📋 Overview
If you fulfill your own orders, two metrics decide whether Amazon considers you operationally reliable: Late Shipment Rate and Pre-Fulfillment Cancellation Rate. Both are easy to breach on low order volume, and both are driven by settings you configured once and probably have not looked at since.
This article is for merchant-fulfilled (FBM) sellers. It covers what each metric actually counts, how your expected ship date gets calculated, and what to do when you cannot ship an order on time.
⏱️ What Each Metric Counts (and What It Doesn’t)
Both metrics live in Account Health under shipping performance. Both apply only to merchant-fulfilled orders — FBA orders are excluded, because Amazon controls the shipping on those.
Late Shipment Rate (LSR)
LSR is the percentage of your orders where shipment confirmation happened after the expected ship date. Amazon’s target is below 4%.
The trigger is the ship-confirm timestamp, not the carrier’s first scan. An order you confirm on time but hand to the carrier a day later does not become a late shipment — it becomes a potential Valid Tracking Rate or late-delivery problem instead. Amazon publishes LSR over both a 10-day and a 30-day window, so a bad week shows up in the short window first while the long window is still green.
Pre-Fulfillment Cancellation Rate (PFCR)
PFCR is the percentage of orders you cancel before confirming shipment. Amazon’s target is below 2.5%. Once an order is ship-confirmed, cancelling is no longer possible — a refund at that point is a return or a refund, and it affects different metrics.
The reason code you select when cancelling does not change whether the cancellation counts. Amazon’s published guidance is that cancellations the buyer initiates are excluded, but cancellations you perform yourself are counted. Confirm the current treatment in Amazon’s seller performance metric definitions before you rely on it.
Why small accounts breach these first
The denominator is your order count in the window, not your lifetime volume. At 25 orders in a 10-day window, a single late ship-confirm puts you at 4%. At 40 orders, one seller cancellation puts PFCR at 2.5%. Neither of these metrics is forgiving on a new or seasonal FBM account, which is why the fix has to be structural rather than reactive.
🗓️ The Expected Ship Date Is the Whole Game
You never set the expected ship date directly. Amazon calculates it from the order time plus the handling time on the shipping template attached to that listing, counted in business days, with your configured cut-off time and operating days applied. Every LSR problem traces back to one of these four inputs being wrong.
1. Audit handling time against your slowest realistic day
Most sellers set handling time to the fastest they have ever shipped. The correct input is the slowest you can reliably hit when a supplier is slow, a staff member is out, or an order lands at 4:58 p.m. on a Friday. Open your shipping templates and compare the handling time on each one to your actual ship-confirm history over the last month or two. A good result is a handling time you hit comfortably on your worst day, not your best one.
Lengthening handling time costs you some delivery-speed competitiveness and can affect how your offer displays. That trade-off is real, but it is smaller than the cost of an account health notification.
2. Check your cut-off time and operating days
Your cut-off time determines whether a late-afternoon order is treated as today’s order or tomorrow’s. Your operating days determine whether weekends count as business days for handling. If you set up your account assuming Saturday pickups and your carrier has since changed its schedule, every Friday order is now running on a clock you cannot meet.
3. Plan for closures before they happen
Holidays, moves, inventory counts, and vacations all produce the same failure: orders keep arriving against a handling time nobody is there to honor. The compliant options are to extend handling time ahead of the closure, remove the operating days you will not be shipping, or set your listings to inactive for the period. All three must be done before the orders arrive, because the expected ship date is locked in at the moment of purchase.
4. Know whether automated handling time is on
Amazon can adjust handling time for you based on your recent ship performance. If that setting is active, the number you see in your template may not be the number being applied. Check its status on each template so you know which value is actually governing your expected ship dates.
📦 Why Pre-Fulfillment Cancels Happen — and the Alternatives
Almost every seller-side cancellation comes from one of three causes, and each has a fix that is not “cancel the order.”
Overselling from stale quantities
If you sell the same stock on other channels or in a physical location, your Amazon quantity is a snapshot that goes stale between updates. The levers are update frequency and a quantity buffer — holding back a few units per SKU so the last physical unit is never the one Amazon sells. Buffers cost you a small number of sales; cancellations cost you the sale and the metric.
Pricing errors you want to walk back
Cancelling an order because the price was wrong is a seller-initiated cancellation and counts the same as any other. If a repricing mistake or a feed error creates underpriced orders, the usual choice is to honor the orders and fix the price immediately. Correct the input, not the order.
A buyer asks you to cancel
Direct the buyer to request the cancellation from their own order page so the request is recorded as buyer-initiated, then action the request that appears in Manage Orders. If you cancel the order yourself because a buyer emailed you, you are the initiating party on Amazon’s record.
💡 Pro Tip: Work the Unshipped view in Manage Orders at a fixed time every working day and sort by ship-by date rather than order date. The order about to go late is rarely the oldest one — it is the one that landed under a short handling time on a SKU with a two-day template.
When you genuinely cannot ship on time
If stock will arrive within a day or two, shipping late is usually the better outcome: you take an LSR hit on one order, the buyer gets the product, and the sale survives. Message the buyer through Buyer-Seller Messaging with a realistic date and ship as soon as you can. If the item is genuinely unavailable, cancel promptly rather than letting the order sit — a buyer waiting past the delivery estimate with no product is a path to negative feedback and an A-to-z claim, which feed Order Defect Rate (target: below 1%) instead.
🚫 Mistakes That Quietly Inflate Both Rates
Ship-confirming before the package exists
Confirming shipment early to beat the expected ship date is the most common “fix” and the most damaging one. It is a false shipment confirmation, it starts the buyer’s delivery clock before the item moves, and it converts a contained LSR problem into tracking, late-delivery, and claim problems. Ship the order, then confirm it.
Fixing the setting but not checking the window
These are rolling metrics. Extending handling time today does nothing for orders already in the window — the breach has to age out. Sellers who change a setting, see no movement in 48 hours, and change it back end up with neither the old behavior nor the new one. Make the change, then watch the short window first.
Running one shipping template across a mixed catalog
A single template means your slowest SKU runs on the same handling time as your fastest. Products that ship from a supplier, require assembly, or are made to order need their own template with a longer handling time. Group listings by how long they actually take to get out the door, not by category.
❓ FAQs
Does a late shipment count toward my Order Defect Rate?
No. Late Shipment Rate is its own metric with its own target. The connection is indirect: late orders generate negative feedback and A-to-z claims, and those do feed ODR.
I fixed the problem — why is my rate still over the limit?
Because the window is rolling. The offending orders stay in the calculation until they age out of the measurement period, and on a low-volume account a small number of good orders will not dilute them quickly. Keep shipping on time and watch the shorter window for the first sign of recovery.
Do cancellations for suspected fraud or bad addresses count against me?
A cancellation you initiate generally counts regardless of your reason. That does not make cancelling wrong — shipping to an address you believe is fraudulent carries its own risk. Treat it as a cost you accept occasionally, not a routine tool, and check whether an address-validation step at order review can catch these before they become orders you have to cancel.
What happens if I go over the target?
Amazon flags the metric in Account Health and may require a plan describing the root cause and what you have changed. Sustained breaches can put selling privileges at risk. The notification itself is the prompt to fix the setting that caused it — handling time, operating days, or inventory accuracy — and to document that change in your response.