Temu built its US and European growth on a promise that reads very simply to a shopper: if the item is wrong, you get your money back, and you probably do not have to fight for it. That promise is a marketing asset for the platform. For the merchant behind the listing, it is a cost line, and the size of that line depends almost entirely on which selling model you are in.
This piece walks through how refunds actually flow on the platform, how the cost splits between Temu and the seller under the fully managed and local seller models, what return rates look like in the low price segment, and how to turn all of that into a provision you can price for rather than a surprise you absorb.
In short
- The buyer experience is deliberately frictionless. Temu advertises a long return window and, for most orders, a free first return, which means disputes rarely reach the seller as an argument.
- Your model decides your exposure. Under fully managed, Temu owns most of the reverse flow and absorbs much of the handling. Under local-to-local and semi managed, the operational and shipping cost sits much closer to you.
- Return rates are category driven, not platform driven. Apparel, footwear and anything sized runs far above hard goods, and the gap is wide enough to change which products are worth listing.
- Refund without return is common and often rational. Below roughly the $10 to $15 order band, the reverse shipping and inspection cost usually exceeds the recovery value of the unit.
- Price the provision in, do not discover it. A returns reserve expressed as a percentage of gross revenue belongs in your cost of sale next to commission and freight, and it needs a category specific number.
How the buyer refund experience works
Temu’s public positioning on returns has been consistently aggressive since its US launch. As of the time of writing, the platform advertises a return window measured in months rather than weeks for most categories, and a free first return per order, with subsequent returns in the same order potentially carrying a shipping charge. Those terms have changed more than once, and they differ by market, so treat any specific number you read anywhere, including here, as something to confirm on Temu’s own policy pages before you build a pricing model on it.
The mechanics from the shopper’s side are intentionally shallow. A buyer opens the order in the app, selects a reason, and in a large share of cases receives either an immediate refund decision or a prepaid label. There is no seller approval step in the way an experienced eBay or Amazon third party merchant would recognise it. The platform decides.
Why the platform resolves rather than arbitrates
This design is not generosity for its own sake. Low average order values make dispute handling uneconomic: a human reviewing a $7 claim costs more than the claim. Automating approval is cheaper than adjudicating, and it buys conversion on a platform where the main objection is trust in unfamiliar sellers.
The consequence for merchants is that you lose the right to contest individual cases in most situations. What you retain is influence over the inputs: product accuracy, sizing data, packaging and photography. That shift, from arguing about outcomes to engineering out causes, is the single most useful mental adjustment when moving onto the platform. It is part of the broader pattern we traced in how Temu is rewriting the rules of low-price e-commerce, where platform control replaces seller discretion almost everywhere.
What the buyer sees versus what gets logged
A refund in the app can mean several operationally different things. It can mean the unit comes back to a Temu facility. It can mean the unit comes back to you. It can mean nobody collects anything and the stock is written off in place. All three look identical to the shopper and very different to your P&L.
If your reporting does not separate those three cases, you cannot calculate a true return rate, and you will tend to overestimate recoverable inventory. Pull the reason codes and the disposition codes separately, and reconcile them to actual received units at your warehouse rather than trusting the refund count alone.
Fully managed versus local seller responsibility
Temu runs more than one merchant model, and the labels matter because the returns economics diverge sharply between them. Anyone evaluating the platform should settle this question before modelling anything else, which is where our walkthrough on selling on Temu as a US merchant starts as well.
Fully managed
Under the fully managed arrangement, you ship inventory into a Temu designated warehouse and the platform takes over pricing, listing, marketing, outbound shipping and customer service. Returns land back at the platform’s facility. Temu handles the inbound, the inspection and the decision on whether the unit is resaleable.
Your exposure here is mostly indirect. You carry the inventory risk on units that come back damaged or get written off, and you carry the revenue loss on the refunded sale, but you are not paying per-parcel reverse freight or staffing a returns desk. For small sellers this is the attraction: the platform absorbs operational complexity in exchange for control over price.
Local-to-local and semi managed
Under local-to-local, you hold stock in the destination market and fulfil orders yourself, which gives faster delivery and makes you eligible for categories that cross-border shipping cannot serve well. It also puts the reverse flow back on your side of the line. You receive returns, you inspect them, you decide disposition, and in many cases you are paying for the label or absorbing the cost of a return-free refund.
Semi managed sits in between and the exact split has shifted as the programme has evolved by market. The practical rule is to read your own agreement rather than a general description, because this is the clause that determines whether your returns provision is two percent or twelve.
