“Amazon takes 15 %, we have plenty of room.” The spreadsheet was open on my screen, theoretical margin in bright green. I took one single SKU with him and rebuilt it line by line against the platform’s own settlement report. On that one, the marketplace was collecting 27 % of what the customer paid.
Nobody had lied. The rate card really does say 15 %. It was everything else that appeared nowhere.
A marketplace does not sell your product. It rents you a customer, one order at a time. And rent is worked out per SKU, never per channel.
1. The rate card never tells you what it applies to
The headline percentage is the easy part. What decides everything is the base it is calculated on.
Amazon charges its referral fee on the “total sales price”, meaning the full amount paid by the buyer, item price and shipping included, as stated in its official rate card. eBay applies the same rule: the final value fee covers the total amount of the sale, shipping, VAT and other taxes included, per its business seller fee page. Cdiscount goes as far as printing the formula on its pricing page: commission = (sales price incl. VAT + shipping costs) × category rate.
First consequence, and this is where it bites: free shipping is not free. You pay the carrier, then you pay a commission on the amount you chose not to invoice.
Take a €24.90 item with €4.90 shipping, Home category, on eBay. Headline rate: 9 %.
- 9 % of €29.80 (the real base): €2.68
- fixed fee of €0.35 per order: €0.35
- regulatory operating fee, 0.35 % of the total: €0.10
Total taken: €3.13. Against the item price alone, that is 12.6 %. The rate card said 9. And I have not yet mentioned the box, the carrier or the advertising.
2. Fixed fees decide the bottom of your catalogue
Run the same maths on a €6 item with €3 shipping: €0.81 commission, €0.35 fixed, €0.03 regulatory. That is €1.19, or 19.8 % of the item price. The thirty-five cents alone account for almost six points.
Amazon pulls the same lever in another shape: a minimum referral fee of €0.30 per item in most categories. On a €3 product in an 8 % category, the theoretical fee is €0.24; €0.30 is what leaves your account, so 10 %.
I often see whole catalogues pushed across because the feed already exists, so why not send everything. That is the surest way to fund the visibility of your expensive products with the margin of your cheap ones, and never notice it happening.
3. The rent, before the first sale
| Platform | Monthly subscription | Headline rate | Calculation base |
|---|---|---|---|
| Amazon | €39 excl. VAT (Professional plan) | 8 to 15 % in most categories; 7 % in computing and consumer electronics; 45 % on Amazon device accessories | Item price + shipping, with a €0.30 minimum |
| Cdiscount | €39.99 excl. VAT | 15 % by default; 7 % on TVs and computers; 20 % on jewellery; 2 points more on non-new items | Price incl. VAT + shipping costs |
| eBay | €0 without a store, €19.50 to €149.50 excl. VAT with one | 5 % in computing, 9 % in home and DIY, 12 % in fashion | Total amount paid, shipping and VAT included, plus €0.35 per order |
Thirty-nine euros a month is €468 a year before you have sold anything at all. At €3 net margin per order, you need thirteen orders a month just to pay the entry ticket. That is the first number I put in front of a client hesitating over a new channel, before we even discuss listings and the single product record behind them.
4. Advertising is not a marketing line
Marketplace Pulse went through the profit and loss accounts of a sample of sellers and landed, in its February 2023 study “Amazon Takes a 50% Cut of Sellers’ Revenue”, on more than half of seller revenue captured by the platform: roughly 15 % referral fee, 20 to 35 % fulfilment for those using Fulfilment by Amazon, and up to 15 % advertising. A few sellers in the sample reached 60, even 70 %. The study is three years old and the detail has shifted since, fulfilment costs in particular. The mechanism has not.
What matters in that breakdown is where advertising sits. On a marketplace it is not a growth investment you review each quarter. It is the price of the shelf. The best-converting slots on the results page are sold; not buying means accepting that you will not be seen. So I treat it as a cost of sale, ranked alongside the commission, never as the line you cut when the month is tight. It is the exact opposite of the reasoning behind a brand campaign, where the question is whether the ad creates anything. Here it creates nothing. It buys the shelf. How to steer that spend, I covered on the Sponsored Products side.
5. What appears on no rate card at all
Three lines only show up once they have hurt. I look for them in the first settlement report a client sends me.
The performance penalty, first. eBay documents it plainly: a seller rated “below standard” on the 20th of the month sees the final value fee rise by 6 percentage points the following month. An 11 % fee becomes 17 %. After four consecutive months, it is 7 points. A dispatch backlog during a peak can cost more than the peak brought in.
Cross-border and currency, next. Still on eBay: 1.6 % international fee when delivery leaves the euro zone, 3.3 % for the rest of the world, and 3 % withheld on any currency conversion. Three lines that stack on the same sale.
Cash, finally. Not a fee, but it behaves like one. Cdiscount remits sales proceeds every ten days and holds a guarantee reserve sized on your flows. You front the stock, the freight and the advertising before you get paid. On a seasonal range, that gap weighs more than a point of commission.
The most expensive mistake
Matching your marketplace price to the price on your own site because “we have to stay consistent”. The same price does not carry the same costs: on your site you pay for acquisition; on the marketplace you pay for acquisition, the rent and the fixed fees. The channel grows, overall margin slips, and the dashboard reports healthy growth. Redo the maths SKU by SKU: at worst you confirm your current prices and sleep better, at best you find a handful of references funding their own competition. Price consistency across channels I treated separately in the article on cannibalisation; it is the commercial counterpart of this margin calculation, and the heart of a marketplace engagement as much as of an e-commerce strategy.
6. The threshold: when I pull a SKU
My test fits in one line. I work out what a SKU actually returns on the marketplace once commission, fixed fees, fulfilment and advertising are out, and I compare it with what the same SKU returns elsewhere.
A SKU comes off when the net margin left no longer covers the cost of handling an order in your own warehouse. Not when it turns negative: well before that. An order leaving forty cents ties up the same picker, the same box and the same after-sales desk as an order leaving eight euros.
Two exceptions, which I stand by. First: a loss-making SKU used as a launch ramp for a range, long enough to build the sales history and the reviews that will lift the others. That gets decided with an end date written down somewhere, otherwise the ramp becomes a conveyor belt. Second: a traffic-driver that triggers multi-item baskets, where you look at the order and not the line.
Outside those two cases, keeping a SKU that does not pay its rent means handing catalogue space to a channel you do not own. Better to be subtle about it: you do not leave a marketplace, you select what deserves to be on it.
In the crucible: what is left once the platform has helped itself
A marketplace is a remarkable accelerator and a till with a false bottom. Both at once, on the same sale. French e-commerce was worth €196.4 billion in 2025 according to the Fevad annual review, €76.1 billion of it on products alone. Nobody walks away from that flow on principle.
But flow is not margin. Every SKU has to go through the crucible, let the commission, the fixed fees, the shipping and the advertising burn off, and show what settles at the bottom. On some, there is gold. On others, you ran the warehouse all year for lead.