The ROAS in Ads Manager isn't your profit. How to calculate break-even ROAS

Most conversations about ads start with the same question: "What's our ROAS?" It's a good question. The problem is that the answer comes from a place that cannot know the truth.

The short version, for anyone who wants the answer fast: the ROAS reported by Meta or Google counts the conversions the pixel sees, not the money that lands in your bank account. Delivery refusals, returns, discounts and tracking errors can move the real number by 20–50% in either direction. The figure that matters is break-even ROAS: average order value divided by gross margin per order. Below it, you lose money on every sale, whatever colour the report is.

Three situations from accounts I've managed this year show why.

Case 1: refused parcels don't show up in ROAS

A supplement brand in Romania. April: the platforms report over 10,000 RON in sales. Accounting says just under 8,000. The difference: 18.6% of parcels were refused at delivery. The customer ordered, the courier called, the customer changed their mind.

To Meta and Google, each of those orders is a successful conversion. The pixel fired on the thank-you page and the story ended there. For the business, every refusal means shipping out, shipping back, stock tied up for two weeks and zero revenue.

Platform ROAS: 1.08. ROAS on money actually collected: under 0.9. Same account, same month.

Case 2: the sales that never happened

Same industry, different account. When I took over reporting, I found roughly 5,000 RON of "sales" in a single month that matched no real order. Test orders, duplicates from form resubmissions, Purchase events firing on the checkout page instead of the confirmation page.

Nobody lied. Tracking had been set up in a hurry and nobody had checked it. Three months of budget decisions made on numbers inflated by 40%.

Case 3: two truths for the same store

A fashion store selling in four countries. In the Romanian market, Google Ads reported a ROAS of 1.0. GA4, on the same data, showed 5.07.

GA4 was right. But the campaign was optimising against the wrong number, because conversions weren't flowing back into Google Ads. The algorithm saw a campaign losing money and treated it accordingly: lower budgets, timid bids, shrinking volume. A profitable store, sabotaged by its own tracking.

Why platform ROAS differs from real profit

The platform counts what it can see. A click, a thank-you page, a cart value. It does not see:

• delivery refusals and returns
• the shipping cost you absorb
• the discount applied at checkout
• marketplace commission, if you also sell there
• duplicate or test orders
• conversions blocked by ad blockers and declined consent

Each of these pulls the real number in a different direction. Some inflate the report, others understate it. The outcome is the same: you make decisions on a number that isn't yours.

How to calculate break-even ROAS

The one figure every online store owner should know before opening Ads Manager: below what ROAS do you lose money on each order?

A full example with round numbers: - Average order value: 250 RON - Product cost (purchase or production): 90 RON - Shipping, packaging, payment processing: 25 RON - Delivery refusal rate: 15%, which works out to about 6 RON lost per delivered order (round-trip shipping on refusals, spread across the orders that stick)

Gross margin per order: 250 − 90 − 25 − 6 = 129 RON

Break-even ROAS = average order value ÷ gross margin per order = 250 ÷ 129 = 1.94

In other words: at a ROAS of 2.0 you make 4 RON per order. At 1.8, however green the platform's report looks, you lose money on every sale.

Then add the fixed costs you want ads to cover (salaries, rent, subscriptions, your own time) and you have your target ROAS. For the store in the example, it usually lands between 2.5 and 3.5. One client with fashion margins needed 6. Another, selling supplements with high margins, was profitable from 1.7.

There is no such thing as a good ROAS in general. There is only ROAS above or below your number.

What to do with the number: harder than it looks

This is the part the "5 steps" articles skip. Knowing your break-even is mandatory. Acting on it correctly is something else.

Target ROAS doesn't work everywhere. The temptation is to set tROAS on every campaign at your target and wait. It doesn't work like that. The strategy needs conversion volume to learn, typically 30–50 per month per campaign. Below that, a strict tROAS chokes delivery and the campaign stops spending altogether. I've seen an account where an agency switched overnight from Target ROAS to Maximize Conversions, with no test. ROAS dropped from 4.5 to under 2 and took months to recover. The reverse change, made just as abruptly, does the same damage.

When to stop a campaign is not a formula. Two weeks below break-even could mean a bad campaign. It could also mean a slow season, an attribution change, a competitor cutting prices, or a new campaign still learning. I had an account where ROAS jumped from 3.5 to 5.2 in three days after a conversion-goal fix. If someone had paused it a week earlier, it would have looked like a perfectly rational decision.

Products can't be treated the same. A product with a 70% margin is profitable at a ROAS of 1.5. One with a 30% margin needs 3.5. If they sit in the same campaign, the algorithm sends the money to whichever converts more easily, not whichever earns more. Separating them sounds simple, but it changes the account structure and resets learning.

What I'm getting at: the number is easy to calculate, the decisions around it are not. Each of the situations above looks exactly the same in Ads Manager for the first few days. The difference is made by someone who has seen the pattern before, knows which question to ask before pressing pause, and has no reason to show you the report in a particular light. Not necessarily the person running the campaigns. Someone who sits next to you while you read them.

The basic check you can do today

One exercise. Take last month and put three numbers side by side: sales in Ads Manager, sales in your e-commerce platform, and actual money collected in your bank account after refusals and returns.

If they differ by less than 10%, your tracking is good and your store is healthy. If they differ by 30–50%, every budget decision of the past few months was made on a wrong number.

Most owners I've worked with had never made that comparison. Not out of neglect. Nobody had told them they should.

If you're not sure where you stand, the 2-minute test here tells you whether your problem is measurement, audience or execution. For most, it's the first. (Note: the diagnostic is currently available in Romanian only.)