Skip to content

Open-source SEO, GEO, and marketing skills for AI agents.

19.3 KB
Raw Blame

PPC Optimization Pattern Playbook for NotFair Skills

Research date: 2026-05-29
Scope: generalized paid-search operating patterns distilled into reusable Google Ads workflows for NotFair skills.

1) Pattern inventory overview

  • Coverage: search term management, n-grams, negatives, broad match, RSA/ad testing, Quality Score, bidding/budgets/impression share, conversion tracking, Search Partners, account audits, assets, landing pages, and PMax/Search overlap.
  • Most reusable pattern families:
    • Search term + n-gram mining for negatives, new keywords, ad-group routing, and product/content insights.
    • Negative keyword hygiene: inconsistent coverage, conflicts, malformed terms, root-word negatives, and shared-list governance.
    • Broad match readiness and containment: only use when measurement, bidding, budgets, landing pages, and search-term review can support wider matching.
    • RSA testing: test business outcomes, not just Google ad strength/CTR; control ad-group granularity and pinning where message-to-query alignment matters.
    • Quality Score diagnosis: break down expected CTR, ad relevance, and landing-page experience; interpret trends with changes in ads, landing pages, targeting, budgets, and broad match expansion.
    • Impression share and budget/rank triage: separate lost IS due to budget from lost IS due to rank, then map to budget, bid, quality, asset, and targeting interventions.
    • Conversion tracking audit: validate action consistency before trusting automated bidding, ad tests, negative decisions, or budget scaling.
    • Search Partners evaluation: segment by network, compare profitability and search-term quality, then disable or split/monitor when partner traffic diverges.
    • Audit automation: automate deterministic checks first, escalate judgment-heavy findings with evidence and likely financial impact.
    • Landing-page experience and ad/LP testing: align message, trust, navigation, speed, and conversion action; test full message chains when ad and page interact.

2) Top repeatable patterns

  1. Aggregate before judging sparse queries

    • Individual search terms often lack enough clicks for decisions. Roll up one-, two-, and three-token n-grams across queries, then sort by spend, clicks, conversions, CPA/ROAS, bounce-like engagement if available, and campaign/ad-group distribution.
    • Default candidate filters: conversions = 0 sorted by cost; or high-click/no-conversion terms (example threshold: clicks > 150). Treat these as review triggers, not automatic negatives.
  2. Classify bad search terms by action, not just performance

    • Irrelevant intent -> add negative keyword.
    • Relevant but wrong ad group/campaign -> add exact/phrase keyword to correct group and/or negative-route from wrong group.
    • Relevant but poor message fit -> rewrite RSA/landing page or split ad group.
    • Repeated relevant demand not sold/offered -> flag product/service/content opportunity.
  3. Use negatives as steering controls for automation

    • Smart bidding and looser match types can keep spending on poor intent even after lots of clicks. Negatives are how the manager constrains the machine.
    • Watch question words, research phrases, consumer-vs-B2B modifiers, jobs/free/DIY, geography mismatch, and duplicated search terms served from lower-converting ad groups.
  4. Govern negative keyword scope deliberately

    • Account-wide disqualifiers belong in shared negative lists linked consistently to campaigns.
    • Campaign/ad-group negatives should be reserved for routing, brand/non-brand separation, or product-line differences.
    • Audit for conflicts where a negative blocks a positive keyword or high-value search term.
    • Remember negative match types are not the same as positives: plurals/misspellings often require explicit variants.
  5. Broad match requires a readiness gate

    • Before adding broad match, confirm conversion tracking is trustworthy, search-term management cadence exists, landing pages are focused, budget can absorb exploration, and bidding strategy fits the data volume/risk.
    • Prefer broad match for longer buying journeys or when smart bidding has meaningful conversion data; be cautious for short buying cycles, dynamic pages, broad service-list pages, low budgets, or weak conversion signals.
    • Do not assume exact always beats broad: if broad match plus smart bidding converts better because it can use unseen user-journey signals, keep broad and use exact selectively.
  6. Budget increases need opportunity proof

    • If lost IS due to budget is near 0%, more budget is unlikely to add meaningful traffic.
    • If budget is capped, forecast incremental spend/conversions using lost IS, recent CPA/ROAS, and marginal query quality. New budget can be absorbed by worse broad-match inventory, search partners, cannibalization, or operational bottlenecks.
    • Increasing budget without negative/search-term controls can scale waste.
  7. Rank loss has four common levers

