkedIn.
The formulas behind the three calculators, and four queries every brand should be able to run.
| For | Formula | Notes |
|---|---|---|
| Cost of a refund | H + (1 − r) × C | H: cost to bring back and process. r: share resold at full price. C: product cost, P × (1 − m). |
| Value of an exchange over a refund | −F + (1 − q)(P − C − B(1 − m)) − q(H + (1 − r)(C + B(1 − m))) | F: replacement shipping. q: share of exchanges returned again. B: bonus credit, costed at product cost. |
| Break-even exchange share for a bonus | e₀ × D(0) / D(B) | e₀: exchange share without the bonus. D: the value above, with and without the bonus. |
| Repeat lift a change must buy | cost / (first-time customers × contribution per repeat customer) | Cost: orders × return rate × extra cost per return, plus orders × extra return rate × cost per extra return. |
| Holdout size per group | 16 × p(1 − p) / d² | p: current repeat rate. d: lift in absolute terms. 80% power, 5% two-sided. |
| Serial-returner line | (a × m − F) / (a × m + h + a(1 − m)w) | a: order value. F: shipping and fulfillment per order. h: cost per return. w: share of returns written off. |
-- value returned / value sold, by product, last 12 months
-- order_lines: order_id, line_id, product_id, quantity, price
-- refund_lines: refund_id, line_id, quantity, amount, created_at
WITH refunded AS (
SELECT line_id, SUM(amount) AS amount FROM refund_lines GROUP BY line_id
)
SELECT ol.product_id,
SUM(ol.quantity * ol.price) AS sold_value,
COALESCE(SUM(rl.amount), 0) AS refunded_value,
COALESCE(SUM(rl.amount), 0)
/ NULLIF(SUM(ol.quantity * ol.price), 0) AS return_rate
FROM order_lines ol
JOIN orders o ON o.order_id = ol.order_id
LEFT JOIN refunded rl ON rl.line_id = ol.line_id
WHERE o.created_at >= CURRENT_DATE - INTERVAL '12 months'
GROUP BY ol.product_id
ORDER BY refunded_value DESC;
Refunds miss exchanges; add them from your returns app’s export.
-- first-time customers 12 to 24 months ago, split by first-return outcome
-- returns: return_id, order_id, customer_id, requested_at, outcome
-- ('refund','exchange','credit','keep'), first_scan_at, refunded_at
WITH firsts AS (
SELECT customer_id, MIN(created_at) AS first_at
FROM orders GROUP BY customer_id
HAVING MIN(created_at) BETWEEN CURRENT_DATE - INTERVAL '24 months'
AND CURRENT_DATE - INTERVAL '12 months'
),
first_return AS (
SELECT DISTINCT ON (r.customer_id) r.customer_id, r.outcome
FROM returns r JOIN firsts f ON f.customer_id = r.customer_id
WHERE r.requested_at < f.first_at + INTERVAL '90 days'
ORDER BY r.customer_id, r.requested_at
)
SELECT COALESCE(fr.outcome, 'no return') AS group_,
COUNT(*) AS customers,
AVG(CASE WHEN EXISTS (
SELECT 1 FROM orders o2
WHERE o2.customer_id = f.customer_id
AND o2.created_at > f.first_at
AND o2.created_at <= f.first_at + INTERVAL '12 months')
THEN 1.0 ELSE 0 END) AS repeat_12m
FROM firsts f
LEFT JOIN first_return fr ON fr.customer_id = f.customer_id
GROUP BY 1;
Some returns apps create an order for each exchange. Exclude those from the repeat count, or every exchanger looks like a repeat buyer.
