Your Meta ads audit will inform you of where your money is going out of the account without purchasing anything and how you can correct it. The following seven audits should be conducted in this sequence: conversion events, click-through rate, budget concentration, frequency, audience, structure and account problems.
I did this on two live accounts that morning. Together they had spent AED 615K in the last 90 days, both got a B, and there was nothing red on the report. Four out of the seven leaks listed below were active on those two accounts. The other three came back clean on them, but you have to test them too since being clean on two accounts doesn’t necessarily mean being clean on yours.
Why a B score can still leak money every day
These two results are both a B grade, scoring 84 out of 100 and 80. On the dashboard, nothing seems to be broken, and that is precisely the issue. You are spending money; the numbers are green, but you do not know whether your agency is doing well or is average.
A score adds up the parts of an account that are set up correctly. It does not show you the single campaign that has been spending for three months with nothing to learn from. That is why I read the findings, not the grade.
Three of the seven checks turned out to be negative for both. I will not try to convince anyone that there were actually any leaks when there were no leaks there at all. Four of these are what I found that day; three of them are the negative ones.
If you only want to know whether your account is wasting spend, start with the first check below. It is the most expensive one in both accounts, and it takes a few minutes in Events Manager.
The two accounts, and how the score breaks down
Account A spent AED 361K and recorded 2,728 results. Account B spent AED 254K and recorded 1,704 results. Same window for both: the last 90 days.
| Area | Account A | Account B |
|---|---|---|
| Budget | 100 | 100 |
| Audience | 100 | 100 |
| Tracking | 100 | 85 |
| Structure | 100 | 100 |
| Creative | 80 | 90 |
The audit splits every account into five areas.
Every finding in Account A, a AED 361K account, was a creative problem. Account B lost its points on creative and tracking, and the 85 on tracking is where its most expensive finding sat.
A perfect area score only means nothing in that area tripped a check. Read each area's findings before you trust its number.
Leak 1: a campaign spending with zero conversions
In Account B, one campaign spent AED 4,645 over 90 days and recorded zero conversions. Not a low number. Zero.
This is almost always a good campaign. This is a campaign that has been designed to take advantage of an event it never gets. Meta requires that the conversion event be reported back to it from your website using either the pixel on the browser or the Conversions API on your server. If the event specified on the ad set is not the one that is fired by your website, the delivery process continues, but there is nothing to optimize for, so it optimizes for the cheapest audience.
- Read the optimisation event
Open the ad set and write down the exact conversion event it is optimising for.
- Check it in Events Manager
Find that same event and confirm it has received activity in the last 7 days.
- Compare names and counts
If the names do not match, or the count is flat, everything that campaign spent bought nothing.
If you send both browser and server events, each pair needs a matching event ID. Without it Meta cannot deduplicate them and your numbers get inflated instead. Meta documents this in its Conversions API deduplication guide.
If your tracking is the problem, fix it before you touch a single ad. A clean conversion tracking setup is what every other check in this list depends on.
Leak 2: campaigns under the click-through floor
Account A had three campaigns under 1% click-through rate. Account B had one.
One percent is the floor my audit uses, and it is not a vanity metric. A low click-through rate means Meta pays more to place each impression, so your cost per thousand climbs, your click cost climbs and the same budget buys fewer conversions.
At 0.58%, about six people in a thousand found the ad worth touching. In that account the audience scored 100, so this was not an audience problem. It was the ad.
- Pull your top performer
Look at what its first two seconds actually say.
- Rebuild the losers around that hook
Fewer, better hooks, not more creative.
- Cut the long-term losers
Switch off anything that has had 10,000 impressions and still sits under 1%.
Leak 3: one ad ate the budget
In Account A, one ad set sent 94% of its spend to a single ad. Nobody chose that.
Meta picks a winner early. Once spend concentrates like that, the other ads never get enough impressions to prove anything. You think you are running a four-ad test; you are running a one-ad campaign with three ads sat in the corner.
Sometimes that is fine, because the winner really is the winner. In this account it was also one of the sub-1% click-through ads, so Meta had locked onto the best of a weak set and stopped looking.
- Sort the ad level by amount spentInside each ad set, see where the money actually went.
