Key takeaways
- On average, Amazon takes 19.4% of every sale in fees, a 2.1 point increase from 2025.
- Storage fees, attributed advertising, and returns together consume 18.5% of sales, nearly matching product costs.
- 61% of sellers unknowingly pay the aged-inventory surcharge after 271 days in stock.
- One in four ad campaigns spend more than the product’s unit margin, risking profitability.
- Sellers lose an average of €3,200 per year in unreimbursed returns and damaged inventory.
Every seller knows Amazon charges. Very few know how much, precisely, on every unit they ship. The gap between those two things is why so many accounts grow revenue and stall on profit.
For this report we crossed the settlements, advertising reports and product costs of 4,812 European accounts connected to SellerKey. No surveys, no estimates: just the real waterfall, from the gross sale to the euro that lands in the bank account.
19.4%
of every sale goes to Amazon fees, on average
+2.1 pt
increase vs. the 2025 report
4.3%
recoverable with logistics and Ads adjustments
Where the money goes, line by line
Take an average €34.90 order in the Home category, with FBA logistics and standard size. This is the breakdown almost no dashboard shows in one place:
| Line item | Amount | % of sale |
|---|---|---|
| Gross sale | €34.90 | 100% |
| Referral fee | −€5.24 | 15.0% |
| FBA fulfillment fee | −€3.66 | 10.5% |
| Storage + aged inventory | −€0.71 | 2.0% |
| Attributed advertising | −€3.84 | 11.0% |
| Returns and refunds | −€1.92 | 5.5% |
| Product cost (COGS) | −€11.86 | 34.0% |
| Net profit | €7.67 | 22.0% |
Sample median. VAT is excluded from all calculations.
The headline isn't the referral fee. That one is well known and hard to move. The interesting part is the three line items most sellers treat as background noise: storage, attributed advertising and returns. Together they eat 18.5 points, almost as much as the product cost itself.
“I thought my margin was 34%. When I saw the full waterfall it was 21%, and my best-selling product was losing money.”
Marta G. · Home seller, €1.8M/yr
The three leaks that weigh the most
1. Aged-inventory surcharge
It's the line item that grew the most in 2026: +38% year over year. It kicks in silently on stock past 271 days and shows up in no sales report. In our sample, 61% of accounts pay this surcharge without ever having detected it.
2. Advertising attributed to unprofitable SKUs
The sample's average ACOS is 24%. That is sustainable for a product with 40 points of gross margin and lethal for one with 22. One in four active campaigns spends above the unit margin of the product it promotes. \n
3. Returns never restocked
When a customer returns, Amazon pulls the sale instantly. The inventory reimbursement, though, takes time, and sometimes never arrives. We detected an average of €3,200 per account per year in lost, damaged or never-restocked units nobody claimed.\n\n## What you can actually cut
Of the 19.4 fee points, we estimate 4.3 are recoverable without switching suppliers or lowering prices. In order of effort vs. return:
- Claim pending reimbursements. Immediate return, zero risk. It's money that is already yours.
- Move old stock out before day 271. Clearance, bundles or removal: any exit is cheaper than the surcharge.
- Set a target ACOS per SKU, not per account. The limit is set by the product's margin, not the portfolio average.
- Review the declared size and weight. A size-tier misclassification costs between €0.40 and €1.10 per unit.
Key takeaways
- The average seller hands 19.4% of every sale to Amazon, 2.1 points more than last year.
- The big leaks aren't in the referral fee, but in storage, Ads and returns.
- There are 4.3 margin points recoverable with decisions that don't touch price.
- Without the per-SKU waterfall it's impossible to know which product is financing which.
Methodology
Sample of 4,812 accounts with at least 12 months of connected history, based in Spain, Germany, France and Italy, between January and May 2026. Amounts are normalized to euros excluding VAT. Medians are computed per account, not per order, to avoid high-volume-account bias. Data is aggregated and anonymized; no individual account is identifiable.
