AcademyPro: what insiders knowReimbursements and fee forensics: the money Amazon owes you
Pro: what insiders know

Reimbursements and fee forensics: the money Amazon owes you

Lesson 7/13 ⏱ ~12 Min. By Enes Kurt Updated August 2026
What you'll take away

FBA is a warehouse with millions of movements a day. In any system that size, things get lost, damaged, miscounted or mismeasured. Amazon reimburses such cases — but only part of them automatically, and only while deadlines run. This lesson turns chance finds into a process. It is the logical continuation of lesson 6: that one was about what you pay, this one about what you overpaid.

Important up front

Reimbursements are a legitimate claim, not a bag of tricks. Inventing claims, filing the same case repeatedly or firing off automated bulk requests risks your account — Amazon treats that as abuse. A clean process is entirely sufficient: in a well-run account the genuine cases are numerous enough.

1The claim types
CaseWhat happenedWhere you see it
Lost in the warehouseUnits vanished from inventory without being soldInventory ledger: receipts minus outbound do not reconcile
Damaged in the warehouseAmazon rendered stock unusable or booked it as damagedInventory adjustments with a matching reason
Receiving discrepancyFewer units booked than deliveredShipment reconciliation: shipped versus received
Refunded but never returnedThe customer got their money; the goods never reached the warehouseReturns report against the inventory ledger
Returned but booked unsellableStock came back and was graded defective without customer damageReturns report, condition column
Wrong dimensions or weightThe stored size tier does not match realityFee preview against your own measurement
Removal order lostRemoval requested, goods never arrivedRemoval order report against your own goods-in
In plain words

Imagine handing your stock to an enormous warehouse someone else runs. At month end you get a receipt. Almost nobody checks it — and that is exactly where the errors sit. This is not about suspecting the warehouse: with millions of movements, errors are normal. What is unusual is how few customers do the arithmetic.

2The valuation rule: manufacturing cost, not selling price

For losses occurring before a customer order, Amazon has since a policy change in spring 2025 reimbursed the manufacturing cost value — your bare purchase or production price, without freight, duty, handling or margin (evidence level: practice; the policy sits in Seller Central behind the login). Previously the selling price was the usual basis.

From that follows the single most important action in this lesson: maintain your cost values in the account. Without stored costs Amazon estimates — and estimates rarely fall in your favour. Concretely:

  • Enter a cost value per SKU and update it with every new purchase price.
  • Only include costs that match the definition. Folding in freight and duty invites queries that stall the whole claim.
  • Keep evidence to hand: the supplier invoice per batch. Without it, every figure is an assertion.
Example

Fourteen units of the AURELO set vanish in the warehouse. With no cost value stored, Amazon reimburses on its own estimate — €3.10 per unit in this example. With a stored cost value of €4.50 (the actual supplier invoice, excluding freight) it is €63 instead of €43.40. The €20 difference sounds small; extrapolated across a year with a shade under one percent shrinkage over the whole range it becomes a three- to four-figure amount — for an entry made once per SKU.

3Dimensions and weight: the silent permanent overpayment

The fulfilment fee follows Amazon's own measurement of the ready-to-ship unit (lesson 6). If that measurement differs from reality you pay the difference on every single sale — permanently and without warning. That makes this the most rewarding case of all: one corrected value applies to every future order.

  1. Measure yourself, properly. Packed and sealed: length, width, height, weight. Several units from different batches, and note the worst value.
  2. Compare with the stored values. The fee preview shows which size tier Amazon is using.
  3. Request a re-measurement on any discrepancy. With photos: product against a tape measure, scale with the reading visible, both in the same frame as the label.
  4. Always check after packaging changes. New packaging does not automatically mean a new measurement — otherwise the old tier stays.
4Deadlines: the point where claims die

Deadlines differ by claim type, and the publicly circulating figures contradict each other: 60 days from the loss or damage notice for warehouse cases, markedly longer periods for other claim types and 90 days for removal orders are all cited (evidence level: disputed — what counts is the figure in your account). For practice that means two things, both independent of the exact number:

  • A monthly audit run beats any deadline debate. Check every month and you are never in the position of having to invoke a deadline.
  • Plan against the shortest circulating deadline. The same logic as the conservative corridor in Listing L16: work to the strictest figure and you lose nothing if a more generous one applies.
Most common mistake

“Tidying up” once a year and discovering half the cases have expired. Reimbursements are not an annual project but a 30-minute monthly routine. The second most common mistake is the opposite: filing the same matter repeatedly because the first answer was unsatisfying. Duplicate claims count as abuse — you follow up WITHIN the existing case, with new evidence.

5The monthly audit run in 30 minutes
  1. Pull the inventory ledger (previous month) and reconcile per SKU: opening stock plus receipts minus sales minus returns received = closing stock. Every discrepancy is a candidate.
  2. Put the reimbursements report beside it. Amazon reimburses many warehouse cases automatically today (evidence level: practice) — anything already listed there is settled and is not filed again.
  3. Check returns against inventory receipts: which refunded order has no return into the warehouse?
  4. Reconcile open shipments: shipped versus received, per shipment.
  5. Spot-check the fee preview against your own measurements — two or three SKUs a month give rolling coverage.
  6. Open cases: one per matter, with a report extract, period, SKU, unit count and amount. Bundled claims across several matters get a blanket answer.
Example

March's audit run shows a gap of 9 units on the AURELO set. The reimbursements report already contains 5 of them automatically — 4 remain. For those 4 the inventory ledger supplies date, reason and quantity. The case is opened with exactly those four lines, not with the sentence “units are missing”. Processing time: eleven minutes. Result: €18 — unexciting on its own, but it is one of an average five cases a month, and the routine behind it costs nothing extra.