- Over 80% on one ad?Either accept it and build new challengers, or move the other ads into their own ad set with their own budget so they get a fair chance.
- Cross-check click-throughIf the winner is also under 1%, the test needs better ads, not more time.
Leak 4: frequency fatigue
One ad set in Account B was running at a frequency of 3.3: the average person in that audience had seen the ad more than three times in the window.
Frequency on its own is not a failure. Frequency plus a falling click-through rate plus a rising cost per result is fatigue, and you can pull all three in the same breakdown report in Ads Manager.
✅ Your two moves
- Put new creative in front of the same audience
- Put the same creative in front of a wider audience
❌ The move that makes it worse
- Raising the budget on the tired ad set
If your results fell off a cliff rather than slowly, the cause may be elsewhere; I cover that in why Meta Ads performance drops.
Leak 5: stacked and stale audiences
This check came back clean in both accounts. In yours, open the audiences list sorted by date created and count the lookalikes. The warning sign is several 1% lookalikes for the same country, built in different months, off custom audiences that are basically the same people.
A 1% lookalike is the top 1% of a country by similarity to your seed. Build several from near-identical seeds and you have several ad sets bidding into the same pool. You pay to compete with yourself, and the conversions that should be teaching one ad set get split between them.
Then read the size column. Rows that say not available, or below 1,000, are lists built to a fixed date months ago. A custom audience from a file is a snapshot; it does not refresh unless you refresh it.
- Keep two lookalikesOne narrow and one broad, built from your strongest seed: purchasers, not page viewers.
- Archive the restDuplicates only split your learning.
- Replace tiny listsAnything under 1,000 people goes back to a source that rebuilds itself.
If you run lead campaigns, my Meta Ads lead generation structure guide shows how I lay out campaigns and audiences.
Leak 6: zombie campaigns in the structure
In the campaigns view over the last 90 days, one campaign read Active and the rest read Off. Dead campaigns do not spend, so this one came back clean too, but it still costs you clarity and signal.
Every time somebody reopens an old campaign to test it again, conversions get split across ad sets that never collect enough events to exit the learning phase, and delivery comes back as learning limited.
- Filter the last 90 days
List every campaign and ad set that has not spent in that window.
- Archive them
The account you read should be the account you run.
- Consolidate
Fewer campaigns and fewer ad sets, each with more conversions to learn from.
Leak 7: outstanding account issues
This one takes thirty seconds. Go to Account Overview, open recent account issues and the Outstanding tab, then check your support cases.
Outstanding means nobody dealt with it. Rejected ads and restricted assets sit there quietly while budget keeps flowing to whatever still delivers.
Check this before you touch anything else. Half the creative fatigue problems I get called about turn out to be half the ads switched off. If the issue is the account itself, read what to do when a Meta Ads account is banned.
The order to run the seven checks, every month
Run them in this order, because each one changes how you read the next. There is no point judging creative on a campaign that is optimising for an event it never receives.
By hand, all seven used to take me about three hours per account. That is why I built a tool for it. Whether you use a tool or a spreadsheet, the order is what matters.
Running Google Ads as well? The same idea applies there; my Google Ads account audit checklist follows the same money-first order.
What a healthy account looks like after the checks
You are not aiming for a perfect score. You are aiming for an account where every dirham has something to learn from and every ad gets a fair chance to prove itself.
Every ad set optimises for an event that fired in the last 7 days.
Nothing with 10,000 impressions sits under 1% click-through.
No single ad takes over 80% of an ad set unless you chose it.
Two lookalikes, no stale lists, nothing reopened from the archive.
If you would rather have someone run this for you, this is the kind of work I do as a Meta Ads specialist in Dubai, and you can see results in my case studies.
Get the 47-point Meta Ads audit checklist
The video goes through these seven live, on the two accounts. For the rest of the account, open the free 47-point Meta Ads audit checklist. It starts with tracking, like this article, and there is no email wall.
Open your most expensive campaign, read its optimisation event, and check that event in Events Manager. If the count is flat, stop there and fix tracking first.
Want a second pair of eyes on your account? Book a call and I will tell you which of the seven is costing you the most.
