agentleFS
vidual-labsvidual-labs/digital-marketing-skills/skills/marketing/google-ads-diagnostics/SKILL.md

google-ads-diagnostics

Skill3 starsChanged 45 days agoMITView on GitHub

Use when analyzing Google Ads performance data and recommending optimizations — Search campaign structure, keyword health, search terms and negatives, Quality Score, responsive search ad strength, assets, Smart Bidding strategy and budget scaling. Also use when the user shares a Google Ads export, screenshot or report and asks why CPC is high, conversions dropped, or how to scale. Don't use for generating keyword lists (google-ads-keywords), landing page fixes (landing-page-funnel), or conversion tag bugs (gtm-debugging).

The file

We split this skill into the rules it gives an agent

6 rules. About 0 are common, 5 are distinctive, and 1 is very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈5

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 1

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Don't use for: keyword list generation (use google-ads-keywords), landing page conversion problems (use landing-page-funnel), conversion …

    line 29
  2. Distinctive:

    Before optimizing, confirm the conversion data is real:

    line 64
  3. Distinctive:

    Add as negatives at the right level — account shared list for universal junk, campaign for topic, ad group for cross-theme routing

    line 100
  4. Distinctive:

    Check asset group performance by channel (Search, Shopping, Display, YouTube, Discover, Gmail, Maps) in the campaign report and search …

    line 168
  5. Distinctive:

    Add brand exclusions so PMax does not buy your brand queries at inflated cost

    line 169
  6. Distinctive:

    Add campaign-level negative keywords for junk queries surfaced in the search terms insights

    line 170

≈ “Common” is an estimate: at least 1 in 5 of the 364,980 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Google Ads Performance Diagnostics

Overview

Turn Google Ads performance data into a prioritized list of concrete changes: structure fixes, keywords to pause or scale, negatives to add, Quality Score work, ad and asset improvements, bid strategy and budget moves. Every recommendation names the current state, the change and the expected effect, in an order that avoids resetting Smart Bidding learning.

When to Use

  • Reviewing campaign, ad group, keyword or search term reports
  • Diagnosing high CPC, low CTR, falling conversions or rising CPA
  • Deciding whether a campaign is ready for Target CPA / Target ROAS
  • Planning budget increases without destabilizing delivery
  • Reviewing Performance Max or Demand Gen campaigns at the level the reports allow

Don't use for: keyword list generation (use google-ads-keywords), landing page conversion problems (use landing-page-funnel), conversion tracking or consent bugs (use gtm-debugging), or Google Shopping feed issues.

Inputs

Request at least 14 days of data (30 preferred) at these levels:

LevelMetricsWhy
AccountSpend, conversions, conversion value, CPA/ROAS, conversion-tracking statusHealth and trust in the data
CampaignType (Search / PMax / Demand Gen / Shopping), bid strategy, budget, spend, impressions, clicks, CTR, CPC, conversions, CPA, ROAS, Search impression share, "Limited by budget" statusEfficiency and headroom
Ad groupSpend, CTR, CPC, conversions, CPATheme relevance
KeywordMatch type, impressions, clicks, CTR, CPC, conversions, CPA, Quality Score with its three componentsIndividual performance
Search termsQuery, match type that triggered it, clicks, cost, conversionsWasted spend, negative candidates
AdsAd strength, impressions, CTR, conversions, asset performance labelsCopy quality
ConversionsWhich actions are "primary", attribution model, whether values are importedWhether optimization targets are right

Data rules: work only from the user's numbers. If a level is missing, say which phase you could not evaluate rather than guessing. Benchmarks below are industry starting points; the account's own history beats them. Use the account's currency. Google renames features often (extensions became assets, Enhanced CPC was retired, Dynamic Search Ads are migrating into AI Max) — if a report label differs, map it to the nearest concept and note it.

