One figure comes up in every conversation about paid advertising: "a ROAS of 3 and you're fine." It's the most expensive half-truth in the business. A ROAS of 3 can be an excellent company or an orderly way to lose money every month, and the difference isn't in the campaign — it's in the margin of what you sell.
The one number to work out before spending a euro
Break-even ROAS — the point where the campaign neither wins nor loses — is the inverse of your gross margin:
Break-even ROAS = 1 / gross margin
If every €100 of revenue leaves you €30 after product cost, shipping, payment fees and the average return, your gross margin is 0.30 and your break-even ROAS is 3.33. At a ROAS of 3, that campaign loses money. Not a little: it loses on every order.
If your margin is 70% — software, services, digital products — break-even sits at 1.43, and a ROAS of 3 is a very good business.
Same number, opposite realities. Which is why copying someone else's figure is worth nothing.
The real margin is usually worse than the spreadsheet margin
When someone tells me "my margin is 40%", they have almost always counted the cost of the product and stopped there. The margin that governs this calculation includes:
- Product cost, landed in your warehouse, inbound freight included.
- Shipping to the customer, at the real carrier cost, not what you charge.
- Payment processing, typically 1.5–2% on cards plus a fixed fee per transaction — and the fixed fee hurts badly on low tickets.
- Returns, spread across every order. An 8% return rate doesn't subtract 8% of margin: it subtracts the lost or refurbished product plus two legs of shipping.
- Tax you don't pass on, which in the Canary Islands has a story of its own.
Strip all that out and a margin that looked like 40% frequently lands at 26%. And break-even ROAS climbs from 2.5 to 3.85.
The mistake of watching only the platform's ROAS
The figure Meta or Google reports isn't your ROAS: it's the ROAS the platform attributes to itself. Three reasons it's almost always inflated:
The attribution window. By default a sale counts if the person saw or clicked the ad in the preceding days. Some of those sales would have happened anyway — and the better known the brand, the bigger that share.
Overlap between campaigns. When prospecting and retargeting run at once, the same sale can be counted in both. Adding up the ROAS figures in the dashboard gives a total that doesn't exist.
What the platform can't see. A customer who discovers the product in an ad, searches the brand name and buys from the organic result doesn't count as a campaign sale, even though the campaign caused it.
The honest comparison is simpler: take total revenue for the month against total ad spend for the month, and compare it with a month when you didn't advertise.
That's MER, or account-level ROAS, and it's the one that pays salaries.
How to decide what to spend
With break-even in hand, the decision stops being a hunch:
- Work out break-even using the real margin, the true one.
- Add a cushion. If break-even is 3.33, the operating target is 4: the gap is what pays for overhead, not just for the product.
- Decide what you're willing to lose while learning. A new campaign needs data before it can be optimised. That learning budget is an investment, not a failure — but it needs a ceiling written down before you start.
- Set the cut-off. "If after 14 days and €X spent we haven't reached a ROAS of 2.5, it stops." Written beforehand, not argued afterwards.
That fourth point is the one almost nobody writes down, and it's what stops a bad campaign from running for three months "because it's still learning".
When ROAS isn't the metric
If you sell something that gets repurchased — consumables, subscriptions, a recurring service — measuring only the first order will make you shut down campaigns that were good. There the calculation runs against customer value over a horizon you can finance: if a customer is worth €180 over a year but you need to collect in 30 days to pay for stock, the useful horizon isn't twelve months, it's whatever your cash can carry.
And if you sell a service that closes manually, the platform's ROAS simply doesn't exist: what it measures is cost per qualified enquiry. The rest is decided by your close rate, which isn't in any dashboard.
What I usually do on a first account review
Before touching a single ad: rebuild the margin with the person who knows the real costs, calculate break-even, compare revenue against spend month by month for the last six months, and find out whether the account actually makes money.
More than once that calculation has ended with a recommendation to spend less rather than more, or to fix shipping before fixing the ad.
It's less sellable than promising a ROAS of 5, and it's what makes the next campaign worth running.
If you're running campaigns and aren't sure of your break-even number, send me your figures and we'll work it out on the first call. If you'd rather see how I work first, it's all in services.