How to audit a Meta Ads account

Step 1

Configuration — establish what the account is trying to do

Start by reading the conversion event from each ad set's configuration, never from its name. Group ad sets by event. An account mixing purchase, lead and registration campaigns needs three separate KPIs, and blending them produces an audit that is wrong from the first table.

Watch for a registration or add-to-cart event sitting on a purchase objective. That is sometimes a genuine misconfiguration and sometimes a third-party tracker routing real purchases through a different event — opposite conclusions, so confirm which before writing it up as a finding.

Step 2

Tracking — before you judge any performance number

Compare purchase event volume against its own daily baseline. If events dropped while traffic held, the measurement is broken and every performance number downstream is unreliable. Auditing performance on top of broken tracking produces confident, wrong recommendations.

• Pixel and CAPI both firing, with event volume at its usual level
• Attribution window — note what it is set to, and whether it changed in the window you are auditing
• View-through share, especially on retargeting
• Duplicate conversion counting where two campaigns target the same audience

Step 3

Structure — how the account is organised

Check budget type, bid strategy and learning status. The most common structural finding is a bid strategy that is not actually constraining anything — a cost cap set at three times the real CPA, or a target ROAS set at half what the account already achieves. Flag those: the account is effectively running unconstrained while appearing controlled.

• CBO or ABO per campaign, and whether the budget sits where you think
• Ad sets that never exited learning — our gate is spend of at least 2× AOV or 4× CPA
• Ad sets with too little signal to judge at all, below 3–5 conversions a week
• Audience overlap between ad sets competing in the same auction

Step 4

Delivery — read against the account's own baseline

Pull 90 days to establish normal, then compare the audit window against it. CPM, CTR and frequency, always with spend as context. The goal is not to grade the account against the industry but to find what has drifted from its own recent history.

Step 5

Creative — the ads carrying the spend

Restrict to the ads making up the first 90% of cumulative spend. For each, check the trend rather than the period total, and look specifically for frequency rising alongside CTR falling over 14 or more days, which is the fatigue signature. For video, split hook rate from hold rate — they fail for different reasons and need different fixes.

Before recommending any pause, check whether the ad is the top converter in its ad set. If Meta has concentrated budget on one ad and that ad produces most of the conversions, that is the algorithm working, not a problem to fix.

Step 6

Segments — where delivery actually landed

Check placement and position, device, and age and gender. The recurring finding is delivery outside the intended target when Advantage+ audience expansion is on, consuming budget at poor conversion rates. Quantify the gap rather than noting it qualitatively.

Remember what you can and cannot do with the finding. Breakdowns are not budget levers — you cannot shift spend between placements or age brackets. The available actions are building a new ad set with specific targeting, or applying exclusions.

Step 7

Waste and dependencies

Finally, list what is spending without producing, and check what each cut would break before recommending it. An ad set past twice the ad set's CPA target with zero conversions is a defensible pause. An ad set with one or two conversions at high CPA is insufficient data, not a verdict.

Map the funnel before cutting. Prospecting campaigns feed the audience pools that retargeting converts, so pausing a prospecting campaign often raises retargeting CPA within one to two weeks. Say so in the audit rather than discovering it afterwards.