6Service providers: when a percentage model pays

Providers exist who run the reconciliation automatically and keep a share of the reimbursement. The sober assessment:

  • In favour: completeness on large ranges. With 300 SKUs you will never find everything by hand.
  • Against: access to your account and responsibility for the claims. In an abuse case your account is liable, not the provider.
  • The line sits, practically, where your own audit run takes more than an hour a month. Below that, doing it yourself is cheaper and more instructive — you see problems no reimbursement report shows.
Reimbursement checklist
  • Cost value per SKU stored in the account and updated with every purchase.
  • Supplier invoices per batch filed within reach.
  • Monthly audit run as a fixed calendar task.
  • Automatically reimbursed cases reconciled before every claim.
  • One case per matter, with a report extract rather than a description.
  • Dimensions and weight checked on a rolling basis, immediately after any packaging change.
  • No duplicate claims, no automated bulk filing.
7Expert insight: the chain of evidence that makes a case undeniable

The difference between “rejected” and “reimbursed” almost never lies in the facts but in the form. A claim an agent can follow in two minutes gets processed; one that raises questions gets closed. A solid chain of evidence has four links:

LinkWhat it provesWhere it comes from
1. The goods existedThey were delivered and bookedShipment closure with received quantity
2. They were not soldNo outbound through an orderSales report for the period
3. They are gone anywayInventory adjustment with reason and dateInventory ledger
4. They were not already reimbursedNo entry in the reimbursements reportReimbursements report for the same period

Evidence all four links with lines from Amazon's own reports and you are not making a claim, you are presenting an invoice. Link 4 is the one most people omit — and it is the most common rejection reason: “already reimbursed”. A claim that proves for itself that it is not a duplicate takes the easiest rejection out of the agent's hands.

Three subtleties separate a good process from a very good one:

  • Pick generous periods and precise numbers. A loss is sometimes booked weeks after the event. Pull reports across two months but name exact lines in the claim — the wide window finds the case, the tight figures make it checkable.
  • Keep your own numbering. Every case gets an identifier in your own sheet, with date, amount, case number and outcome. Without that sheet you will eventually file a duplicate — and that is the one way a legitimate routine turns into an account risk.
  • Analyse rejections rather than filing them away. When the same reason comes back three times, your claim is missing a link. That is process information, not a run of bad luck.

And one thought beyond reimbursement: forensics is also an early warning system. If damage clusters on one SKU, the warehouse is rarely at fault — the packaging is. If receiving discrepancies cluster with one supplier, Amazon is rarely at fault — the count at packing is. Recover only the money and miss the pattern, and you recover the same money every month instead of fixing the cause once.

The automation trap

Tools that file claims automatically produce exactly the bulk patterns Amazon reads as abuse — especially when they refile cases that were already reimbursed automatically. Automating DETECTION is entirely safe: a script that reconciles inventory and hands you a list is pure legwork. Filing stays a human with a spreadsheet.

This bonus track unlocks later

The pro track assumes the other three and repeats nothing from them. It opens once you have completed all three in full — every lesson quiz at 80 % or better, and every final quiz passed.

Not logged in? Your progress only counts with a free account.

Check yourself

6 quick questions — one at a time, instant feedback. With a free account your progress is saved.

How does Amazon value stock lost before a customer order?
Since a policy change in spring 2025, losses before an order are valued at the bare purchase or production price — without freight, duty, handling or margin.
Why should a cost value per SKU be maintained in the account?
With no stored value Amazon applies its own estimate. Those are rarely generous, and the difference applies to every single loss case.
Why is auditing dimensions and weight especially rewarding?
Other cases are single events. A wrong size tier applies to every future order — so a one-off correction has a permanent effect.
What do you check before opening a reimbursement case?
Amazon reimburses many warehouse cases automatically today. Skip the reconciliation and you file settled cases — which is exactly the pattern that counts as abuse.
Which link in the chain of evidence is most often forgotten?
“Already reimbursed” is the most common rejection reason. A claim that shows for itself that it is not a duplicate removes the easiest rejection.
What may you automate in the reimbursement routine?
Reconciliation is pure legwork and a script may do it. Automatically filed claims, by contrast, produce exactly the bulk patterns that get read as abuse.

Frequently asked

How much does a clean audit run typically recover?

That depends entirely on range size and stock movement — a general percentage would be invented. What is reliable is the structure: the amounts per case are small, the annual total is not, and the real gain is often not the money but the pattern the audit makes visible.

What do I do when a case is rejected?

Follow up in the EXISTING case, never open a new one. And take the reason seriously: if a link in the chain of evidence is missing, supply exactly that — with lines from Amazon's own reports. If the same reason comes back three times, it is your claim format, not bad luck.

Do I have to audit dimensions for every product individually?

Not all at once. Two or three SKUs a month give full coverage across a year and keep the effort small. Immediate checks are needed on only two triggers: after a packaging change, and when the fee per unit rises with no visible reason.

← Previous lessonThe 2026 FBA fee map and the capacity question Next lesson →Inbound depths: shipment splits, bulk storage and multi-channel
🛠 Check the theory against your own ASIN

The free Listing Check scores any ASIN from 0 to 100 in a minute — the fastest way to see whether the mechanics from this lesson actually hold on your own listing.

Go deeper: the complete guide to optimizing Amazon listings →

Enes Kurt
Amazon seller for over ten years · founder of Listimo

Everything in this academy comes from day-to-day selling practice — the same playbook behind Listimo, the tool that turns product photos into complete Amazon listings.