Meta Ads · 9 min read
The 7-angle Meta ads account audit
An account audit is not a list of metrics. It is seven questions asked in a fixed order, where each answer rules out the next. Done in order, most audits stop early: the first broken link explains the symptom, and the fix is that link, not a new creative and a new budget on top of a tracking problem.
The order below is the one a senior buyer uses and the one AdsBud's agent runs when you ask it to audit an account: signal, structure, delivery, creative, measurement. Meta's own definitions are quoted where they exist; the thresholds are working practice and are marked as such.
Angle 1, tracking: is the signal real?
Everything downstream depends on this, so it comes first. Open Events Manager before Ads Manager and confirm the conversion events the campaigns optimise for are still firing. A broken event looks exactly like a dead campaign.
Then know which attribution setting the numbers are on. Ads Manager's ad set setting offers click-through within 1 or 7 days after a link click, view-through within 1 day after an impression, and engage-through within 1 day after a non-link-click action on the ad. Reporting also offers 28-day click. Two 2026 changes matter here. On January 12, 2026, the Ads Insights API stopped returning the 7-day view and 28-day view attribution windows, so any dashboard built on those columns lost conversions that day while Ads Manager itself was already counting without them. And in March 2026 Meta renamed engaged-view attribution to engage-through and cut the video threshold from 10 seconds to 5, so engaged-view counts moved with the definition.
If the account uses incremental attribution, note Meta's description: it optimises delivery for incremental conversions using models that predict whether a conversion is caused by an ad, and Meta says results cannot be compared across ad sets with different attribution models. Do not compare an incremental ad set with a standard one and call it a drop.
Angle 2, structure: can the ad sets learn?
Look for ad sets competing for the same people, and for ad sets too thin to leave learning. Meta's delivery system wants around 50 optimisation events in the week after the last significant edit. An ad set shows Learning limited when it is unlikely to get there, and Meta lists the causes: small audience size, low budget, low bid or cost control, high auction overlap, an infrequent optimisation event, or running too many ads at the same time.
Meta's own first fix is consolidation: combining ad sets and campaigns helps you get the results you need faster. Its public performance guidance says the same thing, combine ad sets and minimise changes during this phase, and pairs it with developing a variety of ad creatives so the system can deliver the right message to the right people. Any advice you read that attributes those rules to a named Meta ranking system is commentary; the advice itself is Meta's.
While you are here, check what restarted learning recently. In Meta's words, a significant edit is when you pause your ad set or make a change to the optimisation event, audience or creative; bid strategy and budget changes may also be significant depending on the magnitude. Meta publishes no percentage for budget changes, so treat the common 20 percent rule as practice, not policy. The account's activity history in Ads Manager shows who changed what, and when.
Angles 3 and 4, delivery, placements and countries: where did the money go?
Delivery first. Is spend pacing to budget? Meta recommends checking the status of your ad at all levels if you are experiencing delivery issues, and the statuses tell you why: Learning, Learning limited, In review, Rejected (the ad does not comply with Meta Advertising Standards), Scheduled, Completed, Off, plus Error and Warning states. Three quiet killers: a failed payment pauses ads until the balance is paid; an account spend limit that has been reached stops delivery; and a bid cap set too low shows the bid limited status, where Meta's advice is to increase or remove the bid control. A cost per result goal, which Meta's older pages still call cost cap, may not spend the full budget once you hit the cap, and Meta's help center says the daily budget should be at least five times that cap.
Then placements and countries. Pull the breakdown by placement and by country for the last 7 days and since launch. By default, Meta chooses placements for you, and one Audience Network format, rewarded video, shows full-screen videos to people who choose to watch them in exchange for an in-app reward. On one real account about 97 percent of a video campaign's spend went there, and about 87 percent of another campaign's budget went to one of four target countries, while the top-line metrics looked healthy. For Awareness, Traffic and Engagement campaigns, placements can still be set at the ad set; for Sales, Leads and App promotion, Meta is removing ad-set placement choice and offers account-level placement exclusions by objective, rolling out gradually.
Angles 5 and 6, audience, frequency and creative: fatigue or auction?
Meta defines frequency as the average number of times each Meta Account saw your ad, calculated as impressions divided by reach, and notes the metric is estimated. Read it with CPM and CTR per ad, 7 days against since launch, because the three together tell you what kind of problem you have.
As working practice, not a Meta rule: CPM up with CTR and frequency steady means the auction got more expensive, and the ad is fine. Frequency up with CTR down and CPM up on the same ad means the same people keep seeing it and ignoring it; that is fatigue, and the fix is new creative, not a bigger budget. CTR down with frequency flat means new people are not responding either; that is the message, not the wear.
The audit output for these two angles is a short list: which ads are tiring, which audiences are saturating, and which problems are simply the market getting more expensive this week.
Angle 7, economics: against the target, not against last week
Meta defines cost per result as the total amount spent during the selected time period divided by the number of results. The audit question is not whether it went up, but whether it is above the target the business set: the most a result is worth. An account can be cheaper than last week and still losing money, or more expensive than last week and perfectly healthy.
Two caveats before any scale or cut decision. Platform-reported results are attributed by the platform; reconcile them against the business's own revenue or lead counts before a big move. And the attribution setting from Angle 1 decides what the results column even means.
All seven angles are what AdsBud's agent runs when you say audit this account. It pulls today, the last 7 days and since launch, reads placements, countries, frequency and per-ad numbers, and compares cost per result with the target stored in the account's business profile. When it proposes a change, it cites those numbers; you approve, and one click undoes it. Every change across every connected account lands in one activity log with who decided it and why.
Frequently asked
Why audit in this order instead of starting with the ads?
Because each angle rules out the next. A tracking problem looks like a creative problem; a learning reset looks like an audience problem; an attribution change looks like a performance drop. If you start with the ads you fix the most visible symptom and keep the cause. Stop at the first broken link.
Which of these thresholds are Meta rules and which are practice?
Meta rules, quoted from its help center: around 50 optimisation events in the week after the last significant edit; the significant-edit list; the statuses; the five-times-your-cost-cap budget guidance; the metric definitions. Practice, not policy: the 20 percent budget-step habit, the fatigue-versus-auction reading of CPM, CTR and frequency, and any frequency ceiling you set for yourself.
Did Meta delete historical attribution data on January 12, 2026?
Meta's wording is that the 7-day view and 28-day view windows will no longer be available and will return no data in the Ads Insights API, for all versions. It does not say history was deleted, and Ads Manager was already consistent with the remaining windows. Say the windows are no longer returned for any date range, and compare like with like when you report.
See what AdsBud catches on your account
AdsBud runs this check for you around the clock, flags the leaks, and proposes each fix for your one-click approval. Read-only by default, cancel any time.
Keep reading
Tools we have compared
- AdsBud vs AdAmigoAdAmigo is an AI media buyer for Meta ads with creative generation and bulk launching; AdsBud is the transparent, approve-first alternative that also covers Google Search.
- AdsBud vs PixisEnterprise-grade AI advertising platform that auto-optimizes bids, budgets, and targeting across Meta and Google, with a generative creative suite and a conversational layer.
- AdsBud vs AdStellarBetter for teams who want chat-first, approve-first management of live Meta and Google Search ads, with every change justified by fresh plus historical data, rather than high-volume AI creative generation.