P
Knowledge Pack Files
Performance Marketing Knowledge Pack Files
Browse the source files that power the Performance Marketing MCP server knowledge pack.
Available free v1.0.0 LLM
$
sidebutton install marketing Marketing Skill Pack v1.0 — Content Provenance
Visual map of what was generated from domain knowledge vs adapted from public research.
Legend:
[GEN]— Generated from scratch (domain expertise, no specific source)[RES]— Adapted from specific public research finding[VAL]— Generated, then validated/confirmed by research
Root & Manifest
skill-pack.json .......................... [GEN] SideButton format
_skill.md ................................ [GEN] Context protocol, taxonomy, module catalog
ATTRIBUTION.md ........................... [RES] ~50 research sources compiled from 21 agent runs
Roles
_roles/media-buyer.md
├── Campaign lifecycle ................. [GEN] Standard media buying process
├── Campaign principles ................ [GEN] Industry fundamentals
├── Quality Gates ...................... [RES] claude-ads — 3x Kill Rule, budget floors
│ [RES] google-ads-skills — CEP protocol
├── 70/20/10 budget heuristic ......... [RES] claude-ads — budget allocation rule
└── Output format ...................... [GEN] Original deliverable structure
_roles/analyst.md
├── Hypothesis-driven analysis ......... [RES] McKinsey/Stratechi — MECE principle, Issue Trees
├── Anomaly detection (3-tier) ......... [RES] Improvado — 10-20%/20-40%/40%+ thresholds
│ [RES] AdsScripts — z-score 2σ/3σ method
├── Four Levers diagnostic ............. [RES] Pixis — Audience/Creative/Bids/Budget MECE model
├── Reporting cadence table ............ [RES] 2POINT Agency, Sidekick Accounting
├── Analysis principles ................ [GEN] Industry fundamentals
└── Output format ...................... [GEN] Original deliverable structure
Module: paid-search
paid-search/_skill.md
├── Campaign hierarchy ................. [GEN] Standard Google Ads structure
├── Campaign types table ............... [GEN] Platform knowledge
├── Match types ........................ [GEN] Platform knowledge
├── Quality Score components ........... [VAL] Weights confirmed by research
├── Bidding strategy framework ......... [VAL] Maturity progression from research
├── Naming convention .................. [GEN] Industry patterns
├── RSA Testing Methodology ............ [RES] Optmyzr 93K RSA study — pinning data, 8-10 headlines
│ [RES] GrowthSpree — conv requirements: 50%→100, 30%→250
│ [RES] PPC.land — asset label thresholds 500/2K/5K impressions
├── Keyword research (5-step) .......... [RES] PureSEM — universe building methodology
│ [RES] seoClarity — Plant/Prune/Prioritize framework
├── KOS scoring formula ................ [RES] HM Digital Solution — weighted opportunity score
├── Search term Four-Bucket Model ...... [RES] KeywordMe — Converters/Prospects/Junk/Brand
├── Negative keyword 0-100 scoring ..... [RES] Negator.io — 5-dimension scoring system
├── Decision thresholds ................ [RES] Negator.io, AdLabs — 2x CPA, 60 clicks rules
├── Tips ............................... [GEN] Domain expertise
└── Gotchas ............................ [GEN] Domain expertise
paid-search/ad_copy_generator.yaml ....... [GEN] Original workflow
paid-search/keyword_research_engine.yaml . [GEN] Synthesized from research frameworks
paid-search/references/campaign-structures.md
├── STAG/SIAG/Hagakure ................ [RES] MeasureU, PPC Hero
├── 70-15-15 budget rule ............... [RES] HopSkipMedia
├── Campaign templates ................. [GEN] Domain expertise
├── RSA headline templates ............. [VAL] Validated by ad-copy-skill-pack
└── Negative keyword lists ............. [GEN] Domain expertise
paid-search/references/keyword-research.md
├── 5-step universe building ........... [RES] PureSEM, seoClarity, KlientBoost
├── Intent classification .............. [RES] Broder taxonomy, Content Harmony 9-type
├── KOS formula ........................ [RES] HM Digital Solution
├── Negative keyword scoring ........... [RES] Negator.io — 5 dimensions, action thresholds
├── N-gram analysis .................... [RES] Brainlabs script, WordStream
└── Quick decision rules ............... [RES] Negator.io, AdLabs — threshold compilation
Module: paid-social
paid-social/_skill.md
├── Funnel architecture ................ [GEN] Standard funnels