    • Quality Score/ad relevance/landing page changes.
    • Bid or target changes, including forgotten bid adjustments after strategy changes.
    • Targeting/match-type expansion changing auction mix.
    • Asset coverage and expected impact.
  8. Diagnose conversions as traffic × conversion rate

    • Conversion change comes from click volume change, conversion-rate change, or both.
    • Isolate by campaign/ad group/device/network/search term/landing page/date, compare year-over-year seasonality, and overlay recent account changes.
  9. RSA tests should measure the advertiser’s goal

    • Google may serve RSA assets/combinations based heavily on CTR; the advertiser may care about CVR, CPA, ROAS, conversion per impression, revenue per impression, or profit per impression.
    • Use experiments or structured before/after with enough minimum data. Pause losing RSAs only after the chosen metric has sufficient volume and confounders are checked.
  10. RSA structure matters more than asset volume alone

  • A broad ad group with many intents can cause Google to mismatch assets to queries. Split ad groups when query intent requires specific copy or landing pages.
  • Ad strength is a diagnostic of Google’s asset diversity expectations, not a final performance objective. Pinning can reduce ad strength but may improve message control.
  1. Quality Score work should be componentized
  • Expected CTR: compare SERP/message attractiveness and query fit.
  • Ad relevance: ensure query themes appear in ad assets without forcing irrelevant repetition.
  • Landing-page experience: match query/ad promise, improve trust, transparency, navigation, speed, mobile UX, and conversion clarity.
  • Trend patterns matter: cliffs often indicate broken pages/disapprovals/tracking/offer changes; slow declines suggest relevance drift or competitor movement.
  1. Search Partner traffic is a separate diagnostic segment
  • Compare Google Search vs Search Partners for cost, clicks, conversions, CPA/ROAS, conversion rate, search terms, and brand/non-brand composition.
  • If partners are materially worse and volume is meaningful, disable Search Partners; if better, keep but monitor because site mix is opaque and can shift.
  1. Conversion tracking is a prerequisite, not a postscript
  • Audit conversion action status, primary/secondary inclusion, attribution/counting consistency, duplicate tags/actions, imported action ownership, values/currencies, phone/form/offline coverage, and campaign-specific goal consistency.
  • Do this before optimizing bids, budgets, negatives, or ad tests.
  1. Landing page tests can be independent or bundled with ads
  • If testing message chains (e.g., discount vs free shipping), pair ad+LP variants.
  • If testing which ad and which page independently wins, use a full factorial setup: each ad to each page.
  • Geographic/service specificity may improve CTR only, CVR only, or both; measure both.
  1. PMax can obscure Search diagnostics
  • When PMax serves on overlapping queries, Search impression share and search-term opportunity can look different because eligibility/auction priority shifts.
  • Audit overlap between PMax search terms, Search keywords, and Search search terms; compare CTR/CVR/value and use negatives/keyword coverage/routing where controls exist.

3) Recommended NotFair skill modules

Module A — search-term-ngram-triage

  • Primary skill home: google-ads/manage; secondary audit integration.
  • Triggers: “wasted spend,” “search terms,” “add negatives,” “query analysis,” “low conversions,” broad match cleanup, PMax/Search overlap review.
  • Workflow:
    1. Pull search terms for 30–90 days by campaign/ad group/network; include cost, clicks, conversions, conversion value, keyword/ad group, and campaign type.
    2. Tokenize terms into 1-, 2-, and 3-grams; aggregate performance and count distinct queries/campaigns/ad groups.
    3. Generate action buckets: negative candidates, keyword/add-to-exact candidates, route-to-better-ad-group candidates, ad/LP mismatch candidates, product/content opportunity candidates.
    4. Apply safeguards: minimum cost/clicks, ignore brand/protected terms unless user confirms, check existing negatives and conflicts, surface examples for each n-gram.
    5. Recommend or execute bulk negatives/keywords only after user approval and experiment-impact checks.
  • Decision rules: no-conversion high-spend n-grams are review candidates; high-conversion terms missing as keywords are keyword candidates; same query in multiple ad groups with divergent CVR/CPA is routing candidate.
  • Sources: n-gram posts; search query decisions; negative steering; PMax/Search overlap.

