Return Policy by Channel: Amazon and MercadoLibre Don't Cost You the Same

Almost every multichannel seller computes a single return cost and applies it flat across the catalog. It’s understandable: a return feels like a generic event, something that simply happens and has to be absorbed. But it isn’t generic. The same product, returned for the same reason, costs you different amounts depending on which channel sold it.

The difference isn’t in the price or the commission, which you already track. It’s in three decisions the platform makes and you don’t: how long the buyer has to change their mind, who pays the return freight, and who decides whether the return goes through. Those three variables combine differently on Amazon and on MercadoLibre, and the result is that your cost per return — and therefore your real margin per channel — is not the same.

When that isn’t separated in your numbers, the usual thing happens: you believe a channel is better because it sells more or because the commission looks lower, and it turns out that after returns the ranking flips. Or the opposite: you penalize a channel over a high rate that is actually cheap for you, because that’s not where you absorb the return.

This article organizes the real differences, and it starts with a warning that is meant seriously: both channels’ policies change frequently and vary by category, by fulfillment program, and by country. What follows is the conceptual frame and what public documentation indicated at the time of writing. The exact windows, percentages, and category exceptions have to be confirmed in Seller Central and in MercadoLibre’s seller help, not in an article.

the three variables that set the cost

Before comparing channels it helps to be clear about what is being compared, because the cost of a return is not a number: it’s a sum of pieces that switch on or off depending on the policy.

• The refund to the buyer. The obvious part, and the only one everybody counts.
• The channel commission. Do you get it back in full, partially, or not at all? This changes across platforms and across return reasons.
• The fulfillment fee. What FBA or Full charges to process the order, and what they charge on top to process the return.
• Outbound freight. Already spent, and almost never recovered.
• Return freight. The variable most sensitive to channel and to reason.
• The unit’s destination. Whether it comes back sellable, comes back as used at a discount, or doesn’t come back at all.
• Your team’s time. Real, even though it shows up in no report.

Of all of those, the ones that differ structurally between Amazon and MercadoLibre are the window, the return freight, and who decides. The rest vary more by category and by fulfillment model than by channel.

the window: how long you stay exposed

The window doesn’t look like a cost, but it is one in two ways.

The first is directly financial: while the window is open, the sale isn’t fully yours. If your cash cycle depends on reinvesting what you sold this week, a long window means part of that money carries a probability of reversal you can’t ignore.

The second is operational: the window defines how long a buyer can use your product before sending it back. In categories where use leaves a mark — apparel, footwear, tools, seasonal electronics — every additional week of window raises the share of units that come back unsellable.

Public documentation from Amazon Mexico for buyers indicates a general window of roughly 30 days from receipt for most products, with category exceptions and extended windows in certain seasons, plus an additional period for the buyer to actually ship the return once they have the instructions. On MercadoLibre Mexico, official help also points to a window on the order of 30 days from receipt for the buyer to open the return.

The two numbers are similar, which is why the window is rarely the big difference between channels. What matters is what comes next: what happens once the return is in motion.

Glossary: real net margin, with everything deducted →

who pays the return freight

Here there are real structural differences, and they aren’t “Amazon pays” versus “Meli pays.” They are differences in how fault gets assigned.

On Amazon, the public logic is reason-driven: returns attributable to the seller — wrong product, damaged, not as described — are charged to the seller, while buyer-remorse returns typically have the shipping cost deducted from that buyer’s refund. In parallel, Amazon has been pushing the use of prepaid return labels issued by the platform itself, which takes the choice of how the return ships away from the seller but also shortens the cycles. Which categories and which marketplaces fall under that scheme changes, and it needs to be verified in Seller Central.

On MercadoLibre, the public scheme is more buyer-centric: the platform generates the label, the buyer drops the parcel at a dispatch point, and Meli handles the transport back. The seller generally absorbs the return cost, and official help notes that when the return is confirmed, the selling fee and the absorbed shipping cost are credited back, with different treatment depending on the reason: remorse and a defective product are not resolved the same way.

The practical consequence for your costing is important and counterintuitive: on one channel the return may reach you already netted in the settlement, and on the other it may appear as a charge and a credit at different moments. If you only look at the deposit, you see neither.

who decides, and why that costs money

The third variable is the least discussed and the most frustrating: who resolves a disagreement.

