What is a good ROAS in your sector

"I have a ROAS of 3. Is that good?"

A client asked me that a few months ago, and the honest answer was an awkward one: I can't tell you yet. I'm missing a number of yours, and it isn't a marketing number.

The figure we're talking about — the one dashboards label ROAS — is simply how much you sell for every dollar you put into ads. A 3 means that for every $1,000 you invested, $3,000 in sales came back.

Sounds fine. For half the businesses I know, a 3 is a party. For the other half, it's the slow way to go broke.

The difference isn't the industry. It's the margin.

The number that decides is yours, not your industry's

Selling isn't earning. Of those $3,000 that came in, part was never yours: it's the cost of the product, the supplies, the hour of the professional who did the work.

What's left after that is the only money that can pay for ads.

So your floor — the point where you neither gain nor lose — comes from one simple division: 1 divided by your margin.

At a 50% margin, you break even at 2. At 25%, you need 4 just to get to zero. At 70%, you break even at 1.4 and everything above that is yours.

Put it in money, it reads better. Two businesses, both showing the same 3, both spending $1,000 a month on ads:

The store that resells. It sells $3,000 thanks to those ads. Since it buys the product and resells it, 25% of that sale stays: $750. It paid $1,000 to the platforms. It lost $250 that month, with a 3 on the screen.

The clinic. It sells the same $3,000. Since what it sells is professional time, close to 70% stays: $2,100. Minus the $1,000 in ads, it made $1,100.

Same number. One is growing and the other is quietly eating its working capital.

That's why "is a 3 good?" has no answer in the abstract. If you don't know your margin, you're not measuring your marketing: you're looking at a nice-looking number.

Now, your sector

With your own floor clear, the industry table is good for the second question: whether you're being charged more than everyone else for the same thing.

Before I show it, the fine print, because almost nobody prints it.

These are medians: half of all advertisers sit below them. They don't come from Google or Meta — each vendor builds them from the accounts it manages, and none of them are audited. They're useful as an order of magnitude, not as an official yardstick.

With that said, here's the median return by industry in 2026:

Industry	Google	Meta
Beauty & personal care	6.1x	3.2x
Fashion & apparel	4.8x	2.9x
Home & garden	4.2x	2.8x
Ecommerce (general)	4.0x	2.8x
Automotive	3.6x	2.4x
Food & beverage	3.2x	2.1x
Travel	2.8x	2.3x
Sports & fitness	2.7x	2.2x
Education	2.4x	1.8x
Healthcare & medical	2.1x	1.4x

Look at the last row. Healthcare has the lowest multiple on the table, and not because it's a bad business: it's because margins there are high and the floor sits low. A medical practice at 1.8 can be profitable. A clothing store at 1.8 is giving money away.

There's another table I find more useful for a small business, because it doesn't talk about multiples but about what a customer costs you. It comes from 13,474 search campaigns measured between April 2025 and March 2026, with the methodology published:

Sector	Per click	Of every 100 who arrive, leave their details	Each new customer
Dentists	$8.00	11	$72.97
Physicians & surgeons	$4.76	12	$40.04
Attorneys	$9.87	6	$131.63
Beauty & personal care	$4.62	10	$39.25
Health & fitness	$6.17	7	$67.36
Personal services	$7.17	12	$54.60
All sectors	$5.42	8	$66.69

A new client for a law firm costs more than three times what one costs an aesthetics clinic. Same country, same month, the same Google. There goes the myth of the universal number: nobody is failing, they're just different businesses.

And one figure from that same measurement is worth the rest put together: cost per customer dropped for the first time in five years, and it didn't drop because clicks got cheaper. The opposite — the click went from $2.32 to $5.42 in a decade. It dropped because pages started converting better.

Translated: the ground left to gain is on your site, not in the auction. I learned that the expensive way. My ads had better numbers than the industry standard and each scan still cost me too much. The hole was in the page people landed on, not in the ads.

