Meta Ads
How to diagnose a Facebook ad account in under fifteen minutes
Short answerSkip the individual ads at first. Check that the sales Ads Manager reports match your payment processor, then read blended CPA (total spend divided by total purchases) month over month against target and breakeven. Then check frequency on your main concepts, walk the chain from click to purchase, sweep the settings, and sort your last ten ads into real concepts versus variations. That’s the whole Facebook ad account audit, and it tells you which layer is broken.
Most people start a Facebook ad account audit in the wrong place. Somebody opens Ads Manager, sorts by cost per purchase, and starts pointing at the worst ad on the screen.
That’s the slowest way to find the problem, and it’s how good ads get killed. Individual ad numbers are the most misleading data in the whole account.
When I look at an account, I want to know one thing first: which layer is broken? The tracking, the market, the traffic, the settings, or the ads themselves. Here’s the order I check them in, with rough times, so you can run it yourself with a coffee.
Minutes one to three: make sure the numbers are real
Pull the last 30 days of purchases from your payment processor and put that next to the purchases Ads Manager reports for the same window.
They won’t match exactly. The data inside Facebook is always a little skewed, which is why I keep my own count. What you’re looking for is a gap that suddenly got wider, or reported purchases that dropped while real sales didn’t. That’s a tracking problem, and there’s no point diagnosing ads off broken numbers.
While you’re there, check the value your purchase event is sending. If it doesn’t match what people actually pay (and the backend usually isn’t in it at all), your ROAS column is telling you a story, and you’ll make the wrong call off it.
Minutes three to six: read blended CPA as a trend
Blended CPA is total ad spend divided by total purchases across the whole account. Not per ad. The account.
Look at it month over month. Is it going up, going down, or flat? That trend is your truth. A few bad days inside a healthy month is a dip. A climb over several weeks is a real problem.
For lookback windows, I start with all time, then 30 days, then maybe 7. Most businesses run about a month at a time, and a 7-day view can hurt you if your buyers take two or three weeks to convert. My rough rule for how long buyers take: low ticket, mostly within the first five to seven days. Book-a-call, more like seven to fourteen.
Then put that blended number against two lines, not one. Your target CPA (what you’d love to pay) and your breakeven CPA (where you stop making money once you count bumps, upsells and backend). An account can be over target and still above breakeven, which isn’t winning, but isn’t bleeding either. Know which one you’re in before you change anything.
Minutes six to nine: check frequency and engagement
Frequency is the average number of times each person has seen your ad (Meta calculates it as impressions divided by reach). Look at it on your main concepts over the last 30 days.
My rough read is that above about 2.5, I’m watching it closer, and somewhere around 3 to 3.5 I’m thinking about turning that concept off for a bit. That depends on the audience, though. A complex audience that needs a lot of convincing will run higher frequency before it buys, so I expect it to climb. A simple mass-market offer, I get concerned sooner.
Then look at engagement over time. Comments, shares, click rate, week against week. One of the cleanest signs a concept has worn out its pocket is engagement sliding while CTR holds. People are still clicking out of habit, but nobody’s reacting.
If blended CPA is rising, frequency is climbing and engagement is fading, you probably don’t have an ad problem. You’ve fished that part of the lake. I wrote more about telling those apart in how to tell ad fatigue from a broken offer.
Minutes nine to twelve: walk the chain after the click
Now read the funnel as a chain. Link clicks, landing page views, checkout visits, purchases. Divide each by the one before it and find the link that got weaker.
One signal I lean on: if your cost per landing page view is rising faster than your cost per click, the quality of the traffic is slipping, even when click volume looks healthy. And if clicks held but landing page views dropped, something is wrong with the page or the link, not the ad.
If people land fine and just don’t buy, the leak is on the page or in the offer. I walked through this chain step by step in what to do when your CPA doubles but CTR stays the same, so I won’t repeat all of it here.
Minutes twelve to fourteen: sweep the settings
This part of a Facebook ad account audit is boring and it catches more problems than it should.
First, what is the campaign optimizing for? Whatever you ask Facebook for, that’s what you get. If you’re selling a low ticket product and the campaign is optimized for leads or engagement, you’ll get people who like to click and comment, and a sales campaign can’t turn them into buyers.
Second, open a few ads and check the creative enhancements. Some time ago I had an account that was doing exceptionally well and then one day it just flopped. A friend asked me whether text improvements was off. Sure enough, that little box had been flipped on. I flipped it off, CTRs went through the roof, cost per click got better, and I got a better quality customer at a lower price. I turn all of those AI enhancements off by default now, and I re-check them, because they have a habit of coming back on.
Third, look at targeting. If there are interests stacked three deep, that’s worth a note. In almost every account I run, location, age and gender is all the targeting I use, and the creative does the rest.
Minute fifteen: count your real concepts
Last one. List your last ten ads and sort them into two columns. Genuinely different concepts (a different kind of person, a different problem, a different moment) and creative variations of the same idea (new image, new hook, same person).
Most people who think they’re running ten concepts are running two or three with variations stacked on top. If that’s you, and your frequency is climbing, you’ve found the reason the account feels heavy. New images won’t fix it. A new concept aimed at a pocket you haven’t reached will.
And notice what isn’t on this list: killing ads. I only kill an ad when its reported CPA is three to four times my target and its engagement is weak too. A high CPA with strong CTR is often an ad feeding the ones that get the credit, and nothing in fifteen minutes justifies pulling it.
One caveat on all of this. If the Facebook ad account has less than two full days of data, none of this audit means much yet. Wait.
What to do this week
Block fifteen minutes on your calendar for a Facebook ad account audit and run this exact order once: processor versus Ads Manager, blended CPA trend against target and breakeven, frequency and engagement on your top concepts, the click-to-purchase chain, a settings sweep, and the ten-ad concept sort. Write one sentence at the end naming which layer is broken. Only then open an individual ad.
What is a good frequency for Facebook ads?
It depends on the audience. For a simple mass-market offer I start watching closely above about 2 to 2.5 over 30 days. Around 3 to 3.5 I’m usually ready to rest a concept. Complex audiences that need more convincing often need higher frequency before they buy, so judge it alongside blended CPA and engagement.
How do I calculate blended CPA?
Divide total ad spend across the whole account by total purchases for the same period. Use purchases from your payment processor if Ads Manager is undercounting. Compare the trend month over month, and judge it against both your target CPA and your breakeven CPA, not against any single ad.
Why did my Facebook ad account suddenly stop working?
Check tracking first, then settings like creative enhancements and the optimization event, then whether anything changed after the click, such as price or page. If none of those moved and frequency has been climbing for weeks, the pocket of buyers your concepts reach has likely been worked through, and you need new concepts.
P.S. Keep a simple sheet with the blended CPA for each month. It takes two minutes to update, and it’s the one number that stops you from panicking over a bad Tuesday.
