Your real Amazon fees.
Not the estimate.

Paste your settlement or payments report and see what Amazon actually deducted: real per-unit fees, your measured referral rate, and the break-even ACoS your ads should be judged against. Free, no signup.

100% in your browser — nothing uploads No signup, no cookies US · UK · CA · AU

Break-even ACoS calculator

Three numbers in, three decisions out. If you don't know your real fees per unit, the fee checker below reads them from your own report.

Fees = referral + fulfilment (+ per-unit storage if you track it). From the settled numbers, not the fee preview — the two disagree more often than you'd think.

Contribution per unit
price − fees − COGS, before ads
Break-even ACoS
ad spend above this loses money on the sale
Max bid (break-even CPC)
contribution × conversion rate

Settlement fee checker

Reads the Date Range transaction report (Payments → Reports Repository → CSV) or the V2 settlement flat file (.txt). English-header marketplaces.

Drop your report here or click to choose a file
.csv or .txt — parsed locally, never uploaded

Privacy: this page has no upload endpoint. Parsing happens in your browser — check the network tab while you use it, or read the page source.

Orders / units settled
Measured referral rate
referral fees ÷ product sales
Fulfilment fee / unit
FBA fees ÷ units
Net per unit (pre-COGS, pre-ads)
sales − all fees, ÷ units

Referral rate by month

A step between months usually means Amazon changed your rate — or your category — without telling you.

MonthUnitsProduct salesReferral %FBA / unit

Where the money went

LineAmount% of sales

Why the settlement file, not a calculator

The method behind this tool, learned on our own account.

Every fee calculator — including Amazon's own preview — returns an estimate. The settlement file records what Amazon actually deducted, order by order, fee line by fee line. On our account the estimate ran about $0.70 per unit kinder than the settled truth. Price your ads against the kind number and every campaign looks better than it is.

Worse: fees move and nothing tells you. In July our referral rate went from ~5.6% to ~12.6% overnight — no notification, no dashboard change. For a month every bid was priced against a break-even that no longer existed. The by-month table above exists because of that month.

Two traps if you ever do this by hand in a spreadsheet: in the V2 flat file, filter on amount-type and amount-description together — "Principal" alone matches both the item price row and the negative promotion row, which double-counts units and manufactures a plausible-but-wrong referral rate. And check multi-unit orders separately: Amazon writes them in two valid shapes, and one of them quietly collapses everything past the first unit.

What this page deliberately does not do: it never estimates. Ad spend is billed outside settlements, so it isn't here. COGS is yours, so it isn't here. What's left is the one thing a dashboard can't give you — the measured number.

FAQ

Is my data uploaded anywhere?

No. The file is parsed by JavaScript in your browser. This site has no upload endpoint — verify it in your browser's network tab while using the checker.

Which reports work?

The Date Range transaction report CSV (Seller Central → Payments → Reports Repository) and the V2 settlement flat file (tab-separated .txt). Headers must be in English — US, UK, CA and AU marketplaces are fine.

Why is my measured referral rate different from the table rate?

Category reassignment, a promotional rate expiring, VAT interactions on some marketplaces, or a plain rate change. The by-month view shows when it moved, which is the first step to finding out why.

Does it include ad spend?

No — Amazon bills advertising outside the settlement, so it can't be measured from this file. Judge ads against the break-even ACoS above, using your ads console spend.

Who built this?

An Amazon seller (AU/US/CA) who got burned by a silent referral-rate change and started re-measuring fees from settlement line items on a schedule. The longer-form tooling lives at agentxray.ai.