Why buyers return your products on MercadoLibre: the MeLi reason vs. the real cause
The reason a buyer picks when opening a claim isn't the real cause of the return. A method to measure the rate per product, read the messages, sort causes (size, color, listing details, shipping damage, defect, buyer's remorse) and know what to do about each one.
A claim comes in. On the MercadoLibre screen it says "Different color, size or model." You handle it, accept the return, and the next week another identical one shows up for the same product. Is the problem the color? The size? The photo? A buyer who didn't read the listing? The official reason doesn't tell you, and that's why the claim keeps coming back.
In this guide, you'll learn how to separate the reason MercadoLibre records from the real cause the buyer writes, how to measure returns per product without fooling yourself with percentages, and what to do about each cause.
The MeLi reason isn't the cause
When a buyer opens a claim, MercadoLibre asks them to pick an option from a list: the product arrived damaged, it's different from the listing, parts are missing, it never arrived, they changed their mind. That option is the reason: it's official, it's on every claim, and it's useful for organizing operations.
The problem is that it's generic on purpose. "Different from the listing" lumps together issues that get fixed in opposite ways:
- The buyer expected 40 cm and the product is 40 cm, but the photo made it look bigger.
- The color in the photo isn't the real color because the shot was taken under warm light.
- The listing details say "stainless steel" and the product is painted steel.
- The wrong variation shipped because someone mixed it up in the warehouse.
Four causes, one reason. If you only look at the reason, all you can do is "improve the listing," and that's not a plan.
The real cause lives somewhere else: in the messages the buyer writes inside the claim. That's where they say "I thought it was bigger," "it looks blue in the photo and it's purple," "it came scratched," "I ordered the wrong model." That text is the information that matters, and almost nobody reads it systematically because it's a lot of messages spread across a lot of claims.
The method in three steps
1. Measure the rate per product, not per account
Your MercadoLibre reputation measures claims against your account's total sales. It's useful for knowing whether you're at risk, but it hides the detail: one product with a high rate gets buried among ten that are doing fine.
To make decisions, you need the numbers per product:
- Claim rate = claims ÷ units sold in the period.
- Return rate = units returned ÷ units sold in the period.
Two things to watch so you don't fool yourself:
- No sales, no rate. If a product had no paid sales in the period, the percentage can't be calculated. It isn't zero: it's "no data."
- Without a sample, the percentage means nothing. With 3 sales and 1 return you get "33%," which sounds like a disaster and is really just noise. Until you have a few dozen sales, look at the absolute number of claims, not the percentage.
2. Read the messages, not just the reason
For each claim, note what the buyer wrote in addition to the reason. With only a few claims, a spreadsheet is enough: product, MeLi reason, what the buyer said in one sentence. With a lot of them, it's impossible by hand, and that's exactly the kind of work where AI helps: reading the messages and grouping them into comparable categories, without making up anything the buyer didn't say.
3. Sort the causes into actionable categories
The key is that each category leads to a different action. These are the ones we use:
| Real cause | What the buyer says | What to do |
|---|---|---|
| Size | "I thought it was bigger," "it doesn't fit" | Measurements in the title and on the first photo; a photo next to a reference object (a hand, a chair). |
| Color | "it looks like a different color in the photo" | Photos under neutral light; describe the color in everyday words, not just the trade name; one photo per variation. |
| Doesn't match the listing details | "it says X and it's Y" | Check the listing details attribute by attribute against the physical product; fix it before the claim escalates. |
| Shipping damage | "it arrived broken," "the box was crushed" | Packaging: filler, a sturdier box, double boxing for fragile items. Check whether the damage clusters in one region or one shipping method. |
| Manufacturing defect | "it doesn't work," "it came defective" | Quality control before shipping; talk to your supplier; if the batch is bad, pause the listing until it's sorted out. |
| Missing parts | "the charger wasn't included" | A packing checklist; state clearly in the listing what's included and what isn't. |
| Wrong product | "they sent me a different model" | Picking process: labels per variation, a photo of the package before sealing it. |
| Late or never arrived | "it never arrived," "it took weeks" | Almost always logistics, not the product: review your handling times and shipping method. |
| Buyer's remorse | "I don't need it anymore," "I bought it by mistake" | There's nothing to fix in the product. Reduce friction: accepting the return quickly protects your reputation. |
The table turns "I have returns" into concrete tasks and, more importantly, tells you where not to spend your energy: if your main cause is buyer's remorse, changing the photos won't move the needle.
How to prioritize
Once the causes are sorted, the question is where to start. Three criteria, in this order:
- Products with a real sample and a high rate. Start with the ones that sell a lot and get returned more than the rest. That's where every point of improvement shows.
- Actionable causes. Size, color, listing details, packaging and picking are in your hands and can be fixed in days. Buyer's remorse and, partly, logistics can't.
- What repeats. A cause that shows up across several products at once is usually a process problem (how you take photos, how you pack orders), and fixing it once fixes all of them.
And after you change something, measure again. If you corrected the measurements in the listing and the "size" cause is still the same weeks later, the problem was somewhere else.
What almost nobody does: close the loop
The most common mistake isn't ignoring claims, it's resolving them one at a time and never looking at them together: each one gets handled and the information the buyer left behind gets lost. Reviewing them together once a month, by product and by cause, turns post-sale into improvements to your listing, your packaging and your supplier, and it costs less than paying for the same return over and over.
How Ficanaz does it
If you sell on MercadoLibre and don't want to do this by hand, Ficanaz has it built in:
- Claims: a table by SKU with units sold, claims, returns and their rates, MercadoLibre's official reason and the real cause read from buyer messages. Rates with no sales show as "—" and those with a small sample show in gray, so a 33% from three sales doesn't scare you. The data comes from your MercadoLibre account and refreshes automatically every 6 hours; you can also sync on demand. Viewing the table and the causes doesn't use credits.
- AI report: when you ask for it, the AI sorts pending messages into the causes in the table above, looks at your claims from several angles (period, causes, products, logistics, resolution, impact, trend) and pulls them together into a summary of what to tackle first. It's charged per use, with a cap you see before running it, and if there are no claims in the period it doesn't run or charge. Buyers' personal data is stripped before the text reaches the AI.
- Assistant: ask "which claims do I have open?" and it answers with yours, along with what MercadoLibre lets you do on each one. With your permission, it proposes the reply, the refund or accepting the return, and you confirm before anything happens.
To read claims, Ficanaz needs your MercadoLibre connection to include post-sale permissions; if they're missing, the screen tells you and takes you to reconnect.
If you're just getting started and don't have claims to analyze yet, start with the basics: how to sell on MercadoLibre in 2026 explains how to pick a product and run the numbers before you list.
Bottom line
The MercadoLibre reason tells you there's a problem. The real cause, the one the buyer writes, tells you what it is and what to do. Measure by product, read the messages, sort them into categories that each lead to a different action, and measure again after you change something. It takes work, but it's the work that keeps the same claim from coming back.
Want to see your claims by product and cause?
Connect your MercadoLibre account and open Claims: the table is free.
Create a free account →