google-search-audit
Daily Google Ads Search morning briefing — compares last 3 days vs prior 3 days, segments brand vs non-brand (with different expectations for each), and surfaces specific recommendations across budget
SKILL.mdskills/google-search-audit/SKILL.md
GitHubGoogle Ads Search Audit — Morning Briefing (Read-Only)
For the Google Ads account the user names, run a daily Search-campaigns operator pass. Compare last 3 days vs prior 3 days. Read-only — output is recommendations only, no mutations applied.
If the user didn't name an account, list their connected accounts and ask which one to use.
Step 1 — Brand vs Non-Brand classification (do first)
For every Search campaign, classify as Brand or Non-Brand before analyzing. Use campaign name, keyword content, and whether top queries contain the brand name.
| Type | Healthy state | Focus |
|---|---|---|
| Brand | IS > 85%, low CPA, query intent clean | Efficiency, waste reduction. Scaling is demand-limited, not budget-limited. |
| Non-Brand | Growth runway in IS, CPA at-or-below target | IS growth, budget scaling, volume expansion. |
Surface recommendations against the correct expectation per type.
Step 2 — Metrics to pull
Per campaign, both 3d windows: CPA vs target, Conversion volume, Search Impression Share, Lost IS (Rank), Lost IS (Budget), Top-of-page rate, Absolute-top %, ROAS, Conversion Value (AOV trend).
Step 3 — Analysis tasks
Budget + efficiency
- Campaign with rising CPA + stable/improving IS → likely demand issue, recommendation: reassess targeting or pause
- Campaign with low IS + strong CPA → suggest budget increase with exact %
- Campaign where CPA improved but ROAS declined → flag AOV / mix-shift issue
Impression share
Lost IS (Budget) > 20%→ suggest budget increase, state %Lost IS (Rank) > 20%→ suggest bid adjustment or relevance work
Search-term waste
- Identify queries spending > 2× target CPA with poor conversion intent
- Suggest exact-match negative keywords (do not add them — recommend only)
Query-level opportunities
- Identify high-converting queries not yet in exact match
- Suggest exact-match keyword additions
Alerts
- Flag any sudden drop > 25% in CVR, IS, conversions, or ROAS — surface as "investigate first"
Output format
Three sections, each a vertical list of recommendation lines:
🛑 CUT / WASTE
Suggest adding negative exact: "[query]" — spending $X at CPA $Y (3.2× target)
🔧 FIX
Suggest bid +15% on [ad group] — Lost IS (Rank) 28%, strong relevance signal
📈 SCALE
Suggest budget +20% on [campaign — Non-Brand] — Lost IS (Budget) 35%, CPA 18% under target
Skip a section entirely if it has no recommendations. End with: _All recommendations — no changes applied._