Who touches what
| Step in the reverse flow | Fully managed | Local-to-local / semi managed |
|---|---|---|
| Refund decision | Temu, automated | Temu, automated |
| Return label issued to buyer | Temu | Usually Temu, cost often passed on |
| Physical receipt of the unit | Temu warehouse | Seller or seller’s 3PL |
| Inspection and grading | Temu | Seller |
| Restock or dispose decision | Temu, per policy | Seller |
| Reverse freight cost | Largely absorbed by platform | Largely seller, per agreement |
| Disposal or liquidation cost | Platform process | Seller arranges |
| Inventory write-off | Seller | Seller |
| Practical control over causes | Low (platform owns listing) | Higher (seller owns more of the listing) |
Read that table as a trade. Fully managed converts a variable operational cost into a loss of control over the very inputs that drive returns. Local-to-local gives you the inputs back and hands you the bill. Neither is strictly better, and the right answer tends to follow category: sized and fit-sensitive goods argue for control, commodity hard goods argue for outsourcing the hassle. The wider version of this decision, across every major marketplace, is laid out in the complete guide to selling on global e-commerce marketplaces.
Return rates by category on low price platforms
There is no audited public figure for Temu’s return rate, and any specific percentage attributed to the platform in press coverage should be read as an estimate rather than a disclosure. What does exist is reliable category level data for US retail generally, and it is a reasonable starting point as long as you adjust for the things that make this segment different.
What the published benchmarks measure
The National Retail Federation publishes an annual returns study with Appriss Retail that has put total US retail returns in the mid-teens as a share of sales in recent editions, with online returns running materially above the store average. That study covers all of retail, so it blends furniture, grocery and electronics into one number. Check the current edition directly rather than citing a figure from memory, because the headline moves each year.
Two adjustments matter for low price cross-border goods. Returns tend to run lower in absolute percentage terms on very cheap impulse purchases, because the shopper’s effort threshold for initiating a return is high relative to the refund. At the same time, dissatisfaction runs higher, because expectation gaps are wider. The result is a high complaint rate paired with a return rate that can look deceptively modest, and a large share of resolutions that never involve a physical return.
Indicative bands by category
The table below is a planning aid built from general retail patterns, not a Temu disclosure. Use it to decide which categories need a thicker provision, then replace every number with your own observed rate as soon as you have ninety days of data.
| Category | Indicative return rate band | Dominant reason | Resale likelihood after return |
|---|---|---|---|
| Apparel (sized) | High | Fit and sizing | Moderate if unworn and sealed |
| Footwear | High | Fit, width, box damage | Low once the box is opened |
| Jewellery and accessories | Medium | Appearance versus photo | Moderate |
| Consumer electronics and accessories | Medium | Compatibility, failure | Low, testing cost is high |
| Home and kitchen hard goods | Low to medium | Size or quality expectation | Moderate to high |
| Tools and auto parts | Low to medium | Fitment | High if packaging intact |
| Toys and novelty | Low | Damage in transit | Low, packaging is the product |
| Beauty and personal care | Low | Reaction, expectation | None, hygiene rules apply |
The last column is the one most sellers skip, and it is where the money hides. A high return rate on a product that comes back fully resaleable is an annoyance. A medium return rate on a product that is worthless once opened is a margin problem, because every return is a total loss of unit cost plus the freight you already spent.
Why the comparison with Shein is instructive
Shein’s catalogue is weighted heavily toward apparel, which is the single highest return category in retail, while Temu’s mix spans a much broader set of hard goods. That difference alone explains a lot of the divergence in how the two platforms handle after-sales, as we set out in the comparison of Temu versus Shein for shoppers and for competing retailers. If you sell into both, do not carry one returns assumption across both catalogues.
Who pays return shipping and disposal
The honest answer is that the label cost and the disposal cost are two separate questions, and sellers routinely budget for the first while forgetting the second.
On the label: under fully managed, reverse freight within the destination market is generally part of what the platform absorbs as the cost of its own refund promise. Under local-to-local, the commercial terms vary, and in several markets the pattern has been that the platform issues the label for buyer convenience and recovers the cost from the seller account. Your settlement report is the authority here, not any general description. Reconcile a month of deductions line by line at least once so you know which code represents reverse freight.
Disposal is the cost nobody models
A returned unit has four possible fates: restock as new, restock as graded, liquidate in bulk, or dispose. Each has a different cost and only the first recovers full value. For low value goods, the inspection labour alone can exceed the unit’s wholesale cost, which pushes a large share of returns straight to liquidation or disposal regardless of condition.
If you are running local-to-local at any volume, this is the point at which the reverse logistics market becomes relevant to you rather than academic. The specialist processing sector has been consolidating steadily, a trend we examined when looking at European returns platform consolidation, and the practical upshot is that outsourcing grading and liquidation is now a realistic option well below the volumes it used to require.
The duty question on cross-border returns
When a unit was imported and is then refunded, two separate things can happen to the duty you paid. Nothing, which is the default, or a recovery through a formal customs process. In the United States, US Customs and Border Protection administers drawback, which allows duty refunds in defined circumstances including certain unused merchandise exported again. The administrative threshold is real: drawback is paperwork-heavy and rarely economic on individual low value parcels, though it can be worth examining at aggregate volume.