Benchmarks (Search, starting points only)

MetricStrongAverageWeak
CTR> 8%4–8%< 3%
Quality Score8–106–7≤ 5
Search impression share> 80%50–80%< 50%
Absolute top impression share> 50%25–50%< 25%
Conversion rate> 8%3–8%< 2%
CPA vs target≤ 0.8×0.8–1.2×> 1.5×

Industry averages differ widely (legal and B2B SaaS convert at 2–4%; e-commerce brand terms at 10%+). Prefer the account's own trend over these numbers.

Diagnostic Framework

Phase 0: Trust the data first

Before optimizing, confirm the conversion data is real:

  • Are the right actions set as primary conversions? Micro-conversions (page views, button clicks) set as primary inflate volume and mislead Smart Bidding.
  • Is enhanced conversions or offline conversion import in place for lead-gen? Without it, Smart Bidding optimizes toward form fills, not sales.
  • Did conversion counts change suddenly on a date that matches a tag or consent change? If so, route to gtm-debugging before touching bids.

Phase 1: Account and campaign structure

CheckProblemFix
Campaign types mixedSearch, Shopping and PMax competing for the same queriesKeep Search for controlled keywords; let PMax cover Shopping/Display/YouTube; use brand exclusions and campaign-level negatives in PMax so it does not cannibalize brand Search
Too many campaignsBudget fragmented, most campaigns limited by budgetConsolidate to 3–8 Search campaigns grouped by goal, geography or margin
SKAGsSingle-keyword ad groups create management overhead with no Quality Score benefit under close variantsMove to single-theme ad groups of 5–20 related keywords
Brand and non-brand togetherBrand CPA masks weak non-brand performanceSeparate brand campaign with its own budget and bid strategy
Location / language mismatchSpend in regions the business does not serveCheck location options ("Presence" vs "Presence or interest"), exclude unserved regions
Search partners / Display expansion onLow-quality traffic on Search campaignsTurn off unless the report shows profitable partner conversions

Phase 2: Keyword health

SegmentDefinition (30 days)Action
Wasted spendSpend > 2× target CPA, 0–1 conversionsPause, or tighten match type and add negatives if the search terms show salvageable intent
UnderperformersCPA > 1.5× target with ≥ 3 conversionsLower bids (manual) or move to a separate campaign with a looser target; check landing page match
WinnersCPA ≤ 0.8× target and ≥ 5 conversionsAdd exact-match variants from their search terms; ensure not limited by budget or low impression share
Zero impressions30+ days, no impressionsPause; the query space is empty or bids too low — do not keep for "completeness"
Low Quality Score (≤ 5)Any spendDiagnose component (Phase 4) before pausing
CannibalizationSame query served by multiple keywords / ad groupsKeep the best-performing keyword, add cross-ad-group negatives

Phase 3: Search terms and negatives

Review search terms weekly; this is where budget leaks. Note that Google hides low-volume queries for privacy, so the report only shows part of spend — if "Other search terms" is a large share, phrase match and broad match are the usual cause.

Process:

  1. Sort last 30 days by cost descending
  2. Flag irrelevant intent (jobs, DIY, free, unrelated products, wrong location)
  3. Add as negatives at the right level — account shared list for universal junk, campaign for topic, ad group for cross-theme routing
  4. Promote converting search terms that are not yet keywords to exact match

Negative keyword behaviour: negatives match misspellings automatically but not plurals or synonyms — add those yourself. Negatives work at campaign and account level for Performance Max too; use them when PMax captures brand or junk queries.

Phase 4: Quality Score

Quality Score (1–10) is a diagnostic built from Expected CTR, Ad Relevance and Landing Page Experience, each rated Below average / Average / Above average. It is not used directly in the auction, but the same components drive Ad Rank, so low scores mean higher CPC for the same position.

Component below averageLikely causeFix
Expected CTRAd does not match the query's wording or promisePut the keyword theme in at least two headlines; add a concrete benefit or number; test a stronger offer
Ad relevanceAd group too broad for one adSplit ad groups by theme; tighten keywords to the ad's wording
Landing page experienceSlow page, weak relevance, poor mobile, thin contentMatch H1 to the ad, pass Core Web Vitals, make the next step obvious — hand off to landing-page-funnel

Prioritize Quality Score work by spend: a QS 4 keyword spending 30% of budget matters more than ten QS 4 keywords spending nothing.

