How Do I Know If My Meta Ads Are Actually Working?
"Is Meta even working? We're spending a lot."
Every founder I talk to asks some version of this. Usually right after the monthly spend crosses a number that makes them nervous, and usually with a follow-up: "The dashboard says it's working. I just don't feel it in the bank account."
Here's the uncomfortable truth behind that feeling: you cannot answer this question from inside Meta's dashboard. Not because Meta is lying, exactly, but because Meta is grading its own homework. The ROAS number in Ads Manager counts sales Meta believes it influenced, using rules Meta wrote, measured in windows Meta chose. Some of those sales would have happened anyway. Your best customer who buys every month and happened to scroll past an ad? Meta counts her. She was coming regardless.
So the platform number isn't fake. It's just answering a different question than the one you're asking. You want to know "is this spend making me money I wouldn't have made otherwise?" Meta answers "how many sales can I plausibly associate with an ad?" Those are cousins, not twins.
The good news: you can get much closer to the real answer with three checks any owner can run. No new software. No agency. An afternoon, a calculator, and a little honesty.
Check 1: The napkin math (blended ROAS)
Forget attribution entirely for a minute. Take a month. Divide total store revenue by total ad spend across every platform. That's your blended ROAS, sometimes called MER, and it's the most honest number you own because nothing can inflate it. It's just money out and money in.
Now do it for the last six months and put the numbers side by side. You're looking for one thing: when spend went up, did blended revenue follow proportionally? If spend doubled and blended ROAS held steady, ads are probably pulling weight. If spend doubled and blended ROAS sagged while Meta's in-platform number stayed gorgeous, you just found the gap between the story and the bank account.
One number to know before you judge yours: what blended return you actually need to make money, after product costs, shipping, and everything else. Plenty of merchants celebrate a 3x platform ROAS while their real break-even is 2.8x blended and they're quietly losing money on every "win."
Check 2: What happened when the ads stopped?
At some point your ads have paused. A billing hiccup, a creative refresh, a decision to "take a break after BFCM." Go find that week in your revenue.
If revenue barely moved when ads went dark, a chunk of your ad spend was taking credit for demand that exists anyway: your email list, your repeat buyers, people searching your brand name. If revenue dropped hard, your ads are genuinely feeding the machine.
This is the poor man's incrementality test, and it comes with honest caveats: seasonality muddies it, one week is a small sample, and ad effects have a lag. Don't treat it as proof. Treat it as a smell test. But when a founder tells me "we paused for two weeks in February and nothing happened," that is a finding, and it usually costs them nothing to discover because the pause already happened. You're just finally reading the result.
If your ads have never been paused, even briefly, that's worth noticing too. It usually means nobody has ever wanted to know the answer badly enough to risk finding out.
Check 3: Who is the money coming from?
Open your store analytics and split revenue into new customers versus returning customers, then look at the trend over six months.
Paid ads have one job that justifies their cost: bringing in people who did not already know you. If your ad spend has climbed but new-customer revenue is flat, your ads are increasingly "harvesting." They're showing up in front of people who were already yours, taking credit, and charging you for the introduction. Your email platform and your product were going to close that sale for a fraction of the cost.
Growing spend plus growing new-customer revenue is what working actually looks like. Growing spend plus flat new-customer revenue is an expensive victory lap.
What "working" actually means
Run all three and you'll land in one of three places.
The numbers agree with Meta: blended holds up, pauses hurt, new customers are growing. Great. Your ads work. Spend with confidence and stop reading your dashboard with one eye closed.
The numbers disagree with Meta: blended is sagging, pauses were painless, new-customer revenue is flat. Your ads need restructuring before they need more budget, and every dollar you add right now is buying credit, not customers.
The numbers are unreadable: revenue figures that don't match between platforms, no clean way to split new from returning, nobody sure what's connected to what. This is the most common result, and it's its own diagnosis. You don't have an ads problem yet. You have a measurement problem, and no amount of media buying skill fixes a broken scoreboard.
That third case is most of what I find inside stacks. The ads question turns out to be a trust question, and the trust question turns out to be a plumbing question. Which is fixable. It's almost always fixable. But you have to look.
You now have the afternoon version. If you want the deep version, the one where someone verifies the tracking, reconciles the numbers, and puts a dollar figure on every leak, that's the work I do. Either way: stop setting budgets on a number nobody has ever checked.
Sound familiar?
Take your stack's temperature.
The Tech Stack Temp Check: 27 questions, 3 minutes, no logins. Find out if your stack is running cool, warm, or hot — free, and mildly judgmental.