Module B — negative-keyword-hygiene

  • Primary skill home: google-ads/manage and google-ads/audit.
  • Triggers: “negative keyword audit,” “conflicts,” “shared negative list,” “irrelevant searches,” “B2B consumer traffic,” “duplicates.”
  • Workflow:
    1. Inventory campaign/ad-group negatives and shared lists.
    2. Detect inconsistent campaign coverage for common disqualifiers.
    3. Detect positive-keyword/search-term conflicts.
    4. Detect malformed negatives and missing singular/plural/close variants.
    5. Recommend shared-list consolidation by scope: account-wide, brand/non-brand, product line, geography/routing.
    6. Produce a safe mutation plan: create/link list, add terms, remove conflicts, or move ad-group negatives to list.
  • Decision rules: use shared lists for universal exclusions; use ad-group negatives for routing; never remove a conflict without confirming the blocked positive/search term is desirable.
  • Sources: negative primer; common negative problems; steering the machine; n-gram posts.

Module C — broad-match-readiness-and-containment

  • Primary skill home: google-ads/manage.
  • Triggers: “should I use broad match,” “Google recommends broad,” “pause exact,” “broad match wasted spend,” “scale campaign.”
  • Workflow:
    1. Confirm conversion tracking health and bid strategy.
    2. Check conversion volume, budget headroom, lost IS, search-term quality, negative coverage, landing-page focus, sales-cycle length, and current exact/phrase winners.
    3. Classify readiness: safe to test, test with guardrails, or do not expand yet.
    4. If testing, create a controlled experiment or limited ad-group rollout; pre-build negatives and monitoring thresholds.
    5. After rollout, compare query mix, CPA/ROAS, CTR, CVR, and incremental conversions; decide whether to keep broad, add exact, or pause exact if broad is better.
  • Decision rules: broad is riskier with short buying cycles, weak tracking, unfocused/dynamic pages, low budget, and insufficient negatives. Exact is not automatically superior when broad uses valuable journey signals.
  • Sources: broad-match readiness; when to pause exact; broad match vs CTR/QS; budget scaling article.

Module D — rsa-ad-testing-lab

  • Primary skill home: google-ads/copy and google-ads/manage.
  • Triggers: “test RSAs,” “ad copy underperforming,” “low CTR/CVR,” “ad strength,” “pinning,” “ad group organization.”
  • Workflow:
    1. Pull RSA assets, asset performance labels if available, ad-level metrics, ad-group query themes, and landing pages.
    2. Identify broad/mixed-intent ad groups where assets cannot match every query theme.
    3. Choose test metric based on business goal: CTR, CVR, CPA, ROAS, conversion per impression, revenue/profit per impression.
    4. Recommend test type: unpinned RSA optimization, pinned message test, theme-vs-theme test, RSA-vs-existing baseline, or ad+LP message-chain test.
    5. Use experiments when possible; otherwise require minimum data and note confounders.
    6. Generate complete RSA replacement assets when mutating; do not rely on ad strength as the winner criterion.
  • Decision rules: pin when legal/brand/query-message control matters; split ad groups when one RSA contains assets for distinct intents; treat high ad strength as coverage signal, not proof of profitability.
  • Sources: RSA testing; RSA organization; effective RSAs; ad strength; ad/LP testing.

Module E — quality-score-component-diagnosis

  • Primary skill home: google-ads/manage; landing submodule for LPX.
  • Triggers: “quality score,” “expected CTR,” “ad relevance,” “landing page experience,” “CPC rose,” “rank lost IS.”
  • Workflow:
    1. Pull keyword QS components and performance; segment by campaign/ad group/match type.
    2. Bucket issues by component: expected CTR, ad relevance, LP experience.
    3. Overlay recent changes: ads, landing page URL, budgets, bids, targeting, broad-match additions, disapprovals.
    4. For expected CTR/ad relevance: recommend ad group split, RSA copy, keyword-to-ad theme alignment.
    5. For LPX: invoke landing-page checklist (message match, trust, transparency, navigation, mobile/speed, conversion clarity).
    6. Show expected impact as rank/CPA opportunity, not guaranteed QS increase.
  • Decision rules: do not optimize visible QS in isolation; prioritize high-spend/high-impression keywords with below-average components and material rank/cost impact.
  • Sources: QS primer/trends; visible vs auction QS; ad relevance; landing-page experience; rank IS diagnostics.