├── Lookalike optimization ............. [RES] AdEspresso, Meta — % performance table, seed hierarchy
│ [VAL] Seed size 1K-50K, refresh 30-90 days
├── Copy length vs performance ......... [RES] Smartly.io, AdEspresso — CTR/CVR by length
├── Audience layering rules ............ [GEN] Domain expertise
├── ABO vs CBO decision ............... [RES] Anchour Meta 2026 Playbook
├── 3-3-3 Creative Testing ............. [RES] Pilothouse
├── 3-Phase testing .................... [RES] Motion — Pre-Flight/New vs BAU/Scaling
├── CPMr fatigue ....................... [RES] Anchour — threshold, refresh cadence
├── Ad copy PASO formula ............... [GEN] Standard copywriting
├── Hook formulas ...................... [GEN] Domain expertise
├── Bidding & budget ................... [GEN] Platform knowledge
└── Tips / Gotchas ..................... [GEN] Domain expertise
paid-social/creative_test_designer.yaml .. [GEN] Synthesized from 3-3-3 framework
paid-social/references/platform-specs.md . [VAL] Generated, enriched by SearchEngineLand/Sovran
paid-social/references/video-creative-specs.md
├── Platform format comparison ......... [RES] Strike Social, Store Growers — YouTube specs
├── View definitions by platform ....... [RES] DashThis, Google/Meta/TikTok docs
├── Completion rate benchmarks ......... [RES] Amra & Elma — by device/duration/format
├── UGC vs polished data ............... [RES] Emplifi 10.38x CVR, Influee, Revel Interactive
├── Modular hook-body-CTA testing ...... [RES] Sovran — 50 variations framework
├── Optimal length by platform ......... [RES] Shortimize, TikTok — bimodal Shorts pattern
├── Video KPI definitions .............. [RES] RTB House, Simulmedia — VTR/VCR/CPV/CPCV
└── Attribution windows ................ [RES] Google/Meta/TikTok docs — click/view/engaged
Module: display-programmatic
display-programmatic/_skill.md
├── Campaign hierarchy ................. [GEN] Standard programmatic structure
├── Targeting taxonomy ................. [GEN] Domain expertise
├── Retargeting strategies ............. [GEN] Domain expertise
├── Frequency (objective-based caps) ... [RES] Trade Desk, Improvado — per-objective thresholds
├── Diminishing returns curve .......... [RES] Brand Metrics — 80/15/2 rule, 3-7 sweet spot
│ [RES] Brixon/Forrester — B2B industry-specific caps
├── Cross-device frequency ............. [RES] IAS — 32% effectiveness lift with people-based caps
├── Viewability benchmarks ............. [RES] IAS 20th MQR — rates by device/region (2024)
│ [RES] Pixalate Q4 2024 — open programmatic by environment
│ [RES] GroupM — 100% pixel premium standard
├── vCPM formula + worked example ...... [RES] MonetizeMore, Publift
├── Viewability by ad size ............. [RES] Publift — skyscraper 67%, ATF 68%, BTF 40%
├── MFA Detection (5 signals) .......... [RES] ANA — 21% of impressions, 5 consensus signals
│ [RES] DeepSee.io — ACT classification, density thresholds
├── MFA impact data .................... [RES] IAS — non-MFA +278% CVR, 63% more cost-efficient
├── Brand Safety (GARM) ................ [RES] ANA GARM framework — 12 categories + suitability
├── Pre-bid vs post-bid ................ [RES] ClearTrust, Peer39 — trade-off comparison
├── Verification vendors ............... [RES] Oden, Digiday — IAS/DV duopoly post-MOAT sunset
├── Cookieless strategy ................ [RES] Privacy Sandbox retired Oct 2025
│ [RES] GumGum — contextual 41% lower vCPM vs behavioral
├── Universal ID comparison ............ [RES] Avenga, QuickCreator — UID2/ID5/RampID table
├── Data clean rooms ................... [RES] IAB Tech Lab — PAIR v1.1, ADMaP v1.0
└── Tips / Gotchas ..................... [GEN] Domain expertise
display-programmatic/placement_audit_bot.yaml [GEN] Synthesized from MFA framework
display-programmatic/references/placement-quality.md
├── MFA 5-signal checklist ............. [RES] ANA consensus + DeepSee ACT + Coalition for Better Ads
├── ACT classification ................. [RES] DeepSee.io — Arbitrage/Clutter/Template
├── Placement quality scorecard ........ [GEN] Original 4-dimension scoring (viewability/MFA/IVT/perf)
├── Supply chain validation ............ [RES] IAB Tech Lab — ads.txt/sellers.json/SupplyChain
├── IVT classification ................. [RES] MRC — GIVT/SIVT two-tier standard
│ [RES] TAG CAF — 1.41% IVT rate on certified channels
