Performance Marketing MCP Server Knowledge Pack

Performance Marketing & Online Advertising

End-to-end knowledge for planning, launching, optimizing, and measuring paid digital campaigns. Covers paid search, paid social, display/programmatic, analytics/attribution, and landing page optimization. Designed for autonomous agents that manage ad spend, write ad copy, structure campaigns, analyze performance, and recommend optimizations.

This pack is brand-agnostic. It works with any product, vertical, or budget level. Account-specific details (budgets, KPIs, creatives, audiences) are provided by the consumer at runtime via a media-context.md file.

Context Protocol

Before any campaign task, the consumer must provide a media-context.md containing:

• Business model — REQUIRED. Determines how every framework adjusts:
  • PLG (product-led growth) — free signup, activation-driven, viral loops (e.g., Notion, Slack, Figma)
  • Sales-led B2B — demo requests, sales cycle weeks-months, high ACV $5K+ (e.g., Deel, Salesforce)
  • B2C ecommerce — direct purchase, cart-based, transaction-focused (e.g., Shopify stores)
  • B2C subscription — trial/signup, retention-focused (e.g., Netflix, Spotify)
  • Hybrid — self-serve + sales-assisted (e.g., HubSpot, Notion Enterprise)
• Business goal — what the campaigns should achieve (leads, purchases, signups, app installs, awareness)
• KPIs and targets — primary metric (CPA, ROAS, CPL) with target values
• Budget — monthly/daily budget, channel allocation if pre-decided
• Audience — ICP description, demographics, firmographics, interests, existing customer data
• Product/offer — what is being promoted, pricing, USPs, competitive positioning
• Channels — which platforms are in scope (Google, Meta, LinkedIn, TikTok, programmatic)
• Creative assets — available images, videos, logos, brand guidelines
• Tracking setup — pixel/tag status, conversion events defined, analytics platform
• Historical data — past campaign performance if available (benchmarks, winners/losers)

If no media context is provided, ask for one before executing. Without it, recommendations will be generic.

Campaign Taxonomy

Type	Objective	Key Metric	Typical Channels
Search — Brand	Protect brand terms, capture high-intent	CPC, impression share	Google, Bing
Search — Non-brand	Capture demand, drive conversions	CPA, ROAS, conv. rate	Google, Bing
Shopping / PMax	Product-feed campaigns, ecommerce	ROAS, COS	Google, Bing
Social — Prospecting	Reach new audiences, generate demand	CPM, CPA, CTR	Meta, TikTok, LinkedIn
Social — Retargeting	Re-engage visitors, drive conversions	CPA, ROAS, frequency	Meta, TikTok, LinkedIn
Display — Prospecting	Awareness, reach, upper funnel	CPM, viewability, reach	GDN, DV360, programmatic DSPs
Display — Retargeting	Re-engage across web, sequential messaging	CPA, view-through conv.	GDN, DV360, programmatic DSPs
Video	Awareness, consideration, storytelling	CPV, VTR, brand lift	YouTube, Meta, TikTok
Lead gen	Capture contact info directly in-platform	CPL, lead quality score	Meta Lead Ads, LinkedIn Lead Gen

Module Catalog

Module	Purpose	Key Frameworks
paid-search	Google/Bing search campaigns: structure, keywords, bidding, ad copy	SKAG vs themed groups, match type strategy, Quality Score, RSA best practices
paid-social	Meta/LinkedIn/TikTok campaigns: audiences, creatives, bidding	Funnel architecture, audience layering, creative testing frameworks
display-programmatic	Display ads, programmatic buying, retargeting	DSP setup, targeting taxonomy, frequency management, viewability
analytics	Measurement, attribution, reporting, optimization	Attribution models, conversion tracking, incrementality, dashboarding
landing-pages	CRO, A/B testing, message-match, page performance	Message match, friction audit, test prioritization, statistical significance
email-sequences	Post-conversion email automation: welcome, cart, nurture, post-purchase	Sequence architecture, copy frameworks, deliverability, A/B testing

Loading Order

1. This file (_skill.md) — pack overview, context protocol, taxonomy
2. Consumer's media-context.md — account-specific goals, budgets, audiences
3. Role file (_roles/media-buyer.md or _roles/analyst.md) — task lifecycle
4. Module _skill.md — channel-specific methodology
5. Module references/ — detailed frameworks, specs, checklists (loaded on demand)

Cross-Module Dependencies

• analytics is referenced by all other modules — every campaign needs measurement setup before launch
• landing-pages is referenced by paid-search and paid-social — ad-to-page message match is critical for Quality Score and conversion rate
• paid-search and paid-social share audience insights — search query data informs social targeting and vice versa
• display-programmatic retargeting depends on pixel/audience data from paid-search and paid-social campaigns
• email-sequences depends on landing-pages (form capture) and paid-search/paid-social (acquisition source segmentation)

Workflow Catalog

Workflow	Module	Purpose
campaign_audit	analytics	Audit campaign structure, spend efficiency, and identify optimization opportunities
ad_copy_generator	paid-search	Generate RSA headlines and descriptions from product/offer brief
budget_allocator	analytics	Recommend budget allocation across channels based on historical performance
creative_test_designer	paid-social	Design 3-3-3 creative test matrix with hypothesis, kill rules, and measurement plan
friction_audit_engine	landing-pages	Score page across 7 friction types with severity-weighted rubric and prioritized fixes
incrementality_test_planner	analytics	Design incrementality test with power calculation, market selection, and analysis plan
keyword_research_engine	paid-search	Build keyword universe with intent classification, KOS scoring, and ad group clustering
placement_audit_bot	display-programmatic	Analyze placements for MFA risk, viewability, and quality scoring with exclusion list
media_context_builder	analytics	Build complete media-context.md from structured business inputs