Return Rate: How to Tell If Yours Is Normal or an Alarm

“I have 12% returns — is that bad?” It’s one of the most frequent questions among sellers, and it has no direct answer, because 12% can be excellent or it can be an emergency depending on what you sell, on which channel, and against what you compare it. Worse: very often that 12% is calculated wrong, and the entire conversation gets built on a number that doesn’t mean what its owner thinks.

Return rate is one of those metrics everyone believes they understand and almost nobody defines the same way. Some calculate it on units, others on orders, others on pesos. Some measure it in the month the return arrived, others in the month of the original sale. Each of those combinations produces a different number for the same business, and the differences aren’t decimals: they can be double.

This article lands three things: how to calculate it without fooling yourself, what ranges the industry reports by category, and why the external benchmark matters far less than your own history.

first, calculate it correctly

There are three decisions to make before the number means anything, and they’re worth making deliberately and writing down, because comparison only works when it’s against itself.

Units, orders or pesos. The unit rate tells you how often a product comes back: that’s the one for diagnosing the product. The peso rate tells you how much revenue reversed: that’s the one for financial health. They can diverge a lot — if what comes back is your expensive product, the peso rate will be far above the unit rate, and that gap is itself a finding.

The time lag. A return arriving in March almost always corresponds to a February or January sale. If you divide March’s returns by March’s sales, in a growing business you are systematically understating your rate, because the denominator grew while the numerator still comes from the past. In a shrinking business the opposite happens. The correct approach is by cohort: returns from February’s sales, divided by February’s sales, after waiting out the full return-policy window.

What counts as a return. Do pre-shipment cancellations count? Partial refunds for damaged goods with no physical return? Units the marketplace refunded to the buyer but never sent back to you? Each tells a different story and mixing them produces a blurry number.

My practical recommendation: track the rate by units and by cohort to diagnose products, and the rate in pesos for the impact on your results. Two numbers, defined in writing, measured the same way every month.

what ranges the industry reports

Here’s the part to read carefully, because public return benchmarks carry a serious limitation: nearly all of them come from US and European ecommerce, measured mostly on own-brand stores rather than Mexican marketplaces. They work as an order of magnitude and as a reference for the distance between categories; they don’t work as a verdict on your operation.

With that caveat, public 2026 measurements put the overall ecommerce average around 19% to 20% of orders, with enormous dispersion by category:

• Apparel: 20% to 40%. The most punished category, and the dominant reason is sizing. On top of that comes bracketing: the buyer who orders three sizes knowing in advance they’ll return two.
• Footwear: 17% to 30%. Same sizing problem, aggravated by variation between different brands’ lasts.
• Electronics: 8% to 15%. Here returns come not from sizing but from buyer’s remorse and genuine defects. It’s a different problem, attacked differently.
• Beauty: 4% to 12%. The lowest, for obvious hygiene and policy reasons.

There’s also a consistent difference by channel type: the same measurements put marketplace above pure D2C — around 19% against 15% within the same product cohort. It makes sense: the marketplace return policy is more generous and more visible, the buyer trusts the platform more than they trust you, and the friction to return is minimal.

That last point matters for the Mexican seller: if you measure your MercadoLibre rate against an own-store benchmark, you’ll believe you have a problem when what you have is a policy.

Glossary: real net margin, with everything deducted →

why your own baseline is worth more

The external benchmark answers a relatively useless question: “am I normal?” The useful question is “am I getting worse, and where?” And only your own history answers that.

The reason is that your rate is determined by variables no industry average captures: your exact category mix, your shipping policy, your photography and listing quality, your supplier, your packaging, your buyer profile and how aggressive your advertising is. Two apparel sellers on MercadoLibre can sit at 18% and 34% with nominally the same catalog, and the entire difference can live in the size chart.

That’s why the useful reading has three layers:

Your overall rate, month over month. Not to compare yourself with anyone, but to catch the trend. Two consecutive months rising is a signal, even if the absolute level looks fine.

Your rate per SKU, against your own median. This is where the actionable findings appear. In nearly every catalog a handful of SKUs concentrate a disproportionate share of returns, and they’re usually identifiable and fixable: a size that runs small, a photo that misleads, an incomplete description, packaging that doesn’t survive transit.

Your rate per channel, for the same SKU. If a product is returned at 8% on Amazon and 19% on MercadoLibre, the product isn’t the problem. It’s the listing, the policy, the expectation you set, or the kind of traffic you’re buying.

when it’s genuinely an alarm

Three patterns warrant immediate intervention regardless of the absolute level:

A sudden jump on a stable SKU. If a product that sat at 6% for months goes to 15% in a single month, something concrete changed: a supplier batch, an edit to the listing, a competitor who copied your content, a packaging change. It’s traceable and usually fixable.

A rate that climbs while you scale advertising. That’s the symptom of buying progressively less qualified traffic. Advertising that pushes to a cold audience generates buyers with worse-formed expectations, and that comes back as returns. If ACoS improved but returns rose, your campaign may be destroying margin even though the ads report looks good.

A rate that eats your contribution. This is the only threshold that truly matters and it’s different for every business. If your contribution margin per order is $118 and a full return costs you $310, you need fewer than one order in 2.6 to be returned just to break even on that product. Framed that way, “is 12% bad?” answers itself: it depends on how much each good order leaves you.

the three levers that actually lower the rate

Measuring is fine, but the next question is always what to do. In practice, nearly all avoidable returns are attacked with three levers, in this order of return on effort.

The product listing. The cheapest and the most ignored. Most “it wasn’t what I expected” returns are an expectation problem, not a product problem: photos that don’t convey real scale, a description that omits the material, a generic size chart copied from the supplier. Fixing the listing of the SKU that concentrates your returns usually costs an afternoon and moves several points.

Packaging and supplier. When a return is flagged as damaged or defective, the problem sits upstream of the buyer. It’s worth separating those returns from the rest and tracing them by batch: very often they cluster in one specific shipment, and that information is what lets you negotiate with the supplier instead of absorbing the cost in silence.

Traffic quality. If your rate rises when you scale ads, the remedy isn’t in the product but in the targeting. Overly broad search terms bring buyers who weren’t looking for exactly what you sell, and that mismatch returns as a refund several weeks later, when nobody associates it with the campaign anymore.

how it looks in iqseller

In the Profitability module, returns aren’t a separate report: they enter as a provision inside each SKU’s contribution calculation. That is, that product’s historical return rate, multiplied by what a full return costs — refund, commission already paid, outbound freight, return freight and the unit that sometimes comes back unsellable — is deducted before saying whether that product makes money.

That’s the substantive change. A return rate seen alone is an operations data point; seen inside the margin it’s a catalog decision. The product that looked profitable at 22% gross margin and is returned at 19% probably isn’t profitable, and that’s only visible when both numbers live on the same screen.

And because the panel takes data from each channel in the same format, the per-SKU comparison between Amazon and MercadoLibre — the one that reveals whether the problem is the product or the listing — comes out without reconstructing anything by hand.

Glossary: inventory valuation, how much capital is tied up →

the short answer

Is 12% bad? If you sell electronics, probably yes and it’s worth investigating. If you sell apparel on MercadoLibre, it’s a good number. If last month you were at 7%, it’s an alarm regardless of category. And if your contribution per order is thin, it can be unsustainable even when your industry benchmark says you’re fine.

The industry benchmark is good for one thing: knowing whether you’re in the right stadium. Everything else — where the problem is, how fast it’s growing, and whether your margin can absorb it — comes from your own data, measured the same way every month and placed next to what each sale actually leaves.