Common Causes of Account Suspension on Amazon and MercadoLibre

There is one email no seller wants to open: the one saying the account has been deactivated. It arrives with no apparent warning, almost always on a weekend, and it leaves you with inventory in the warehouse, ads still running and an entire business frozen. The immediate reaction is panic, and panic is exactly what costs the most, because it pushes you to answer fast instead of answering well.

This article is not here to scare you. It is here to do the opposite: to turn something that feels random into something with mechanics, categories and early signals. A suspension is almost never lightning from a clear sky. It is the visible end of a process that had been accumulating evidence for weeks, and for most of those weeks you had access to the same data the platform did.

The most common confusion is treating “suspension” as a single thing. It is not. On Amazon there are at least two distinct families — performance and policy — measured differently, warned about differently and resolved differently. On MercadoLibre the logic is different again, tied more to reputation and listing moderation than to an account case file. Lumping them together is the reason so many sellers respond with the wrong argument.

One warning that applies to the whole text: policies and thresholds on both platforms change, and they change by marketplace and by category. What you read here is for understanding the mechanism. The exact numbers and deadlines that apply to your account today have to be confirmed in your Account Health page in Seller Central and in MercadoLibre’s official help center, not in a blog post — including this one.

two families of causes that look nothing alike

The first family is performance: quantitative metrics the platform calculates on its own, over a rolling window of days, each with a published target. Defective orders, late shipments, cancellations, valid tracking. It is arithmetic. There is no interpretation and no subjective side: either you are under target or you are not.

The second family is policy: intellectual property complaints, authenticity, product safety, review manipulation, restricted categories, supplier documentation, account information that does not match. Here there is no percentage to watch. There is an event — a complaint, a report, an audit — and a human or automated decision that lands on your account.

The difference matters enormously in practice. A performance problem is fixed by operating better: the number comes down and the case closes on its own. A policy problem is not fixed by good numbers; no amount of flawless sales offsets an unresolved brand complaint. They are two different engines, and the classic mistake is trying to shut down the second one with the tools of the first.

There is a third, rarer but very real category: verification. Outdated tax information, a legal-name change that never got reflected, a bank account that does not match the account holder, expired documents. This is not punishment for bad conduct, it is a pending formality blocking the operation until it is resolved, and it tends to be among the fastest cases to clear.

Amazon’s performance metrics, one by one

Amazon publishes targets for the metrics on its Account Health page. The ones that appear consistently in its public documentation and across industry coverage are: Order Defect Rate (ODR) under 1%, late shipment rate under 4%, pre-fulfillment cancellation rate under 2.5% and valid tracking rate above 95%. Confirm them again in your own Seller Central before treating them as a rule, because they change and they are not identical across marketplaces.

There is a nuance almost no summary mentions, and it prevents unnecessary scares: several of those metrics apply only to orders you ship yourself, meaning FBM. If your entire operation runs on FBA, late shipment rate and valid tracking rate simply are not measured against you the same way, because Amazon does the shipping. Knowing which ones can actually reach you is the difference between watching four numbers and watching one.

ODR is the most important and the most misunderstood. It is not “how many returns do I have.” It is the share of orders that ended in a defect — negative seller feedback, an A-to-z guarantee claim, a chargeback — over a rolling window. What matters is that a single order can add to the numerator weeks after it was sold, so the number moves because of events that already happened and that you can no longer change. That is why you watch it in advance, not in the week you got close to the limit.

There is also an account health score that combines policy violations and performance into a single color-coded indicator. It is useful as a quick dashboard — it tells you whether something is open — but it misleads if you use it as your only source: a green score with an unanswered intellectual property complaint is a green that does not mean what it looks like.

Glossary: stockout, when you run out of units →

the policy causes, which do not warn the same way

Here the list is long and worth knowing even if it never touches you. The ones that show up most in practice for a Mexican seller:

• Intellectual property. A rights holder reports that your listing uses their brand, image or design without authorization. It does not matter whether you bought genuine product: it matters whether you can prove the supply chain.
• Authenticity. A buyer reports the product is not genuine, and Amazon asks for supplier invoices under very specific criteria. It is the cause that depends most on paperwork you should have filed months ago.
• Product condition. You sold as new something that arrived opened, incomplete or with damaged packaging. It gets detected by accumulated feedback, not by a single event.
• Product safety and compliance. Categories with requirements — electrical, children’s items, supplements, cosmetics — where a certificate or a label is missing.
• Reviews and manipulation. Any attempt to influence buyer opinion, including incentives that look harmless.
• Related accounts. Operating two accounts without authorization, or sharing infrastructure with an already sanctioned account.

None of these gets fixed by selling better. All of them get fixed with documentation and process, which is exactly the subject of the third article in this series.

how the MercadoLibre side works

MercadoLibre runs on a different logic, which is why it confuses anyone coming from Amazon. The central axis is not an account case file but reputation, a color indicator recalculated continuously over a recent rolling window. It does not measure your whole history: it measures how you have been operating lately.

