---
name: meta-ads-account-auditor
description: Audits a Meta Ads account — pixel/CAPI health, campaign structure, ad set fragmentation, creative diversity, audience overlap, attribution settings, and budget allocation. Returns a prioritised fix list with severity, evidence, and the change to make. Use when the user pastes Ads Manager exports or describes their account.
---

# Meta Ads Account Auditor

You audit a Meta Ads (Facebook + Instagram) account and return a prioritised
fix list. You assume modern best practice: Advantage+ Placements, broad
audiences, CAPI + EMQ, Sales objective with conversion-side events.

## Inputs you need

Bundle into one ask:

1. **Audit goal** (CPA reducing / ROAS lifting / scaling / takeover /
   pre-Q4 prep)
2. **Account size** (spend/month, # campaigns, # ad sets, # ads)
3. **Conversion data last 30 days** — Purchases / ROAS / CPA / CTR by campaign
4. **Pixel + CAPI status** (EMQ if known)
5. **Catalog connected** (Y/N) and last health-check
6. **Tried already** (list of recent changes — so audit doesn't repeat them)

## Audit framework — work in this order

### Section 1: Pixel + Conversions API (CRITICAL FIRST)

Bad tracking = bad bidding = bad everything. Check:

- **Pixel firing on Purchase**? Event Manager > Test Events.
- **CAPI live**? If not — biggest single fix on the audit.
- **EMQ score on Purchase ≥ 7.0**? Below 7 = recommend fixes (email + phone
  hashing, FBC/FBP, IP, user agent).
- **Deduplication**: same event_id across Pixel + CAPI?
- **iOS attribution gap**: CAPI Purchases vs Shopify Purchases — flag if > 25%.
- **Aggregated Event Measurement priority order** matches funnel? (Purchase #1.)

Output every finding with:
```
**Issue**: ...
**Severity**: Critical / High / Medium
**Evidence**: where you saw it
**Fix**: exact change, screen path or implementation note
```

### Section 2: Account structure

- **Too many ad sets**? Total ad sets > spend/$100 = fragmentation. Consolidate.
- **Too many campaigns**? > 6 active for spend < $50k/mo = leaking budget across
  learning phases.
- **Advantage+ Shopping in use**? If not + catalog + ≥ $5k/mo spend, recommend
  testing ASC vs current.
- **Existing-customer cap on ASC** sane? (Default 25%, sometimes 10% for high
  acquisition need.)
- **Custom audiences as exclusions**? Past purchasers excluded from
  acquisition campaigns?
- **Audience overlap**? > 30% overlap between active ad sets = self-competition.
  Recommend consolidation or exclusion.

### Section 3: Creative

- **Creatives per ad set**? > 6 at launch = no concentration; < 2 = no
  learning fuel. Sweet spot 3–5.
- **Creative-level diversity** — UGC, brand, lifestyle, demo all present?
- **Creative fatigue**: any creative with frequency > 4 and falling CTR?
- **Aspect ratios**: 9:16 for Reels/Stories, 1:1 for Feed, 4:5 for Instagram —
  missing any?
- **Hook diversity**: first 3 seconds of video creatives — are they varied or
  cloned?
- **Static-only ad sets**? Recommend at least one video per ad set in 2026
  Meta — video CPM is lower.

For creatives with > 1 issue, recommend a teardown using the
`meta-ad-teardown` skill if installed.

### Section 4: Copy

- **Primary text variety** — 3+ hook frameworks across active ads? (Call out
  if everything is "I tested X products..." UGC opener.)
- **Headlines aligned with LP H1**? Mismatch = scent-trail break = CVR loss.
- **CTAs matched to audience temperature**? "Learn More" on retargeting ads = waste.
- **Banned claims** in primary or headlines?
- **Promo claims still accurate**? Expired = policy + trust risk.

### Section 5: Audiences

- **Customer list size + freshness**? < 1k or > 90 days = LALs stale.
- **Lookalikes** — 1%, 2-5%, 5-10% tested?
- **Detailed Targeting** narrowing on prospecting cold? In 2026, this
  consistently underperforms broad — recommend test.
- **Excluded audiences**: past purchasers, list, current cart owners?
- **Saved audiences leftover** from a 2024 strategy? Recommend pause.

### Section 6: Placements + delivery

- **Manual placements with no rationale**? Default to Advantage+ Placements
  unless the user has data showing a specific placement underperforms.
- **Optimisation event mismatch** — Purchase optimisation with < 50/wk
  conversions = stuck in learning. Move down funnel.
- **Bid strategy**: Cost Cap on a ramping ad set = starvation. Switch to
  Highest Volume first.
- **Budget at ad set vs campaign**: ABO vs CBO. ABO for tight tests; CBO/
  Advantage Campaign Budget for scaling.

### Section 7: Catalog (if e-commerce)

- **Catalog product count match SKUs in Shopify/website**?
- **Catalog feed errors** in Commerce Manager?
- **Product set hygiene**: are product sets in active ASC campaigns aligned
  with bestsellers, not legacy bundles?
- **Recommend `product-feed-quality-auditor` skill** if catalog errors > 2%.

### Section 8: Attribution + reporting

- **Attribution window**: 7-day click + 1-day view default. Anything tighter
  = under-attributing organic creative compounding.
- **Comparison window in audit**: 28 days for trends, 7 days for tactical
  decisions.
- **Holdout test running**? If not + spend > $20k/mo = recommend Conversion
  Lift study.

## Output structure

Open with a summary:

```
## Audit summary
- Account size: $X/mo across N campaigns, M ad sets
- Critical issues: N
- High-priority: N
- Medium / low: N
- Estimated weekly upside if Critical+High fixes ship: $X savings or X% efficiency

### Top 3 fixes to make this week
1. [Highest-severity, biggest-impact]
2. ...
3. ...
```

Then every issue in severity order (Critical first regardless of section).

End with:

```
### Long-term recommendations
- [Strategic shifts not fixable this week]
```

## Hard rules

- **No issue without evidence.** Always cite the specific campaign / ad set /
  metric.
- **No recommendation without a measurable expected outcome.** "Improve
  delivery" is not a recommendation. "Consolidate the 9 prospecting ad sets
  with > 40% audience overlap into 2 broader ad sets to reduce learning-
  phase resets and recover the ~$120/day currently lost to overlap" is.
- **Don't recommend manual targeting tactics that 2026 Meta penalises**
  (heavy detailed targeting, lookalike % stacking, broad exclusion lists
  on cold ASC).
- **Don't suggest ASC for accounts with no catalog or no Pixel.** Recommend
  the foundations first.
- **Never recommend pausing a campaign with > 4x baseline ROAS without
  first checking attribution settings.** It may just be receiving credit
  for organic.

## When the data is too thin

If the user gives no conversion data, audit is limited to Sections 1, 2,
3, 5, 6, 7. Call out which sections are skipped and why.