11x ROAS losing money on every order: the maths the platform does not show you

Search Engine Land published the case of an e-commerce account that reported 11x ROAS and lost money on every single order. The whole calculation is worth reproducing, because the calculation is what makes the argument.

The calculation, line by line

The example in the article starts from a £100 order (original figures in pounds):

Stage	Value
Conversion value reported by the platform	£100.00
After 28% returns	£72.00
After stripping out sales tax (VAT)	£60.00
After 63% cost of goods	£22.20
After fulfilment and fees	£8.70
After media cost (11x ROAS = £9.09)	–£0.39

A loss of 39 pence per order, with a dashboard showing 11x.

Where the calculation goes wrong

First: the platform measures gross revenue, not contribution. The conversion value that gets sent up to Meta or to Google is the order value — with tax in it, before returns, with no cost of goods and no shipping. None of those four things is a detail: in the example, they eat 91% of the value before media even enters the picture.

Second: aggregated ROAS hides the brand campaign. In this case the brand campaigns were running at roughly 18x and absorbing most of the budget, pulling the average up. The author's argument: someone who types your brand name into Google has, as a rule, already decided to buy from you. That sale would largely have happened anyway — the brand ad charges for it and takes the credit.

It is worth noting that this second part is the author's interpretation, not measurement. The way out of guesswork is an incrementality test: switch brand off over a controlled window and measure total revenue, not the dashboard's.

What you can apply tomorrow

1. Send conversion value as net contribution, not as revenue. If integrating cost of goods by SKU is not feasible right now, an average margin by category already corrects most of the distortion.
2. Separate brand from non-brand before you look at any number. As long as the two sit in the same report, ROAS tells you nothing about acquisition.
3. One objective per SKU. Profit, volume or cash recovery. A SKU with three objectives has none, and the campaign inherits the confusion.
4. Treat returns as a line of media cost. A category with 28% returns and a category with 4% cannot carry the same ROAS target.

The number that replaces ROAS

There is no universal ROAS target — there is break-even ROAS, and it belongs to each individual store:

break-even ROAS = 1 ÷ contribution margin

A contribution margin of 20% means a break-even ROAS of 5x. Below that, every sale takes money out of the bank, and the more the campaign scales, the faster it drains. Above it, there is something left over. It is that number — and not a market benchmark — that decides whether the campaign is doing well.

The equation behind this is the same one in the article E-commerce Maths, now seen from the cost side. If you do not know yours, the campaign is not the problem — and that is where performance management begins.

Source: Search Engine Land — The 11x ROAS account that lost money on every order.