On both channels the short answer is the same — the platform resolves, not you — but the path differs. On Amazon the seller has a specific mechanism to dispute a return considered improper, the SAFE-T claim, with limited day windows to file and to respond. On MercadoLibre the path is the claim inside the order and, if it isn’t resolved, mediation; official help indicates the platform runs a quality check on the returned product and may cover the seller when there is sufficient evidence of misuse or fraudulent intent.

What matters for cost isn’t so much who wins, but how much work it takes to participate. A channel where every dispute requires assembling a file and waiting for a ruling is charging you hours that appear in no commission line. That cost can be estimated, and it’s worth doing: if over a month your team spent a given number of hours on one channel’s disputes, that number divided by that channel’s returns is part of your real unit cost.

why this changes your cost per channel

Let’s put the pieces together with an invented arithmetic example, with numbers that add up.

Say a product you sell at $1,160 on both channels, with COGS of $520 and outbound freight of $95. Assume, for the example, a channel commission of $174 and a fulfillment fee of $110. Your contribution per good sale, before returns, is $1,160 − $520 − $95 − $174 − $110 = $261.

Now the return. On channel A, the return freight is deducted from the buyer’s refund and the unit comes back sellable: you lose outbound freight ($95) and the fulfillment fee ($110), and you recover commission and product. Cost per return: $205.

On channel B, you absorb the return freight ($120), the commission is credited back but the return processing fee adds $60, and one in three units comes back unsellable, which on average adds a third of COGS, $173. Cost per return: $95 + $110 + $120 + $60 + $173 = $558.

With an 8% rate on both channels, the provision per sale is $16.40 on channel A and $44.64 on channel B. Your real contribution lands at $244.60 versus $216.36. Same product, same price, and a gap of nearly 12% in what each sale leaves you — invisible if you average everything.

The point isn’t the example’s numbers, which are made up. The point is the method: cost per return has to be computed per channel, with the pieces that actually switch on in each one, and then converted into a provision per sale using that channel’s rate.

what you can do with that gap

Knowing a channel costs you more per return doesn’t mean abandoning it. It means having three levers you didn’t have before.

Adjust price per channel. If a channel carries a higher return provision, that differential is a quantified argument for a different price, not a whim. It’s the same logic you already apply with commissions and fees.

Choose which product you push where. SKUs with a high rate and an expensive return are better concentrated where the return scheme penalizes you least, while you push the ones that rarely come back on the other channel.

Prioritize fixes where they hurt. If the same SKU returns far more on one channel, that channel’s listing is the highest-return work available to you this week. Size chart, photos with scale, explicit material, included accessories listed one by one.

Glossary: unified catalog, one real product and many listings →

how this reads in iqseller

The Profitability module computes net margin per SKU taking commissions from the Amazon settlement and from MercadoLibre orders as they arrive, plus FBA and Full fees, shipping, the COGS you loaded, and the VAT and withholding breakdown. The return enters as a provision inside contribution margin, and because the provision is built from each channel’s rate and cost, the result already comes separated: it isn’t the product’s average, it’s what that product leaves on that channel.

On top of that, the per-channel view of the same SKU answers the underlying question. If the rate is even across Amazon and MercadoLibre, the product is the problem and it has to be worked at the source: supplier, packaging, specification. If it spikes on only one, the problem is that channel’s listing or policy, and the work is different.

The Pricing module closes the loop: with the per-channel provision already deducted, each channel’s profitable floor stops being the same number — which is exactly what should happen when the return rules aren’t the same.

what to verify in your own account

Before redoing your costing, it’s worth sitting down for an afternoon with both accounts open and confirming, for your specific categories:

• The current return window and its seasonal exceptions.
• Which return reasons are charged to the seller and which aren’t.
• Whether the prepaid label scheme applies and which categories it covers.
• What gets credited back when a return is confirmed, and when it shows up in your payout.
• The dispute mechanism available, its day window, and what evidence it requires.
• Whether returnless refunds exist in your categories and under what value threshold.

None of those points is administrative trivia: each one moves your cost per return, and therefore your real margin. And all of them change, so they deserve the same review cadence you give commissions.

The conclusion is simple to state and laborious to apply: a return isn’t an event, it’s a policy. And since the policy differs by channel, so does the cost. Counting it flat is the quietest way to be wrong about which channel is working for you.