Late shipments in December: when the carrier's delay becomes your metric

January brings December’s emotional invoice, and it almost always looks the same: a handful of negative ratings saying “arrived late”, a thermometer that dropped a color, or an indicator in Seller Central that turned yellow. And the seller’s reaction is always the same, and entirely fair: I handed the package over on time, the carrier is the one that was late.

That is true. And it is irrelevant to the marketplace. From Amazon’s and MercadoLibre’s side, the seller is the one who chose the carrier, chose the service level, published the estimated date and took the money. The delivery network is your supplier, and your supplier’s problems are your problems. That is the rule, and however unfair it feels, arguing with it changes nothing: the only thing that changes anything is understanding exactly where responsibility is cut and operating accordingly.

Which leads to the distinction that organizes this whole topic. There is a portion of late shipments that is 100% yours, measurable with precision and fixable with decisions. And there is another portion that is not yours, that you cannot prevent, and that gets charged to you anyway: there, all you can do is size it, document it and, where applicable, appeal it with evidence.

This post is about that line: which side is yours, how to measure each one separately, and what to build now so that in January you have evidence instead of a feeling.

a late shipment is not one event, it is three different things

What the buyer experiences as “it arrived late” breaks down into three failures that get fixed in completely different ways and that are worth keeping out of the same number.

• Late dispatch. The order came in and you took longer than your committed handling time to hand it to the carrier. This is one hundred percent yours.
• Late transit. You handed it over on time, with a valid label and inside the pickup window, and the carrier took longer than its own commitment. This is not yours, but it is charged to you.
• A miscalibrated promise. Nobody was late: the date you published was never achievable with the service you offered to that destination. This is also yours, even though it feels like bad luck.

The third case is the most common and the worst diagnosed, because it does not look like an operational error: it looks like a slow carrier. But if your promise says three days and your ground service takes five to that postal code, you do not have a carrier problem, you have an arithmetic problem. How that date gets computed is covered in what lead time is and how to measure yours.

Separating the three is the first piece of real work. If your report says “14 late shipments in December” and you do not know how many of each type, you do not know what to fix.

Glossary: lead time, the real end-to-end clock →

what the marketplaces measure, and over what window

It is worth understanding the mechanics before trying to manage them, and it is worth being honest about which figures are verifiable and which are not.

On Amazon, the direct indicator for orders you fulfill is the late shipment rate, which measures the share of orders you ship-confirmed after the committed date. The third-party material we checked puts the limit at under 4%, with valid tracking rate above 95%, both measured over 30-day windows. Those figures are widely repeated, but Amazon’s public documentation defines the metric without always confirming the live threshold per marketplace, and the targets can vary by region. The number that applies to you is in the Account Health section of Seller Central, and that is the only authoritative source.

There is an important detail in that metric: it measures ship confirmation, not delivery. In other words, it punishes late dispatch — the first case on the list — and not late transit directly. Transit reaches you through another door: buyer dissatisfaction, ratings and order defects, which is the subject of Order Defect Rate and account health.

On MercadoLibre the logic is different. Reputation is computed over a rolling 60-day window and built from three factors: claims that went to mediation, sales cancelled by the seller, and delayed dispatches. The result is painted on the colored thermometer that decides your search exposure and your eligibility for promotions. The exact tolerance percentages per country are not clearly published in open sources: what circulates on seller blogs mentions caps on delayed shipments, but with no reliable attribution to Mexico. There too, the authoritative source is MercadoLibre’s official help center and your own panel.

What can be stated without risk, and this is what changes decisions, is the structure: neither of them measures the day, both measure a window that drags. A bad December does not close on the 31st; it keeps weighing on your number well into February. Which means the cost of a late shipment on December 18 includes the visibility you will lose in January, when the peak is over and traffic is cheaper but you need it to rebuild momentum.

the part you do control, in order of impact

There are four levers, and they are worth addressing in this order because it is the order of how much they move the number.

One: the published handling time. If your packing capacity cannot sustain one day in peak season, the right move is not to try harder: it is to raise the published handling time to two days before it breaks. It costs you some conversion and it saves your metric. It is a cost you choose rather than one you get charged.

Two: the cutoff date per zone. Distant postal codes have longer transits and therefore an earlier cutoff. Having a single nationwide promise calibrated on metro transit guarantees late shipments to the provinces. Differentiating by zone is the highest-impact decision and the one fewest sellers make.

Three: the label and the tracking. An order dispatched on time but with an invalid label, a number that never updates or a carrier that does not report events looks, to the system, exactly like a shipment that never left. This gets fixed with process discipline, not with negotiation.

Four: peak capacity. If your season plan does not say who packs on the Saturday and Sunday of December’s second week, your Monday late dispatch is already decided. The handling-time bottleneck is almost never the carrier: it is a table and a person.

the part you do not control: size it and document it

Late transit cannot be prevented. What can be built is the record that lets you do three things: negotiate with the carrier, appeal with evidence when the marketplace charges you something that was not your failure, and decide next year with data instead of resentment.