└── Verification vendor guide .......... [RES] Oden, Digiday — IAS/DV/Pixalate comparison
display-programmatic/references/targeting-taxonomy.md
├── Targeting methods by platform ...... [GEN] Platform documentation
├── Audience segment categories ........ [GEN] Google/IAB taxonomy
├── Retargeting segment definitions .... [GEN] Domain expertise
├── Brand safety categories ............ [GEN] IAB standard
└── Creative size requirements ......... [GEN] IAB standard
Module: analytics
analytics/_skill.md
├── Conversion tracking setup .......... [GEN] Platform knowledge
├── Attribution models table ........... [VAL] Confirmed by Improvado, Deducive
├── Attribution truth hierarchy ........ [GEN] Domain expertise
├── Incrementality (geo-lift) .......... [RES] GeoLift — 10+ geos, simulation-based power
├── Platform lift requirements ......... [RES] Meta CLS $5K+/100 conv/wk, Google CLS $5K+
│ [RES] TikTok Brand Lift $30K, Google Brand Lift $10K
├── Power calculation formula .......... [RES] Two-proportion z-test, 10K practical minimum
├── Campaign audit (6-dimension) ....... [GEN] Original framework
├── Health Score (0-100) ............... [RES] claude-ads — scoring + quality gate auto-fail
├── Budget allocation methods .......... [GEN] Standard optimization
├── Measurement maturity (5-level) ..... [RES] SiteTracking — levels + advancement criteria
├── Budget pacing formulas ............. [RES] Improvado — target daily, pacing %, projected
├── Waterlevel pacing .................. [RES] Guhl et al. (ScienceDirect) — outperforms even
│ [RES] Zalando — A/B tested: +10% views, -10% CPV
├── Day-of-week CPM patterns ........... [RES] Gupta Media — Meta/TikTok/Snapchat/YouTube data
├── Alert thresholds (3-tier) .......... [RES] Improvado — z-score baseline, 30% false positive cap
├── Dashboard design ................... [VAL] Confirmed by Dataslayer
└── Tips / Gotchas ..................... [GEN] Domain expertise
analytics/campaign_audit.yaml ............ [RES] Health Score from claude-ads + quality gates
analytics/budget_allocator.yaml .......... [GEN] Original workflow
analytics/incrementality_test_planner.yaml [GEN] Synthesized from GeoLift + platform requirements
analytics/media_context_builder.yaml ..... [GEN] Original workflow
analytics/references/attribution-models.md
├── Model comparison matrix ............ [VAL] Confirmed by Deducive guide
├── Platform defaults .................. [GEN] Platform documentation
├── MMM open-source tools .............. [RES] PyMC-Marketing, Robyn, Meridian, Funnel.io
├── Incrementality testing guide ....... [VAL] Enriched with GeoLift + Google geo suite
├── Measurement maturity model ......... [RES] SiteTracking 5-level framework
└── UTM taxonomy ....................... [GEN] Industry standard
Module: landing-pages
landing-pages/_skill.md
├── Page anatomy ....................... [GEN] Standard structure
├── Message match (quantified) ......... [RES] ConversionLab — DTR +31.4% (77-day test)
│ [RES] NextAfter — visual congruence +47.6% (239K sample)
│ [RES] MarketingExperiments — continuity +63.3%
├── Continuity + Congruence ............ [RES] MarketingExperiments — two-principle framework
├── Friction audit ..................... [GEN] Domain expertise
├── ICE framework ...................... [GEN] Well-known
├── PXL framework ...................... [RES] CXL/Speero — binary scoring, 8 criteria
├── Statistical significance ........... [GEN] Standard testing knowledge
├── Form optimization (data-backed) .... [RES] Crazy Egg — 4→3 fields +50% CVR
│ [RES] Imagescape — 11→4 fields +120% CVR
│ [RES] Formstack — multi-step +86% vs single
│ [RES] Marketo — progressive profiling +42% submissions
├── Page speed (per-second CVR) ........ [RES] Portent 100M+ views — 4.42%/second drop
│ [RES] Google/Deloitte — 0.1s = +8.4% conversion
├── Tips / Gotchas ..................... [GEN] Domain expertise
landing-pages/friction_audit_engine.yaml . [GEN] Synthesized from friction framework
landing-pages/references/cro-checklist.md [VAL] Generated, confirmed by Fibr.ai methodology
Module: email-sequences (NEW)
email-sequences/_skill.md
├── Sequence types table ............... [RES] ActiveCampaign, Omnisend — timing blueprints
├── Subject line formulas .............. [RES] Return Path 9B emails — 28-39 chars optimal
│ [RES] Experian — personalization +26% open rate
├── Preheader impact ................... [RES] Litmus — +7% open rate, 40-100 chars