Phase 5: Ads and assets

Responsive search ads (RSAs) are the only standard Search ad type. One strong RSA per ad group, two at most.

CheckStandardFix
Ad strength"Good" or "Excellent"Add headlines (up to 15) and descriptions (up to 4); vary length and angle; include keyword themes
PinningMinimalPin only legally required text; heavy pinning lowers Ad strength and limits combinations
Asset performance labels"Best" / "Good" over "Low"Replace "Low" assets after they have ~5,000 impressions
Sitelinks≥ 4 with descriptionsAdd deep links to pricing, categories, reviews, contact
Callouts≥ 4Short proof points: free shipping, 24h response, certified
Structured snippets≥ 1Types, brands, services
Other assetsCall, location, image, price, promotion, lead form where relevantAdd where the funnel supports it
Automatically created assets / AI MaxUnderstand what is onIf the account uses AI Max for Search or automatically created assets, review the generated headlines and final-URL expansions and exclude pages that should not receive paid traffic

CTR diagnostics: headline does not mirror the query → put the theme in H1; no differentiator → add a number or proof; weak CTA → replace "Learn more" with the specific next step; no assets → add sitelinks and callouts first (cheapest CTR win).

Phase 6: Bid strategy

Current strategies: Manual CPC, Maximize clicks, Maximize conversions (optionally with a Target CPA), Maximize conversion value (optionally with a Target ROAS), Target impression share. Enhanced CPC was retired in 2024–2025; accounts still showing it were moved to Manual CPC.

SituationRecommendedWhy
New campaign, < 15 conversions / 30 daysMaximize conversions without target, or Manual CPC if budget is very smallSmart Bidding needs conversion signal before a target is useful
15–30 conversions / 30 days, CPA roughly knownMaximize conversions with Target CPA set at or 10–20% above recent actual CPAA target below recent reality throttles delivery
30–50+ conversions / 30 days, values trackedMaximize conversion value with Target ROASVolume is enough for value-based bidding
Lead quality variesImport offline conversions or assign conversion values; then value-based biddingOtherwise the system optimizes toward cheap, low-quality leads
Visibility goal (brand defence)Target impression shareOnly for brand campaigns
Pure traffic goalMaximize clicks with a max CPC limitRarely right for conversion accounts

Smart Bidding hygiene:

  • Change targets in steps of ≤ 20% and wait 1–2 weeks between changes
  • Budget changes under ~20–30% do not reset learning; large changes, target changes and conversion action changes do
  • Seasonality adjustments exist for short, predictable spikes (sales, launches) — use them instead of yanking targets
  • Portfolio bid strategies let you set max CPC limits on Target CPA / ROAS campaigns

Phase 7: Budget and scaling

ScenarioReadAction
"Limited by budget" and CPA at or below targetProfitable headroomRaise budget 20–30% per week; watch CPA
Search impression share < 50%, lost IS (rank) highBids or quality too lowFix Quality Score; loosen target 10–20%
Search impression share < 50%, lost IS (budget) highBudget is the capRaise budget or cut waste first
Budget not spent, CPA fineTarget too tight or audience smallLoosen target; add phrase variants; widen location
High spend, poor ROIWasteNegatives, pause wasted-spend keywords, restructure before adding budget

Scaling rule: increase budgets by at most ~30% per week on Smart Bidding so the system adapts without re-entering learning.