Module F — budget-impression-share-triage

  • Primary skill home: google-ads/manage and google-ads/audit.
  • Triggers: “raise budget,” “limited by budget,” “lost impression share,” “more conversions,” “rank loss,” “top IS changed,” “budget pacing.”
  • Workflow:
    1. Pull campaign metrics, budgets, spend, lost IS budget, lost IS rank, search/top/absolute top IS, CPA/ROAS, search terms, and recent changes.
    2. Classify blocker: no headroom, budget-constrained, rank-constrained, demand/seasonality shift, targeting/match expansion, quality issue, asset issue, PMax overlap.
    3. For budget-constrained campaigns, estimate incremental spend/conversions from lost IS and marginal CPA/ROAS; flag if current extra traffic is low quality.
    4. For rank-constrained campaigns, check QS components, bids/targets, bid adjustments, assets, and target changes.
    5. Recommend budget increases only for campaigns with headroom and acceptable marginal economics; otherwise reallocate from waste.
  • Decision rules: lost IS budget near zero -> budget increase unlikely; budget increase with broad match/search partners can degrade marginal efficiency; PMax overlap can make Search IS incomplete.
  • Sources: impression share analysis; lost IS budget; lost IS rank/top IS; budget changes; more budget same results; PMax IS impact.

Module G — conversion-tracking-integrity-audit

  • Primary skill home: google-ads/audit; prerequisite gate for manage actions.
  • Triggers: first-time audit, “tracking,” “conversions look wrong,” “smart bidding,” “CPA/ROAS unreliable,” “offline conversions.”
  • Workflow:
    1. Inventory conversion actions: type, status, category, primary flag, owner, counting type, attribution, value/currency, ECFL/offline/call coverage where available.
    2. Compare campaign goal settings and included actions for consistency.
    3. Flag duplicate actions, missing primary actions, irrelevant primary actions, inconsistent counting/attribution, imported/read-only actions, missing values/currencies, and suspicious zero/huge conversion patterns.
    4. Before bid/budget optimization, summarize whether data is trustworthy enough for automated bidding.
  • Decision rules: if tracking is materially inconsistent, do not make aggressive bid/budget recommendations; fix measurement first.
  • Sources: conversion tracking mistakes; assisted conversions; audit checklist.

Module H — network-partner-evaluator

  • Primary skill home: google-ads/manage and google-ads/audit.
  • Triggers: “search partners,” “network performance,” “bad traffic,” “CPA spike,” “conversion rate changed.”
  • Workflow:
    1. Segment campaigns by ad network type for Google Search vs Search Partners.
    2. Compare cost, clicks, CTR, CVR, CPA/ROAS, conversion value, and search-term quality.
    3. Identify campaigns where partners are material volume and materially worse/better.
    4. Recommend disabling partners, keeping them, or monitoring with alerts.
  • Decision rules: disable only when partner traffic is meaningful and worse on the account’s goal metric; avoid conclusions from tiny volume.
  • Sources: Search Partners evaluation; conversion-change diagnosis; audit checklist.

Module I — landing-page-message-match

  • Primary skill home: google-ads/landing; connected to copy/manage experiments.
  • Triggers: “landing page experience,” “LPX below average,” “conversion rate dropped,” “test landing pages,” “message match.”
  • Workflow:
    1. Map keyword/search-term themes -> RSA claims -> final URLs.
    2. Score message match, offer consistency, trust/transparency, navigation, page speed/mobile usability, and conversion-action clarity.
    3. Decide test design: ad-only, LP-only, paired message-chain, or full factorial ad×LP.
    4. Recommend page variants and experiment setup; track CTR and CVR separately.
  • Decision rules: if CTR improves but CVR does not, landing page may not reinforce the ad promise; if CVR improves only for local pages, scale by geography/service carefully.
  • Sources: landing-page experience; 10 LP steps; ad+LP testing posts.

Module J — automated-audit-orchestrator

  • Primary skill home: google-ads/audit.
  • Triggers: “audit account,” “health check,” “what’s broken,” “quick wins,” recurring monitoring.
  • Workflow:
    1. Run deterministic checks first: disapprovals, broken/final URLs, paused/limited entities, missing assets, missing conversion actions, negative conflicts, budget/rank lost IS, search-term waste, zero-conversion spend, landing-page issues, PMax overlap.
    2. Group findings by financial impact and actionability.
    3. Separate safe automated fixes from recommendations requiring business judgment.
    4. Produce a prioritized action plan with source metric, evidence, suggested mutation, risk, and expected direction.
  • Decision rules: audits should translate findings into likely spend/revenue impact; automate repeatable scans but keep human confirmation for brand, legal, and strategic exclusions.
  • Sources: ultimate audit checklist; audit automation; asset audit; conversion tracking; all diagnostic modules above.