Full Account Audit

Variables

Replace each placeholder before you run the prompt.

• {{account_summary}}High-level facts about the account: vertical, monthly spend, number of campaigns, conversion model.

  B2B industrial laser markers, ~$45K/mo spend, 12 campaigns, offline conversions imported via OCT.

• {{observations}}Raw observations from the account — what you see in the UI. CTRs, CPLs, conversion rates, search-term notes, structural facts.

  CTR 4.2% account avg, CPL $312, 18% search-term contamination, broad match in 6 of 12 campaigns, no negative-keyword list shared across campaigns…

You are a senior B2B paid media auditor. Your job is to diagnose this account ruthlessly and propose fixes — not validate the current setup.

ACCOUNT SUMMARY:
{{account_summary}}

RAW OBSERVATIONS:
{{observations}}

Produce a structured audit covering, in this exact order:
1. Account structure (campaigns, ad groups, naming conventions)
2. Match types & keyword strategy
3. Negative keyword discipline
4. Bidding strategy & conversion model
5. Ad copy & extension coverage
6. Landing page alignment
7. Reporting & attribution

For each section:
- Score 0–10
- List specific issues, ranked by severity (critical / high / medium / low)
- For each issue, propose a concrete fix
- Estimate expected impact (e.g. "10% CPL reduction within 30 days")

End with a Top 3 prioritized action list — what to do first, this week.

Do NOT pad the response. Skip sections that genuinely have no issues. If a section has insufficient data, say so explicitly rather than guessing.

Expected output shape

A 7-section audit with numeric scores, severity-ranked issues, concrete fixes, and a Top 3 action list at the end. Roughly 800–1500 words.

Why we wrote it

Most "AI audit" prompts produce generic best-practice lists. This one forces section-by-section severity rating and Top 3 prioritization, which is what a senior auditor would actually deliver.

How to use

1. Open Claude or ChatGPT. The recommended model for this prompt is claude-opus-4-7 — opus when the prompt requires deep reasoning, sonnet for the rest.
2. Replace every {{variable}} with content specific to your account. The examples above are starting points, not templates to ship as-is.
3. Paste the prompt and run.
4. Read the output against the expected shape above. If the model produced a structurally different response, re-prompt rather than accept the drift.