Three things that make the comparison lie

Before you hold your number up against those tables, here are three common ways to reach the wrong conclusion.

You're comparing your mix against someone else's. Showing an ad to someone who already visited you performs far better than going out to find new people: in online stores, talking to a stranger on Meta runs around 2.2 and talking to someone who's already been on your site, around 3.6. If half your budget talks to people who already know you, your number looks better without your business being better. It's the most common trap, and it's why an "excellent" month doesn't show up in the bank account.

You forget the lag. Whoever buys in September may have seen the ad in August. In slow-decision categories — surgery, a treatment, a lawsuit — comparing this month's spend against this month's sales gives you a number that means nothing. Look at longer windows or you'll panic in the bad month and get cocky in the good one.

You're comparing apples to oranges. Some people calculate on total sales and others on what's left after costs. If your agency shows you a 4 calculated on gross revenue and you compare it against a 2.8 from a table that netted costs out, you're celebrating a gap that doesn't exist.

Before asking whether the number is good, ask whether it's real

Here's what almost none of those tables mention, and what I run into most.

If your sale closes over chat, over the phone or at the counter, the platform never saw that sale. It can't. What it shows you as a result is the last thing it managed to record: a click, a message started, a form. And then it hands you a multiple calculated on that, with total confidence.

In July we scanned 23 sites belonging to small businesses currently running ads. 12 of those 23 couldn't connect a single dollar from Instagram to a customer who actually paid. In dental it was 5 out of 7 with no Meta tracking installed. In physiotherapy and medical centres, 5 out of 5 without the full picture.

That's 23 sites, not a national study. But the pattern repeats and it has a simple explanation: these are sectors where the sale happens in a conversation, not in a shopping cart.

To be fair, one sector came out well: in aesthetics, 5 of 6 were measuring correctly. It can be done. It's just that almost nobody checks until something breaks.

If you're in that group, your number is neither good nor bad. It's fiction. And everything you decide on top of it — raise the budget, kill a campaign, fire the agency — you're deciding blind.

What to do this week

Three steps, in this order. None of them needs an agency.

One. Work out your margin. Of every $100 you sell, how much is left after the direct cost of what you delivered. A rough number is enough to start.

Two. Divide 1 by that margin. That's your floor. Below that number you're losing money even when the dashboard is green.

Three. Check whether what the platform counts as a sale is a sale. If you close over chat and nobody connected that conversation back to the campaign, you know where to start.

Flying blind isn't bad luck. It's a decision you make without meaning to, one month at a time.

Frequently asked questions

What is a good ROAS?
The one that clears your break-even point, which comes from dividing 1 by your margin. At a 50% margin, anything above 2 leaves you a profit. At 25%, you need more than 4 just to break even. There is no single number that's good for everyone.

What's the average ROAS in my industry?
The 2026 medians run from 1.4 in healthcare to 6.1 in beauty, and they change by platform. Use them as an order of magnitude: they're medians, half of all advertisers sit below them, and none of them replace the calculation of your own margin.

Is a ROAS of 3 good?
It depends on what you sell. For a service with a 70% margin, a 3 is a very good business. For a store reselling at 25%, a 3 means losing money every month. Same number, opposite verdicts.

How do I calculate the minimum I need to avoid losing money?
Divide 1 by your gross margin as a decimal: at a 40% margin, 1 divided by 0.4 gives 2.5. That's your exact floor. If you also want to cover rent, salaries and the rest of your fixed costs, aim 25% to 50% above that figure.

Why doesn't the number Meta shows me match my real sales?
Because the platform only counts what it can see. If the sale closes in a chat, on the phone or in person, it never reaches the platform, and the multiple you're shown is calculated on clicks or messages, not on money. The lag between the click and the purchase plays a part too.

If you want to know whether your number is even real, scan your site free: I'll tell you in plain terms what your marketing is measuring today, what's leaking along the way and how your operation looks next to your sector's. No agency and no airplane-cockpit dashboards.