How much can you actually spend on advertising?
“What is a good ROAS?” is the most-asked and least-answerable question in e-commerce advertising. There is no industry figure, because the answer is a property of your product, not of your sector. Fortunately it is easy to work out.
Your break-even is your margin
Not approximately. Exactly.
break-even ACoS = contribution margin break-even ROAS = 1 / contribution margin
If 30% of each sale survives goods, fees and fulfilment, then 30% of revenue is the absolute maximum you can hand to an ad platform before the campaign eats everything it earns. That is a break-even ROAS of 3.33.
Which is why a ROAS of 3 has no meaning on its own. At a 40% margin it is comfortably profitable. At 25% it loses money on every order. Same number, opposite verdicts — and the benchmark you read on a forum has no idea which one you are.
What belongs in the margin
Everything that scales with the sale comes out first: the goods, inbound shipping, marketplace commission, fulfilment, payment fees, packaging, and a realistic allowance for returns.
What does not come out is the advertising itself. That is the thing you are measuring against. Subtracting ad cost inside the margin and then comparing the result to ad spend counts it twice, and makes every campaign look worse than it is.
If you do not know your number, the marketplace calculators on this site produce it directly. Run one with the advertising field set to zero; the net margin it reports is what belongs in the ROAS calculator.
One line to keep clean while you are there: VAT. An invoice from an ad platform registered in another EU country arrives without VAT and is reverse-charged on your own return — see reverse charge explained. The figure that belongs in the calculation is therefore the net one, exactly as it is for the margin. Counting a gross expense against a net margin moves your break-even by the VAT rate and always in the wrong direction.
Break-even is a floor, not a target
Advertising at exactly break-even grows revenue and contributes nothing to rent, salaries or software. You are working for the ad platform.
The gap between break-even and target is where your fixed costs live. If contribution has to cover 8,000 a month of overheads before anything becomes profit, your advertising has to clear break-even by enough to leave that behind. The break-even calculator tells you what "enough" means in units.
There are good reasons to run at or below the floor for a while: launching a product, building review volume, defending a listing under attack. All of those are investments with an end date and a budget. Repeat purchase is another: if a customer comes back, the first order can be bought at a loss on purpose — how large a loss is a question for the LTV to CAC calculator, not for a gut feeling. The failure mode is drifting below it without noticing, which is what happens when ROAS is judged against a benchmark instead of a margin.
Average ROAS hides the number that matters
Your reported ROAS is an average across all the spend. The decision you are actually making is about the next increment of budget, and that performs worse than the average almost by definition — the cheapest, most relevant clicks are bought first.
So a campaign averaging ROAS 5 against a break-even of 3.33 is not necessarily safe to scale. The last slice of spend might already be sitting at 3. Doubling the budget can drag the average down below the floor while the dashboard still shows a healthy-looking figure for the month.
The practical test is to change spend and watch what happens to total sales and to ROAS together. If a 30% budget increase brings 10% more revenue, you have found the edge of your profitable range.
Attributed is not the same as incremental
Platforms report the sales they can attribute to themselves. That includes customers who searched for your brand, already intended to buy, and would have bought anyway.
The honest test is what happens to total sales when the campaign is switched off, not what the panel claims while it runs. Sellers who try this are often unpleasantly surprised by how little total revenue moves — and occasionally pleasantly surprised, which is also worth knowing. The subject has a page of its own: ad attribution.
Treat reported ROAS as an upper bound. A campaign that only just clears break-even on attributed revenue is probably losing money on real revenue.
A workable budget rule
- Calculate contribution margin per product, advertising excluded.
- Take the inverse: that is your break-even ROAS. Write it on the campaign.
- Set your target above it by whatever your fixed costs and desired profit require — for most sellers that means roughly 1.5× break-even.
- Review at the product level, never at account level. A profitable account can easily contain a product quietly funding the rest.
- Re-check whenever costs move. A supplier price rise or a fee change lowers your margin and raises your break-even ROAS the same day.
That last point is the one people miss. Break-even ROAS is not a setting you configure once. It moves every time your cost base does, and a campaign that was profitable last quarter can be underwater this quarter without a single change to the bidding.
Frequently asked questions
Anything above 1 divided by your contribution margin, with enough room on top to cover fixed costs. Any figure quoted without a margin attached is not an answer, it is someone else’s number.
Not when calculating break-even ROAS — that would double-count it. Do include it when you calculate overall product profitability, because it is a real cost per unit like any other.
Because the best-performing inventory is bought first. Extra budget goes to broader keywords, less relevant placements and more expensive auctions. Falling ROAS as spend rises is normal; the question is only whether the marginal spend is still above break-even.
TACoS measures ad spend against total sales rather than attributed ones, which makes it the better long-run indicator. If ACoS holds steady while TACoS falls, organic sales are growing and the advertising is doing work beyond the click.