The rules governing low value imports in both the US and the EU have moved repeatedly over the past two years, and anything written about de minimis thresholds or parcel handling fees dates quickly. Verify the current position with CBP, the Federal Register or the European Commission before assuming your landed cost model still holds.
Refund without return: when it is cheaper for everyone
A refund with no physical return, sometimes called returnless refund or keep-it, is now standard practice across most large marketplaces, and it is heavily used in the low price segment for the obvious reason that the arithmetic demands it.
The arithmetic of a $9 order
Take a unit that retails at $9, costs you $2.60 landed, and generated roughly $1.80 of outbound fulfilment cost that is already spent. If the buyer initiates a return, the reverse parcel in-market costs somewhere between $4 and $8 depending on size and carrier, and inspection plus putaway adds labour on top.
Against that, the best case recovery is a unit worth $2.60 at cost, and only if it grades as new. The expected recovery is well below the cost of getting it back. Writing the unit off in place is the rational decision, and the platform’s automated systems reach it quickly.
| Scenario on a $9 order | Cost incurred | Value recovered | Net outcome |
|---|---|---|---|
| Refund, unit returned and restocked as new | Reverse freight plus inspection, roughly $5 to $9 | Unit at $2.60 cost | Negative, usually clearly so |
| Refund, unit returned and disposed | Reverse freight plus inspection plus disposal | Nil | Worst case |
| Refund with no return | Unit cost plus outbound freight already spent | Nil | Cheapest of the three |
| Partial refund, buyer keeps unit | Share of revenue only | Remaining revenue retained | Often the best available |
Where returnless policy goes wrong
The failure mode is predictable. Once buyers learn that refunds do not require a return, a small cohort optimises against it. Platforms counter with account level pattern detection rather than per-case review, which is the only approach that scales, but the leakage never reaches zero.
From the seller side you have very little control over this and should treat it as a fixed cost of the channel rather than a solvable problem. What you can control is whether your highest value SKUs sit in a channel where return-free refunds are automatic. For anything where unit cost is a meaningful share of retail price, that question deserves an explicit answer rather than a default.
Budgeting a returns provision into your price
The objective is a single number, expressed as a percentage of gross revenue, that you subtract alongside commission and freight when deciding whether a SKU clears your margin floor.
Building the number
Four inputs produce it. First, the return rate for that category, from your own data where you have it. Second, the share of returns that come back physically rather than being refunded in place. Third, the recovery rate on returned units, meaning the proportion that resell at full value. Fourth, the per-return handling cost, which under local-to-local includes freight, inspection and disposal.
Multiply the return rate by the loss per return, where loss per return is the refunded revenue plus handling cost minus recovered value. Express the result as a share of revenue. That is your provision. Most sellers who run this calculation honestly for the first time find a number between three and twelve percent depending on category, and almost always higher than the placeholder they had been carrying.
A worked example
Assume a home goods SKU, $14 retail, $4 landed cost, 6% return rate, 40% of returns physically returned, 70% of those resaleable, and $5.50 of handling per physical return. The refunded revenue is the dominant term: 6% of revenue is gone outright. The physical returns add handling on 2.4% of orders, roughly $0.13 per order on a $14 average, and recover about $0.07 of value back. The provision lands near 6.4% of gross revenue.
Now run the same shape on footwear with a 20% return rate and near-zero resale value once the box is opened and the number moves past 22%. On a category with 30% gross margin before returns, that is the difference between a viable SKU and a loss-making one. This is exactly why a blanket provision across a mixed catalogue is worse than useless: it makes the bad SKUs look acceptable by averaging them with the good ones.
Where the provision belongs
Put it in the same place as commission, payment costs and outbound freight: in the unit economics sheet that gates every listing decision. If your platform economics live only at the account level, you will keep discovering returns as a quarterly variance instead of pricing for them per SKU. The broader framework for stacking these costs, across marketplaces with very different fee structures, is covered in our marketplace selling guide.
Disputes, penalties and quality flags
Beyond the direct cost of refunds, Temu operates seller performance metrics, and after-sales behaviour feeds them. This is the part of the returns picture that can cost far more than the refunds themselves.
How performance metrics bite
The general pattern across large marketplaces, Temu included, is that a product or seller account exceeding a quality or complaint threshold faces graduated consequences: reduced traffic, listing suppression, withheld settlement, mandatory corrective action, and in serious cases account restriction. Specific thresholds, penalty amounts and appeal windows are set in the seller agreement and operational policy documents, they differ by market, and they have been revised repeatedly. Read your current terms in the seller portal rather than relying on any third party summary, including this one.
