An account audit should establish whether the ads can run, whether their results are measured correctly and whether someone can explain the decisions behind the spend. Use the checklist below to inspect those jobs separately.

Start with access and delivery, then measurement. Performance judgments depend on those checks, so record unresolved issues before grading the buying decisions.

The 12-point checklist

Work through it in order. Each check tells you exactly what passing looks like.

Access

1. Open Business Settings and pull up everyone with access to the ad account. Pass: every person and partner on the list still works with you. No ex-employee, no former agency, no freelancer from two years ago still holding admin.
2. Check the role levels. Pass: the person actually running the account has advertiser or admin access, admin is limited to people who own the business relationship, and anyone who only needs to look has analyst access.

Delivery

3. Check delivery status, schedule and recent spend at campaign, ad set and ad level. Pass: anything expected to be delivering has recent delivery evidence, or a documented explanation for why it has not started.
4. Open the live ads the way a customer sees them. Pass: the price, the offer, and the product in the ad match what the store sells today, and nothing being advertised is out of stock.

Measurement

5. Check the Purchase event against actual recent orders and a controlled test order. Pass: the expected event arrives with the correct value and currency. A store with no orders yesterday need not have a Purchase event yesterday.
6. Inspect browser and server Purchase events where both are used. Pass: events for the same purchase share the identifiers required for deduplication, and test evidence shows one purchase is not counted twice.
7. Reconcile Meta purchases with store orders over aligned reporting periods. Pass: the differences have an explanation based on attribution settings, reporting dates and event definitions. Similar totals alone do not validate tracking.
8. Review Purchase event diagnostics and match quality. Pass: material warnings have been investigated, with the outcome documented. A clean indicator alone is not proof of accurate purchase counts.

Decisions

9. Open activity history for budget, status, schedule and bid changes. Pass: the operator can explain the significant changes and any deliberate decision to hold budgets steady.
10. Review paused and underfunded ads. Pass: notes distinguish operational failures, measured losses and insufficient evidence, and explain what will be tested next.
11. Review scaling decisions. Pass: additional spend has a stated economic target and loss limit; promising ads can also be held steady for a documented reason.
12. Compare approved creative with the launch queue. Pass: each ad has launched, has an agreed launch date or has a documented reason for being held.

Record the evidence behind each pass and the action required for each failure. Passing these checks establishes basic control of the account; it does not prove the strategy is profitable.

Why access comes first

Remove access that no longer belongs to a current employee or provider, using the business owner to confirm who still needs it. Review admin permissions separately from permission to run or inspect ads.

Access also tells you who could have made a change you find later. Match the names in activity history with the people currently responsible for the account.

What the delivery statuses actually mean

Check 3 only works if you can read the statuses. Meta shows a delivery status at the campaign, ad set, and ad level, and they can differ. The short version: Error means the entity cannot run at all. Warning means it runs but delivery is limited. Pending means it is not running yet. Inactive means it is not running. Preparing shows up after significant edits. Learning means Meta is still optimizing delivery, so performance can be less stable for a while.

A rejected ad can sit inside an active campaign while other ads continue spending. Inspect the individual ads to find those gaps.

The measurement check, in plain terms

Test the event path and reconcile definitions before judging the totals. For example, 240 attributed purchases against 250 store orders could still hide an error. A count of 240 against 90 requires investigation, but the gap alone does not identify the cause. Check event duplication, order definitions, currencies, attribution and reporting dates.

If measurement fails, stop scoring the account on performance until it is fixed. The optimization runs on those events too, so bad measurement does not just distort the report. It distorts what Meta learns from.

Judging the decisions, not just the settings

The first eight checks cover access, delivery and measurement. The final four examine how the operator uses that information.

For the biggest moves in the last month, ask what the operator observed, what they changed, why they chose that action and what happened afterward. Activity history verifies the change; the accompanying explanation lets you assess the reasoning.

Compare approved creative with launches. If six ads were delivered and two launched, ask why the other four were held. A stock issue or a planned test sequence is different from an approval forgotten in a folder.

How to audit uneven delivery

Uneven spend in a CBO is not automatically a fault. Campaign budget optimization allocates across ad sets, and delivery within an ad set can also concentrate on a few ads. Check eligibility and the amount of evidence before judging the creative. Zero spend does not establish poor conversion.

Ask what the operator plans to do with underfunded concepts and why. Further testing, replacement or leaving the current allocation alone can each be reasonable. The decision needs a purpose. See the CBO delivery guide for the checks behind it.

You found problems. Now what?

A failed check is a fact, not a diagnosis. Some failures are ten-minute fixes you can do yourself, like removing an old agency's access. Some, like a broken deduplication setup or a dead decision trail, tell you the account needs different hands.

If one specific result is bothering you and you want it looked at before you decide anything bigger, that is what a second opinion on your Meta account is for. If the audit convinced you the person running the account is the problem, the harder question is who replaces them, and I wrote a separate guide on how to hire a Meta ads media buyer for ecommerce.

Run it on your agency's last month

Run the checklist against the last 30 days. Performance access covers much of the review; business permissions and event diagnostics may require the owner or someone with access to those assets.