├── Body length by type ................ [RES] Boomerang 40M emails — 50-125 words optimal
├── Single CTA rule .................... [RES] WordStream — +371% clicks with single CTA
├── Segmentation strategy .............. [RES] Omnisend — segmented +101% CTR, +760% revenue
├── Deliverability checklist ........... [RES] Growleads, SalesHive — SPF/DKIM/DMARC requirements
├── Warm-up schedule ................... [RES] Infobip — 14-day schedule with daily volumes
├── A/B testing priority ............... [RES] Mailchimp, HubSpot, Litmus — ranked by impact
├── Sample size minimums ............... [VAL] Standard z-test with marketing-specific thresholds
├── Benchmarks by type ................. [RES] Klaviyo, Omnisend, ActiveCampaign — OR/CTR/CVR
├── Tips / Gotchas ..................... [GEN] Domain expertise
email-sequences/references/sequence-templates.md
├── Welcome sequence blueprint ......... [RES] ActiveCampaign — 89% revenue increase moving pitch
├── Abandoned cart (timing by AOV) ..... [RES] Attribuly — low/mid/high AOV intervals
│ [RES] Closer Apps/Shopify — recovery rates 7-35% by AOV
├── B2B lead nurture ................... [RES] Prospeo — 8-email/30-day with OR/CTR per email
├── Post-purchase ...................... [RES] Omnisend — 22.64% click-to-conversion
├── Re-engagement ...................... [GEN] Standard win-back pattern
├── Branching logic patterns ........... [RES] Klaviyo — flow branching conditions
└── Email-to-LP match rules ........... [RES] MarketingExperiments — mismatch = 50-80% CVR loss
Aggregate Breakdown (conservative — research claims halved)
┌─────────────────────────────────────────────────┐
│ CONTENT PROVENANCE BREAKDOWN │
├─────────────────────────────────────────────────┤
│ │
│ ██████████████████████████████░░ ~70% GEN │
│ ░░░░░░░░░░░░░░████████████░░░░░ ~20% RES │
│ ░░░░░░░░░░░░░░░░░░░░░░░░░░████░ ~10% VAL │
│ │
└─────────────────────────────────────────────────┘
GEN = Generated from domain expertise (no specific source)
RES = Adapted from specific public research finding
VAL = Generated first, then validated/enriched by research
Note: Research claims halved from raw agent output to reflect
actual verification confidence. Many "RES" sections use research
to enrich generated structure — the structure itself is still GEN.
Some benchmarks come from training data (not web-verified URLs).
By Module
| Module | GEN | RES | VAL | Confidence | Key Research |
|---|---|---|---|---|---|
| display-programmatic | 58% | 40% | 2% | Highest | MFA ACT, IAS/Pixalate viewability, Privacy Sandbox (all web-verified) |
| email-sequences | 58% | 40% | 2% | Highest | Cart timing by AOV, warm-up schedule, cadence (2/3 web-verified) |
| analytics | 62% | 30% | 8% | High | GeoLift, platform lift reqs, Gupta Media CPMs (all web-verified) |
| analyst role | 65% | 35% | — | High | Four Levers, 3-tier alerting, z-score (all web-verified) |
| paid-search | 65% | 28% | 7% | High | Negator.io scoring, Optmyzr 93K RSAs (all web-verified) |
| landing-pages | 68% | 27% | 5% | High | DTR +31%, page speed CVR, form field data (2/3 web-verified) |
| paid-social | 67% | 25% | 8% | Medium | 3-3-3 web-verified; LAL %/copy length from training data only |
| media-buyer role | 75% | 25% | — | Medium | Quality Gates web-verified; lifecycle/principles are GEN |
| Root + manifest | 95% | 5% | — | N/A | Structural — doesn't need research |
Research Agent Summary
| Phase | Agents | Focus | Key Sources Found |
|---|---|---|---|
| Initial | 3 | PPC, Social, Analytics/CRO | claude-ads, Pilothouse 3-3-3, CXL PXL |
| Phase 1 | 3 | Display: MFA, viewability, cookieless | ANA/DeepSee ACT, IAS/Pixalate benchmarks, Privacy Sandbox dead |
| Phase 2 | 3 | Analyst, incrementality, pacing | Pixis Four Levers, GeoLift power calcs, Gupta Media CPMs |
| Phase 3 | 3 | Keywords, search terms, RSA testing | PureSEM universe, Negator.io scoring, Optmyzr 93K study |
| Phase 4 | 3 | Lookalike, message match, friction | ConversionLab DTR +31%, Portent page speed, Marketo profiling |
| Phase 5 | 3 | Email sequences, deliverability, copy | Attribuly cart timing, Infobip warm-up, Boomerang 40M emails |
| Phase 6 | 3 | YouTube, short-form video, measurement | Strike Social formats, Amra & Elma completion rates, MRC video |
| Total | 21 | ~50 unique sources in ATTRIBUTION.md |