01 /
Set the scope before reading the numbers
A Meta ads audit is a structured review of delivery, costs and results across an ad account. It should answer three things: what happened, how reliable the evidence is, and what deserves a closer look. It cannot establish incrementality, customer quality or profit from delivery metrics alone.
ZuckerBot’s account audit reads the connected account’s last 30 days of campaign, ad and daily-ad insights. It combines those observations with ad metadata, then produces review flags, creative-fatigue signals and prioritised findings. The browser report shows when it was generated and labels its 30-day scope. That scope is captured at generation; a saved report is a snapshot, not a live dashboard. The audit/API response also includes the exact start and end dates.
- Confirm the account. Check the business and Meta ad account before running the audit, especially if you manage more than one.
- Confirm the goal. A purchase, a lead and a link click are different outcomes. Record what a useful result means for this business.
- Record the context. Note an offer change, tracking interruption, promotion or budget change that could explain the period.
- Check the money units. Use the returned account currency. If currency is unavailable, do not infer it from your location or the dollar symbol.
02 /
Get to a first read-only report
Use Start free, sign in, and connect the Meta ad account you are authorised to review. The audit needs an ad account connection; a Facebook Page is not required for the audit. Page selection belongs to later ad-creation setup.
You can run the audit in the browser. If you prefer a conversation, follow the no-install Claude connection guide and ask for zuckerbot_audit_account. The audit is available on every ZuckerBot tier, subject to the account’s current usage limits.
03 /
Read the report in the right order
1. Coverage and date range
Start with completeness warnings. If Meta returns only part of the data, the observed findings may still be useful, but a clean-looking table does not clear the whole account. Missing evidence is not zero performance. ZuckerBot withholds the account opportunity score when its required inputs are incomplete or there is no assessable spend and ad data.
2. Spend flagged for review
This is historical spend in ads that crossed a review threshold. Read each row’s reason alongside its results. The audit applies a low-CTR flag to traffic, leads, sales and unknown-objective campaigns; it does not apply that same CTR rule to awareness, engagement or app campaigns.
Cost comparisons use the same resolved result type where available: leads against leads, purchases against purchases. Awareness uses CPM; an unknown objective retains a CPL fallback. If Meta reported a different result from the one you intended to optimise, investigate that mismatch before making a budget decision.
The flagged-ad table shows up to 10 of the highest-spend rows. The total can include more flagged ads than are visible in that table. Check the flagged-ad count rather than attributing the entire total to the displayed rows.
3. Fatigue and priorities
A fatigue flag identifies an older ad with a meaningful delivery floor and a decline in click-through rate. It is a diagnostic signal, not proof that people are tired of the creative. Read the fatigue method and competing explanations before proposing a refresh.
The opportunity score, when available, combines flagged-spend and fatigue proportions on a 0–100 scale. A higher number reflects more of these review signals. It is not a performance grade, predicted return, or percentage improvement you can expect.
04 /
Separate the signal from the explanation
| Observed signal | Question to resolve | Next read-only check |
|---|---|---|
| Low CTR, acceptable cost per result | Is the ad still delivering the result that matters? | Compare the actual result type and its cost with like-for-like ads. |
| Clicks, no recorded results | Is this a demand problem or a measurement gap? | Inspect the destination, event reporting and conversion delay before blaming the ad. |
| One campaign receives most spend | Is this a proven performer or an untested dependency? | Review result volume and cost by campaign, with matching objectives. |
| Older ad, declining CTR | Did the creative weaken, or did delivery change? | Compare placement, audience and recent edits before drafting a refresh. |
| Incomplete insights | Which decisions remain unsupported? | Identify the missing data layer; avoid account-wide conclusions. |
These checks may require Ads Manager or business records beyond the audit. Label those inputs as unavailable until you have them. Do not turn a plausible explanation into a reported fact.
05 /
Turn findings into a small decision list
For each priority, record the ad or campaign ID, the observed metric, the comparison period, the reason it matters, and one next check. Add an alternative explanation and the evidence that would change your mind.
- Investigate first: uncertain tracking, incomplete data, an unexpected result type or an unexplained performance shift.
- Draft a test: a supported creative hypothesis with a clearly defined result metric and a bounded comparison.
- Keep observing: an immature or low-volume ad where the evidence is too weak for a decision.
Ranking a finding by spend exposure can help focus attention, but exposure is not the amount you will recover. A recommendation to change an ad needs a separate decision about the exact object, budget and consequences.
06 /
Use a prompt that keeps the audit bounded
After connecting ZuckerBot to Claude, start with this read-only request:
Run zuckerbot_audit_account for my connected Meta ad account. Read-only analysis only: do not create, edit, pause or launch ads, and do not change budgets or settings.
Confirm the account, date range, currency and data completeness. Explain spend flagged for review without treating it as wasted spend or recoverable savings. Check each flag against its campaign objective and actual result type.
Give me the three most useful investigations, the evidence behind each, a competing explanation and the next read-only check. Include the saved report link if one was returned. If evidence or a score is unavailable, say so.
A prompt is an instruction, not a permission boundary. Review tool approvals and disable unneeded write tools in your assistant where available.
07 /
Keep the report, then choose the next step
Return through your account to reopen a saved report when one is available. Reopening a report is different from running a fresh audit. In the assistant flow, report persistence depends on resolving one saved business; if a report link was not returned, do not assume a shareable report exists.
If a session expires or Meta access needs reconnecting, resume at Start free. If the wrong business is selected, correct it before running again. For unresolved connection or report problems, use support and include the error message without passwords or access tokens.
Share report links only with people you intend to see the account’s findings. When you need more usage or another workflow, check current plans and availability; an audit result does not itself authorise a purchase or an ad change.
Reference / FAQ
Common questions
Will an audit change my Meta campaigns?
The account audit reads Meta data and produces findings. It does not edit ads, pause campaigns or change budgets. ZuckerBot can save the resulting report in its own service; other tools have separate capabilities and permissions.
Does flagged spend mean wasted spend?
No. It is spend in ads that crossed a review threshold. An ad can be flagged and still produce useful results efficiently. Inspect the reason, result type and business context before deciding what to change.
Do I need a Facebook Page to run the audit?
No. The audit needs a connected Meta ad account. A Facebook Page is needed for later ad-creation workflows, not this read-only account review.
Why is my opportunity score unavailable?
The audit withholds the score when required inputs are incomplete or there is no assessable ad and spend data. Review the completeness warning and available findings instead of treating the missing score as zero.
Next / Your account
Bring the next question to Claude.
Connect Claude for account questions and supported campaign workflows. Or start with a free, read-only account audit.
Connect Claude Return to my account