Separately, several regulators have scrutinised marketplace practices around product compliance and seller accountability. The European Commission has opened proceedings against large marketplaces under the Digital Services Act, and those are ongoing matters where allegations have been made rather than findings proven. The reason it matters commercially is that regulatory pressure on platforms tends to translate into tighter documentation and compliance demands on sellers.
What to do when a flag lands
Treat a quality flag as a listing problem first and a dispute second. The most common underlying causes are a sizing chart that does not match the goods, photography that flatters, a missing specification that drives compatibility returns, or a packaging spec that fails in the parcel network. All four are fixable, and all four reduce the flag and the refunds simultaneously.
Where you genuinely believe the data is wrong, use the formal appeal route in the seller portal and attach evidence: carrier scans, factory inspection reports, photographs of the outbound pack. Volume of complaint rarely moves an automated system. Documentation sometimes does.
The reporting discipline that pays for itself
Build one monthly view that joins refund records to reason codes, to physical receipt at your warehouse, and to the settlement deductions that correspond to them. Most sellers have all three data sets and never join them, which is why the true cost of returns stays invisible until it shows up as an unexplained gap between gross and net revenue.
Where to verify the rules that matter
This article is general information for merchants and is not legal, tax or customs advice, and it is not a substitute for Temu’s own published terms. Return windows, fee splits, penalty thresholds and duty treatment all change, they vary by market, and the version in your own seller agreement is the one that binds you. For anything with a financial or compliance consequence, consult a licensed customs broker, a trade attorney or a qualified tax advisor about your specific situation.
For the underlying rules rather than commentary, go to the primary sources. Platform terms live in the Temu seller portal and on the platform’s published policy pages. US import and drawback rules come from US Customs and Border Protection, with changes published in the Federal Register. EU customs and parcel rules come from the European Commission and the national customs authority in your market of sale. Background on the platform and its operating model is useful context but is not a policy source. Every figure in this article that is not sourced to one of those bodies is an illustrative planning number, not a disclosure.
FAQ on Temu returns
How long do buyers have to return an item on Temu?
Temu has advertised a return window measured in months for most categories, which is longer than many competing marketplaces, but the exact window differs by market and by product type and has been revised more than once. Check the current return policy page for your market of sale rather than relying on a figure quoted elsewhere.
Do sellers pay for return shipping on Temu?
It depends on your selling model and your agreement. Under fully managed, reverse freight in the destination market is generally absorbed as part of the platform’s own refund promise. Under local-to-local and semi managed arrangements, the cost is more often recovered from the seller account. Your settlement report is the definitive answer, so reconcile a month of deductions to identify the reverse freight line.
What return rate should I budget for?
Use your own observed rate per category once you have about ninety days of data. Before that, anchor on general retail benchmarks such as the National Retail Federation’s annual returns study, then adjust upward sharply for sized goods and downward for very low value hard goods. A single blanket rate across a mixed catalogue hides the SKUs that are actually losing money.
Why do some refunds happen without the item coming back?
Because on low value orders the reverse freight plus inspection cost usually exceeds what the returned unit is worth. Refunding in place is the cheaper outcome for both the platform and the seller, so automated systems reach it quickly. It also means your refund count overstates the inventory you will actually get back.
Can I contest a refund decision?
Individual case-by-case contest is limited by design, since the platform automates resolution to keep dispute costs below claim values. Formal appeal routes do exist in the seller portal for performance flags and for settlement disputes, and they respond to documentary evidence such as carrier scans and inspection reports rather than to volume of complaint.
Do returns affect my seller account standing?
Yes, indirectly. After-sales outcomes feed seller and product quality metrics, and exceeding thresholds can lead to reduced traffic, listing suppression, withheld settlement or account restriction. Thresholds and consequences are set in your seller agreement and vary by market, so read the current operational policy in the portal.
Can I recover import duty on a refunded item?
Sometimes, through a formal customs process rather than automatically. In the United States, drawback administered by US Customs and Border Protection permits duty refunds in defined circumstances, including certain unused merchandise that is exported again. The administrative burden makes it uneconomic on individual low value parcels but potentially worth examining at aggregate volume. Speak to a licensed customs broker before relying on it.
Which categories are hardest to make work given returns?
Anything sized, so footwear and fitted apparel, plus any product where the packaging is effectively part of the product and cannot be resold once opened. The problem is not only a higher return rate but a low recovery rate on returned units, which turns each return into a full loss of unit cost plus the outbound freight already spent.
Is fully managed or local-to-local better for returns cost?
Fully managed shifts most reverse operations and much of the handling cost to the platform, at the price of losing control over the listing inputs that cause returns. Local-to-local gives you that control and hands you the operational bill. Sized and fit-sensitive goods usually argue for control, commodity hard goods for outsourcing the handling.