Amazon ODR, LSR And VTR: The Three Metrics That Trigger Warnings

A seller-fulfilled account can be trading normally at 3.9% late shipments and have listings switched off at 4.1%, and the distance between those numbers is usually one bad week at the packing bench in a month that was otherwise fine.

We understand how that feels, because nothing about the business changed. Same products, same team, same carrier, and an average crossed a line Amazon treats as absolute.

Amazon ODR, LSR and VTR generate most of these warnings, and they get treated as one problem when they are three. Order Defect Rate, Late Shipment Rate and Valid Tracking Rate have different fixes and different response times, and working on the wrong one first is the most common reason a month of effort produces no visible change.

What each metric is actually counting

Worth reading carefully, because two of the three measure something other than their name suggests.

The targets come from Amazon's own help pages: Order defect rate, Late shipment rate and Valid tracking rate, all Amazon primary. On the 10-day and 30-day reporting windows for Late Shipment Rate, two independent third parties agree: eComEngine, updated 2025, and Feedvisor, updated 2026.

Two things there catch people out. Late Shipment Rate is not about delivery, it is about the moment you confirm shipment against the ship-by date your handling time created. And Valid Tracking Rate is assessed by category rather than across the account, so one badly tracked category can produce a warning while the average looks healthy.

Why the order you fix them in matters

They respond at very different speeds, which is worth planning around.

Valid Tracking Rate sits almost entirely in the first column, which is why it is the sensible starting point. Late Shipment Rate is mostly there too, because handling time is a setting even if the capacity behind it is not. Order Defect Rate is the slow one, averaging over 60 days, so today's improvement is diluted by history nobody can edit.

What changed in 2025 and 2026

Valid Tracking Rate got harder in 2025. Amazon's announcement states that from 15 January "VTR will be measured on orders shipped through all shipping service providers and will no longer be limited to those that are integrated with Amazon", and that "you are required to maintain a minimum of 95% VTR at a category level". Source: Amazon Seller Central, Amazon primary, 2025.

Accounts using a local carrier with no Amazon integration found those orders counted from that point.

Then on 28 February 2026 Amazon changed how it enforces the related on-time delivery requirement. The announcement keeps the "90% on-time delivery rate (OTDR) requirement for seller-fulfilled listings" and moves enforcement to item level, so "the listings that have the most impact on your ratings drop will be deactivated. Your other listings will remain active".

Where the shortfall is large or repeated, Amazon "may deactivate all your seller-fulfilled listings". Source: Amazon Seller Central, Amazon primary, 2026.

That is better news than it reads, because a metric problem now tends to cost a handful of listings rather than the whole seller-fulfilled catalogue.

How to tell which metric is actually your problem

Four checks, about twenty minutes.

1. The dashboard says which one is out, the report says why. The path is Performance then Account Health, and each metric links through to the order-level detail. The summary number is the symptom and the list underneath is the cause.
2. Late Shipment Rate concentrates by day of week rather than by product. When late orders cluster on one weekday, the cause is capacity or a cut-off time rather than what you sell.
3. Valid Tracking Rate concentrates by carrier and category. Since the 2025 change, orders sent through a carrier returning no scan data are counted, and because the figure is per category, a low-volume category can breach on a handful of orders.
4. Order Defect Rate has to be split into its three parts before it means anything. An ODR driven by claims for items not received is a shipping problem, while one driven by feedback about the product is a listing or quality problem. In a quiet month two defects can cross 1% on their own, so a rising figure is sometimes arithmetic rather than a decline in service.

What to check before you change your handling time

The first instinct with a Late Shipment Rate warning is to extend handling time. It works, and the cost is worth understanding first, because a longer handling time pushes out the delivery promise shown to shoppers on every listing it touches.

1. The real question is which days fail, not whether handling time is too short. If Monday is clean and Friday is not, the fix is the Friday cut-off rather than an extra day across the catalogue. A template promising Saturday dispatch when nobody is in fails every week.
2. Order cut-off time is a separate setting from handling time. Orders arriving after the cut-off get the next working day as day one, and an optimistic cut-off is a common cause of avoidable lateness.
3. Buying the shipping label through Amazon changes what happens next. Those labels carry tracking that reports automatically, so it is the one change that helps two metrics at once.
4. Where Buy Box performance is already suffering, the trade-off needs thought. Extending handling time to protect a metric can cost the offer position you were protecting, and our page on Buy Box loss covers that.

The order we work these in

Not because it is elegant, but because it moves the dashboard soonest.

1. Valid Tracking Rate comes first. It is fastest to correct, measured over 30 days rather than 60, and usually needs a carrier or label change rather than an operational one.
2. Late Shipment Rate is second, starting with cut-off times. The 10-day figure reflects the change quickly, which is useful evidence before the 30-day number catches up.
3. The three parts of Order Defect Rate get separated next. Until you know whether claims, feedback or chargebacks drive it, any action is a guess.
4. Defects that qualify for removal get challenged. Struck-through feedback stops counting towards ODR, and feedback about delivery on an FBA order is eligible on that basis. A-to-z claims decided in your favour do not count either, so responding with evidence beats refunding reflexively.
5. Only then does anything structural change. Handling time, carrier mix, staffing on the busiest day. Real business decisions, and the first hour of a warning is the worst time to take them.

We will not put a timeline on it, because ODR is arithmetic across a 60-day window and no effort shortens that. What we can say is the sequence matters.

On an account running across three marketplaces, the US, the UK and India, our team cleared 5 account health violations inside a week, restored suppressed listings, corrected FBA dimensions, recovered inbound FBA overcharges and secured further reimbursements. Metrics work sits alongside policy work, which is why our account health management handles them together.

Keeping all three quiet

These metrics reward routine more than effort, and almost every account we see in trouble was doing the right things without a schedule.

1. The dashboard gets read weekly, on a fixed day. The 10-day Late Shipment Rate figure exists so problems get caught before the 30-day number moves, and it only helps whoever looks.
2. Tracking upload is automated, and checked per category. Manual entry at the end of a busy day is where Valid Tracking Rate quietly leaks, and because the figure is assessed per category, the account-wide average is the number that hides it.
3. Handling time is set against your worst week rather than your average one. Promising two days and delivering in one costs almost nothing. Promising one and delivering in two costs a metric. Buyer messages work the same way, since a lot of negative feedback follows an unanswered message rather than a failed product.

What we would do first if this were our account

Open the order-level report behind whichever metric is flagged before changing any setting, because the summary percentage never says whether the cause is a weekday, a carrier, a category or one product.

Then take Valid Tracking Rate first even if Late Shipment Rate looks worse, since it moves fastest and a dashboard that starts improving buys room to deal with the slower metric properly. If a warning is already on the account, the appeal and the metric work are two different jobs, which most sellers discover late. Our policy warnings page covers the appeal side.

We cannot promise a number will fall by a given date, because two of these three average over windows nobody can shorten. What we can tell you, before you spend anything, is which metric is driving your warning and which fixes are settings rather than operations. That is what the free, no-obligation audit covers, and if your shipping is sound and the number is a low-volume artefact, we will tell you that instead.

Related guides

• Handling a Used Sold as New complaint
• Appealing a false Restricted Product flag
• Recovering rank after a stockout

Common questions about ODR, LSR and VTR