Google Ads Search Terms Review

Read first: - google-ads/references/operator-thesis.md - google-ads/references/query-patterns.md - google-ads/references/intent-map.md - google-ads/references/negatives-playbook.md - google-ads/references/deliverable-templates.md

Read workspace if available: - workspace/ads/account.md - workspace/ads/goals.md - workspace/ads/intent-map.md - workspace/ads/queries.md - workspace/ads/negatives.md - workspace/ads/learnings.md

Data Acquisition

Connected Mode (MCP available)

Pull via the search tool on google-ads-mcp:

Primary: Search terms report — last 30 days, by spend:

SELECT
  search_term_view.search_term,
  search_term_view.status,
  campaign.name,
  ad_group.name,
  metrics.impressions,
  metrics.clicks,
  metrics.cost_micros,
  metrics.conversions,
  metrics.conversions_value,
  metrics.cost_per_conversion
FROM search_term_view
WHERE segments.date DURING LAST_30_DAYS
ORDER BY metrics.cost_micros DESC
LIMIT 500

Retrieval ladder — if the primary query returns no rows, follow the shared retrieval ladder in data/search-term-retrieval.md. The ladder walks through account-wide → search-only → campaign enumeration → campaign-scoped classic → PMax campaign_search_term_view → limited visibility. Each step labels its retrieval_mode and the diagnostics shape is defined in that doc. Report the retrieval mode in the output header. In pmax-fallback mode, present query rows but do not fabricate cost/CPA analysis. In limited mode, shift to campaign/asset-group/tracking analysis and request a UI export.

Supplementary: High-spend zero-conversion terms (waste hunt):

SELECT
  search_term_view.search_term,
  campaign.name,
  ad_group.name,
  metrics.cost_micros,
  metrics.clicks,
  metrics.impressions
FROM search_term_view
WHERE segments.date DURING LAST_30_DAYS
  AND metrics.conversions = 0
  AND metrics.cost_micros > 10000000
ORDER BY metrics.cost_micros DESC

Supplementary: Keyword view (cross-reference with targeted keywords):

SELECT
  campaign.name,
  ad_group.name,
  ad_group_criterion.keyword.text,
  ad_group_criterion.keyword.match_type,
  ad_group_criterion.status,
  metrics.impressions,
  metrics.clicks,
  metrics.cost_micros,
  metrics.conversions,
  metrics.cost_per_conversion
FROM keyword_view
WHERE campaign.status = 'ENABLED'
  AND ad_group.status = 'ENABLED'
  AND segments.date DURING LAST_30_DAYS
ORDER BY metrics.cost_micros DESC
LIMIT 200

Why keyword_view matters for search-terms analysis: - Shows which targeted keywords triggered the search terms you're reviewing - Reveals match type expansion: a broad match keyword "recycling" triggering "beer can recycling near me" - Identifies keywords that are generating disproportionate waste (the keyword is the problem, not just the search term) - Informs whether the fix is a negative keyword or a keyword match type change - Cross-reference: if a wasteful search term comes from a single broad-match keyword, narrowing or pausing that keyword may be better than adding negatives

Supplementary: Existing negatives (to avoid duplicates):

SELECT
  campaign.name,
  campaign_criterion.keyword.text,
  campaign_criterion.keyword.match_type,
  campaign_criterion.negative
FROM campaign_criterion
WHERE campaign_criterion.negative = TRUE
  AND campaign_criterion.type = 'KEYWORD'

See data/gaql-recipes.md for additional queries.

Date Range Fallback

If LAST_30_DAYS returns 0 rows or <$5 total spend, fall back to LAST_90_DAYS, then all-time (no date filter). Always state the date range used in the output. See the main skill's Date Range Fallback Protocol for details.

Export Mode (no MCP)

Ask the user for: - Search Terms report: last 30 days - Columns needed: Search term, Campaign, Ad group, Impressions, Clicks, Cost, Conversions, Conv. value - Sort by Cost descending - Include at least top 200-500 terms

Also request existing negative keyword list if available.

See data/export-formats.md for recommended format.

Process

1. Announce mode (connected/export).
2. In connected mode, run the shared retrieval ladder (data/search-term-retrieval.md). Report the resulting retrieval_mode in the output header.
3. If retrieval mode is pmax-fallback, present query rows but do not fabricate cost/CPA analysis. If limited, shift to campaign/asset-group/tracking analysis and request a UI export.
4. Review terms by spend, conversions, CPA/ROAS, and recurring modifiers when those metrics are available.
5. Group terms into meaningful clusters (buyer intent, comparison, informational, junk, branded).
6. Cross-reference against existing negatives — don't re-recommend what's already excluded.
7. Cross-reference against keyword_view when keyword rows exist — identify which targeted keywords triggered wasteful search terms. If a single broad-match keyword is responsible for multiple waste clusters, recommend narrowing/pausing that keyword alongside (or instead of) adding negatives.
8. Cross-reference against the Intent Map — update it if new patterns emerge.
9. Identify: - Keep/scale — high-intent, converting, efficient - Isolate — different intent class, needs its own bucket - Exclude — clear waste, no plausible path to conversion - Watchlist — ambiguous, needs more data
10. Extract messaging clues from high-value language (feeds RSA recommendations).
11. Recommend safest negative match type + scope where warranted.
12. Use the deliverable templates for operator summary + negative recommendations.
13. Update workspace memory files.

Draft Output

Negative Keyword Draft

Trigger: 3+ clear waste terms identified with combined spend > $50 (or any single term > $25 waste).

Create a draft using drafts/templates/negative-draft.md: - Write to workspace/ads/drafts/YYYY-MM-DD-[account-slug]-negatives.md - Include every recommended negative with: keyword, match type, scope, reason, spend wasted, collateral risk - Update workspace/ads/drafts/_index.md

RSA Draft

Trigger: Clear buyer language patterns identified that differ from current ad copy.

Create a draft using drafts/templates/rsa-draft.md: - Write to workspace/ads/drafts/YYYY-MM-DD-[account-slug]-rsa-refresh.md - Include specific headlines/descriptions derived from converting query language - Update workspace/ads/drafts/_index.md

Always update workspace memory:

• workspace/ads/queries.md — notable patterns and clusters
• workspace/ads/intent-map.md — if new intent classes emerge
• workspace/ads/negatives.md — recommended negatives (even before draft approval)
• workspace/ads/findings.md — strategic observations

Output Shape

1. Account Status block — account name, CID, status, date range used, tracking confidence, mode
2. Data summary (terms analyzed, date range, total spend covered)
3. Retrieval mode note: - Classic Search mode — full search-term metrics available - PMax fallback mode — query rows available, but term-level metrics may be limited - Limited visibility mode — Google exposed insufficient query detail; shift to inference and operator next step
4. Cluster analysis (intent groups with performance)
5. Waste identification (specific terms, amounts, patterns)
6. Signal identification (buyer language, emerging opportunities)
7. Isolation opportunities (intent that deserves its own bucket)
8. Messaging clues (language for RSA recommendations)
9. Confidence assessment
10. Drafts created (with paths and summaries)
11. Memory updates

Rules

• Do not recommend broad negatives casually.
• Do not call something junk just because it has low volume.
• Distinguish poor execution from poor intent.
• Favor cluster-level interpretation over row-by-row trivia.
• Always produce a negative draft when waste is clear. Do not leave actionable negatives buried in analysis prose.
• Cross-check existing negatives before recommending — duplicates waste operator trust.