-- median days from request to first scan, and first scan to refund
SELECT date_trunc('month', requested_at) AS month,
percentile_cont(0.5) WITHIN GROUP
(ORDER BY EXTRACT(EPOCH FROM first_scan_at - requested_at)/86400) AS days_to_scan,
percentile_cont(0.5) WITHIN GROUP
(ORDER BY EXTRACT(EPOCH FROM refunded_at - first_scan_at)/86400) AS days_scan_to_refund,
AVG(CASE WHEN refunded_at <= first_scan_at + INTERVAL '1 day'
THEN 1.0 ELSE 0 END) AS share_refunded_at_scan
FROM returns
WHERE outcome = 'refund' AND first_scan_at IS NOT NULL
GROUP BY 1 ORDER BY 1;
-- each customer's share of orders returned, last 12 months
-- set :line from the chapter 11 tool, e.g. 0.66
SELECT o.customer_id,
COUNT(DISTINCT o.order_id) AS orders,
COUNT(DISTINCT r.order_id) AS orders_returned,
COUNT(DISTINCT r.order_id)::numeric
/ COUNT(DISTINCT o.order_id) AS return_share
FROM orders o
LEFT JOIN returns r ON r.order_id = o.order_id
WHERE o.created_at >= CURRENT_DATE - INTERVAL '12 months'
GROUP BY o.customer_id
HAVING COUNT(DISTINCT o.order_id) >= 3
AND COUNT(DISTINCT r.order_id)::numeric / COUNT(DISTINCT o.order_id) >= :line
ORDER BY return_share DESC;
The three-order minimum keeps one unlucky first order from flagging a new customer. The syntax throughout is Postgres.
Policy, portal copy, messages and a test brief. Have counsel review the policy.
RETURNS AND EXCHANGES
HOW LONG You have [60] days from delivery to start a return
or exchange.
HOW MUCH Returns are free. We refund the full price to your
original payment method. [If there is any fee, state
it here, in dollars, and when it applies.]
HOW Start from the link in your delivery email, or at
[site]/returns with your email address.
WHAT Items must be unworn and unwashed, with tags. We can't
take back [final-sale or personalized items]. Faulty,
damaged or wrong items are always on us, whatever the
date.
SWAP Want a different size or color? Choose an exchange and
we'll ship it as soon as your parcel is scanned.
MONEY Refunds are sent when the carrier first scans your
parcel for most orders, and within [5] days of arrival
for the rest.
EU AND UK You also have the legal right to withdraw within 14
days of delivery without giving a reason. [Link to the
withdrawal function.]
HEADING How would you like to sort this out?
OPTION 1 Swap for a different size or color
Ships as soon as your return is scanned. No charge.
OPTION 2 Store credit for something else [+ $10 bonus]
Shop now and use it right away.
OPTION 3 Refund to your original payment method
Sent when the carrier scans your parcel.
RULES All three visible, same size, no pre-selection of
credit, no extra click to reveal the refund.
SUBJECT Your Return Is Started PREVIEW here's your drop-off code and what happens next Hi [first name], Your return for [item] is set up. Take it to [drop-off] by [date] and show this code: [code]. No box or label needed. When they scan it, we'll send your [refund / exchange / credit] the same day. Questions? Just reply. [Brand] [Postal address] This is a service message about your order. Unsubscribe from marketing emails: [link]
SUBJECT Your Refund Is on Its Way PREVIEW $[amount] back to your [card ending 1234] Hi [first name], Your parcel was scanned at [time, place], and we've sent your refund of $[amount] to your [card ending 1234]. Banks usually show it within [3 to 5] business days. You told us the [item] [ran small]. Thank you: we've passed that to the team that writes our size guide. [Brand] [Postal address] This is a service message about your order. Unsubscribe from marketing emails: [link]
[Brand]: Your return was scanned and your $[amount] refund is
on its way to your card ending 1234. Questions? Just reply. Reply
STOP to opt out.
SEND ONLY To customers who agreed to texts about their orders.
Straight apostrophes only.
SUBJECT A Quick Note About Your Returns PREVIEW nothing has changed yet, here's what to know Hi [first name], You've returned [7 of your last 9] orders with us in the past year. That's completely allowed, and we're glad you shop with us. We want to be upfront: when a customer returns more than [70%] of their orders over a year, returns stop being free, and a [$5] fee is taken from each refund. Faulty or wrong items are always free to return. Your current rate is [78%], and you can see it any time here: [link]. If sizing is the problem, reply and we'll help you pick. [Brand] [Postal address] This is a service message about your account. Unsubscribe from marketing emails: [link]
CHANGE e.g. $10 exchange bonus / $6 mail fee
PURPOSE steer / deter / retain
ARMS policy vs holdout, assigned by customer at
first order; share in holdout: %
DISCLOSURE where each arm sees its terms before buying
PRIMARY METRIC 12-month contribution per customer
GUARDRAILS first-order conversion / return rate /
exchange share / tickets / chargebacks
BREAK-EVEN from chapter 4 or 5 tool:
SIZE customers per arm: read at 90 days
and 12 months
DECISION RULE we keep it if:
COUNSEL REVIEW name, date:
Every external source, by chapter. Web sources were read in September 2026.