What to capture per shipment, starting now and not in January:

• Order timestamp and handover timestamp, with the pickup receipt. This proves the dispatch was on time.
• Tracking number and service level purchased, with the day commitment the carrier publishes for that destination.
• Tracking events with date and time, including failed delivery attempts and the reason reported.
• Actual delivery date and the gap against the promised date.
• What claim or rating it generated, if any.

With that record, two conversations become possible that are impossible without it. The first is with the carrier: “in December your ground service to these twelve states missed its commitment on 31% of my shipments” is a negotiating position; “you failed me a lot” is not. The second is with the marketplace: many rating and order-defect appeal processes ask for exactly this evidence, and having it organized per shipment is the difference between being able to appeal and having to absorb it.

And a third use almost nobody makes: using the record to pick carriers by zone. Hardly any seller has only one shipping option, and hardly any knows which of their options performs best in which states. Two months of data settle that question permanently.

the trap that turns a stock problem into a shipping problem

There is a case that gets reported as a late shipment and has nothing to do with logistics: the order came in, but there was no unit to pack.

It happens when the stock one channel publishes is already committed by sales on another, or when the number the channel shows comes from a sync that took hours. Handling time does not stretch because you pack slower, it stretches because you are waiting for merchandise to arrive. And from the buyer’s and the marketplace’s side, it looks exactly like a late dispatch.

So this is worth checking when diagnosing: if your late shipments cluster in a few SKUs and coincide with low-coverage periods, you do not have a packing-operation problem, you have an inventory problem in disguise. The fix lives elsewhere — in the reorder point and in syncing — and attacking the symptom gets you nowhere.

Glossary: real-time sync, why stale data costs you →

why all of this weighs double in December

Three effects stack in the same month and explain why the damage from a December late shipment is greater than from a May one.

Volume concentrates the error. If your late shipment rate is 3% all year, in a normal month that is a handful of orders. In December, at triple the volume, it is triple the affected orders charged to the same measurement window. The percentage holds steady and the absolute number of unhappy buyers multiplies.

The delivery network is saturated. Transit stretches and becomes more variable exactly when your promise has the least slack. And failed delivery attempts rise, because people travel and are not home.

Buyer urgency is at its maximum. A two-day delay in March is an annoyance. The same delay on December 23 makes the product useless. The same operational failure produces a far harsher reaction, and in the month when the most people leave ratings.

The practical conclusion is that December’s margin for error is smaller, not larger, and that the conservative decisions — a longer promise, an earlier cutoff, distant zones closed sooner — pay for themselves.

how this looks in iqseller

Worth stating plainly what the panel does not do: it does not generate labels, does not track packages, does not quote carriers and does not compute your late shipment rate. That information lives in your shipping platform, in the carrier’s portal and in the indicators inside Seller Central and the MercadoLibre panel. If what you need is a shipment tracker, the panel is not that.

What it does solve is the part of the problem that is about inventory and decisions, which is usually what sits behind the late shipments that were avoidable.

The Inventory module keeps real available stock consolidated — FBA, Full, your warehouse, your 3PL — aligned with what each channel publishes, so you do not sell units already committed on another channel. That eliminates the category of late shipment that was really a stockout.

Alerts is the module that applies most here: it flags when a SKU crosses its critical threshold or when sales velocity accelerates enough to pull the stockout date forward. In peak season that flag is the signal to review that SKU’s published promise before it starts producing late dispatches.

Forecast projects when each SKU hits zero per channel, which lets you decide days in advance whether to lengthen handling time, close distant zones or pause. And Profitability gives net margin per SKU — with COGS you loaded, commissions from the Amazon settlement and MercadoLibre orders, FBA and Full fees, shipping, advertising, VAT and withholdings — which is what decides whether paying for an express service to keep the promise is worth it, or whether the SKU simply cannot carry that freight.

the minimum record worth building this week

You do not need a system. You need a sheet with five columns, filled in from the first day of the season, which in January becomes the most valuable document you own:

• Failure type: late dispatch, late transit or miscalibrated promise. One of the three, per shipment.
• Delivery zone and service level purchased, so you can group and see patterns by state and by carrier.
• Committed days against actual days, with the pickup receipt and the tracking events.
• Consequence: claim, rating, cancellation or none.
• SKU and that SKU’s coverage that day, to catch the cases that were an inventory problem in disguise.

By the end of the season that sheet gives you three answers: which part of the damage was yours and is fixable, which part was the carrier’s and is negotiable, and which part was inventory and gets fixed in the forecast. Without the sheet, in January all you have is a feeling that December went badly, and a feeling cannot be appealed or negotiated.