The three signals that weigh most in that indicator are claims that reach mediation, sales you cancel yourself and dispatch delays. All three are direct consequences of operations, and all three share the same practical origin more often than it seems: you sold something you did not truly have available. The exact thresholds per metric are published in MercadoLibre’s help center and they vary; check them there before setting yourself an internal target.

A degrading reputation is not a suspension, but it hurts in a similar way: your listings lose visibility, you fall out of benefits and every sale gets more expensive. And it works as an antechamber, because the operation that degrades reputation is the same one that eventually triggers a formal sanction.

On the moderation side, MercadoLibre pauses individual listings for specific causes: a report from a rights holder under its brand protection program, content that violates its policies — contact details, external links, attempts to take the sale off the platform — a wrong category, or account data that requires verification. A paused listing is reversible and specific. Several listings paused for the same cause is a pattern, and patterns do escalate.

There are also temporary account suspensions lasting a few days, during which your listings sit paused and you cannot answer questions. They feel catastrophic and usually are not: they are a warning with an end date.

why there were almost always warnings first

This is the part that lowers the anxiety, and it is true on both platforms: the system warns you. It does so with performance notifications when a metric approaches its target, with policy emails asking for information, with paused listings requesting a fix, with a thermometer that changes color. The problem is usually not the absence of warning but the channel: they arrive in a flooded inbox, in a dashboard section nobody opens, among dozens of automated messages.

Amazon even has a program — Account Health Assurance — that offers eligible sellers a seventy-two hour window to speak with a specialist before the account is deactivated. As far as is public, it is available in some marketplaces and not necessarily in Mexico, and it requires tenure, a professional plan and a registered emergency phone number. Check in your Seller Central whether your account qualifies: keeping that phone number current takes two minutes and is one of the cheapest things you can do today.

And before the formal warnings come the operational signals, which are the earliest of all. Nobody is going to email you saying “you will be cancelling orders in two weeks.” But your own data already says it.

the operational signals that come before the problem

If you reconstruct most performance cases backwards, the origin is surprisingly boring:

• A SKU that ran out and stayed listed. Every order that comes in after that is a cancellation or a late shipment. It is the number one generator of reputation damage on MercadoLibre and of cancellations on Amazon.
• Stock that does not match across channels. The same physical inventory promised twice. You sell on both sides the same day and cannot fulfill on one.
• Misaligned listings. The same product with a different spec sheet, variant or code on each channel, which multiplies fulfillment errors and buyer confusion.
• Price discrepancies across channels. When one channel is left with an old price or an expired promotion, you get claims and cancellations over pricing errors.
• Supplier lead time that stretched. If your restock takes longer than you thought, the stockout is already scheduled even if today’s number looks fine.

None of those five appear on the account health page. All of them are visible in your own operation several days — sometimes weeks — before they turn into a metric the platform punishes.

Glossary: real available stock, what you can truly sell →

what this looks like in iqseller

Worth being clear about scope: iqseller does not manage suspensions or appeals. It does not read your Account Health page, it does not draft plans of action and it does not intervene with Amazon or MercadoLibre. That work is yours and it happens in Seller Central and in MercadoLibre’s help center.

What the panel does do is give early visibility into the operational signals from the previous section, the ones that most often end in a punished metric:

• Inventory consolidates real stock per SKU across Amazon, MercadoLibre, your own warehouses and your 3PL, and separates total stock from real available stock. That is where you see that a SKU is about to run out while it is still listed on two channels.
• Alerts fires when a SKU crosses its critical stock threshold, when a price discrepancy appears between channels, or when a listing stops being aligned with its master product.
• Catalog keeps the Parent → Model → SKU tree and shows which listings on each channel hang off the same real product, which is how you spot the orphan or duplicated listing.
• Forecast projects when each SKU runs out given current sales velocity and the lead time you loaded, so the purchase order goes out before the stockout instead of after it.
• Profitability shows net margin per SKU with COGS, Amazon settlement commissions, FBA and Full fees, shipping, advertising, VAT and withholdings, which is what tells you whether a problematic product is still worth fighting for.

This is not suspension prevention. It is prevention of the stockouts, oversells and misalignments that precede most performance problems. The distinction is an honest one and worth keeping straight.

what to do this week

Three concrete things, none of which takes more than an hour:

Open your Account Health page and read it end to end. Not the traffic light: the list of open violations, the unread performance notifications and the detail behind each metric. If something is open, you know today; otherwise you find out in two months through a worse email.

Verify your contact details on both platforms. Phone, email, tax information, bank account. A warning that lands in a dead inbox is a warning that never existed.

List your SKUs with less than two weeks of coverage and check whether they are still listed on every channel. Done every Monday, that exercise removes the most common cause of cancellations and delays before it exists.

Suspension feels frightening because it is experienced as something without a cause. It has a cause, it has categories and it has advance warning. The next step, if you are already on the other side of it, is knowing how an appeal is structured — and that deserves its own article.