Phase 8: Performance Max and Demand Gen (when present)

  • Check asset group performance by channel (Search, Shopping, Display, YouTube, Discover, Gmail, Maps) in the campaign report and search themes insights
  • Add brand exclusions so PMax does not buy your brand queries at inflated cost
  • Add campaign-level negative keywords for junk queries surfaced in the search terms insights
  • Confirm the final URL expansion is allowed only for pages that should convert
  • Demand Gen: judge on view-through and engaged-view conversions, not last-click alone; use lookalike segments from first-party lists

Output Format

GOOGLE ADS AUDIT: [Account / Campaign name]
Date range: [Start] – [End] | Currency: [XXX]
Spend: [X] | Conversions: [Y] | CPA: [Z] (target [T]) | ROAS: [R] (if tracked)
Data confidence: [High / Medium / Low — what is missing]

--- HEALTH SCORE: X/10 ---
[One-line rationale]

0. CONVERSION TRACKING
   [Primary actions correct? Enhanced conversions? Anomalies?] → [Fix]

1. STRUCTURE
   [Issue] → [Change] → [Expected effect]

2. KEYWORDS
   Pause / reduce: [keywords with spend, conversions, CPA]
   Scale: [keywords with CPA ≤ 0.8× target]
   Cannibalization: [query → keywords competing]

3. SEARCH TERMS → NEGATIVES
   Account list: "term", "term", [exact term]
   Campaign [name]: "term", "term"
   Promote to exact: [converting search terms]

4. QUALITY SCORE
   [Keyword / ad group] — [component below average] → [Fix]

5. ADS & ASSETS
   [Ad strength, pinning, low assets] → [Rewrite / add]
   Missing assets: [sitelinks, callouts, snippets, ...]

6. BID STRATEGY
   [Current] → [Recommended] → [Why] → [When to change next]

7. BUDGET & SCALING
   [Current allocation] → [Change] → [Expected effect]

8. PMAX / DEMAND GEN (if applicable)
   [Channel mix, brand exclusions, negatives, URL expansion]

PRIORITY CHANGES (implementation order)
1. [Highest impact, lowest risk — usually negatives and tracking fixes]
2. [...]
3. [...]

ESTIMATED IMPACT: [Range, with assumptions stated]
RE-CHECK DATE: [Date, usually 14 days]

Common Pitfalls

  1. Optimizing on broken conversion data. A tracking change that doubled conversions makes every bid decision wrong. Verify Phase 0 before anything else.
  2. Setting Target CPA below what the account has ever achieved. The system responds by not spending. Start at recent actual CPA and walk it down 10–20% at a time.
  3. Changing bid strategy every week. Smart Bidding needs 1–2 weeks and enough conversions to settle. Decide on a cadence and stick to it.
  4. Reading the search terms report as if it were complete. Low-volume queries are hidden. Large "other" spend means match types are too loose, not that the data is clean.
  5. Treating Quality Score as a KPI. It is a diagnostic. Fix the component that is below average on keywords that spend; ignore QS on keywords with no traffic.
  6. Pinning every headline in an RSA. It turns the RSA back into a fixed ad and caps Ad strength. Pin only what legal or brand rules require.
  7. Letting PMax eat brand queries. Brand CPA looks great in PMax while non-brand Search starves. Use brand exclusions and compare blended results.
  8. Optimizing for clicks. High CTR with no conversions is expensive curiosity. CPA, ROAS and conversion quality are the only north stars.

Verification Checklist

  • Conversion tracking sanity-checked (primary actions, enhanced/offline conversions, anomalies) before recommending bid changes
  • Audit based on ≥ 14 days of data, or a stated caveat if less
  • Every phase that had data was evaluated; missing data called out explicitly
  • Negatives listed with match type and the level (account / campaign / ad group) to add them
  • Keywords to pause/scale cited with their actual spend, conversions and CPA
  • Quality Score fixes name the component that is below average
  • RSA ad strength, pinning and missing assets reviewed
  • Bid strategy recommendation matches conversion volume (15 / 30 / 50 thresholds) and includes a wait period
  • Budget scaling stays within ~30% per week on Smart Bidding
  • PMax brand exclusions and negatives addressed when PMax is present
  • Priority changes ordered to avoid resetting learning
  • Estimated impact given as a range with assumptions, in the account's currency

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.