marketing-ai-search-optimization（营销 AI 搜索优化）

AI Search & Answer Engine Optimization (GEO)

Improve how assistants retrieve, summarize, and cite your pages.

For traditional SEO: Use marketing-seo-complete instead.

GEO vs SEO (Overlap Map)

Use this to prevent “GEO-only” work that ignores discoverability and conversion.

GEO is best at

• Making pages easier for assistants to extract, summarize, and cite
• Building entity/proof structures that improve citation probability
• Measuring assistant visibility via query banks and citation share

SEO is still required for

• Getting pages discovered and indexed reliably (crawlability, internal linking, canonicalization)
• Capturing demand in classic search surfaces (SERPs, video, local, forums)
• Avoiding regressions from technical changes (rendering, performance, duplication)

Default operating rule

• Keep classic SEO and conversion work running; treat GEO as a structured overlay on top of high-intent pages.

GEO Monitoring vs GEO Optimization

This skill covers optimization — improving your content so AI platforms cite you more often.

For monitoring infrastructure — building the systems that track whether AI platforms cite you — see project-aeo-monitoring-tools.

Typical workflow: Monitor (track current visibility) -> Optimize (improve content) -> Measure (verify improvement)

Quick start (30–60 min)

1. Build a query bank (30–100 queries for quick start; scale to 250–500 for advanced monitoring): problems, comparisons, "best", "vs", integrations, and pricing questions.
2. Confirm assistants can fetch content (robots/WAF/SSR): use assets/audits/crawler-access-audit.md.
3. Run a baseline visibility audit: use assets/audits/search-visibility-audit.md and assets/audits/ai-search-content-audit.md.
4. Ship one high-leverage page update: use assets/content/ai-search-content-brief.md + assets/content/answer-focused-article-template.md.
5. Set up measurement + retest cadence: use references/measurement-analytics.md and assets/testing/ai-search-testing-protocol.md.

Core workflow

1) Decide scope (avoid wasted work)

• Confirm discovery channel: check whether your ICP uses assistants for research and comparisons.
• Pick one primary platform first (Google AI Overviews vs ChatGPT vs Perplexity) based on your audience.
• Treat GEO as additive: keep classic SEO and conversion work running.

2) Ensure assistants can access your content

• Allow/deny crawlers explicitly: use references/ai-crawler-technical-setup.md and assets/technical/robots-txt-ai-crawlers.md.
• Reduce JS dependency for critical copy (SSR/SSG): use assets/technical/server-side-rendering-guide.md.
• Add llms.txt when useful as a navigation map (not a guarantee): use assets/technical/llms-txt-template.md.
• Review emerging .well-known/ AI discovery standards (llmprofiles.json, mcp.json, agents.json): use assets/technical/well-known-ai-discovery.md.

3) Make pages easy to extract and cite

• Put a direct, quotable answer block in the first screenful (then expand with proof).
• Use stable entities (product, category, competitors, integrations): use references/entity-semantic-optimization.md.
• Use repeatable content structures for questions, comparisons, and "best for": use references/content-structure-patterns.md.

Implementation reference: The AEO monitoring platform's recommendation engine (src/lib/recommendations/engine.ts) automates gap analysis against these patterns. The optimization dashboard (src/app/optimize/page.tsx) surfaces actionable recommendations. See project-aeo-monitoring-tools for the full implementation.

• Create/refresh high-intent pages first (alternatives, integrations, pricing, security, implementation): use assets/strategy/ai-search-growth-plan.md.

4) Build off-site entity presence and earned citations

• Get your brand into third-party sources AI trusts (G2, Reddit, Wikipedia, YouTube, industry listicles): use references/earned-aeo-third-party-citations.md.
• Strengthen Knowledge Graph presence (Wikidata, Google Business Profile, sameAs linking): see Knowledge Graph section in references/entity-semantic-optimization.md.
• Create multimodal content (video, transcripts, audio) for AI platforms that cite non-text sources: use references/multimodal-content-optimization.md.
• For e-commerce: implement Google UCP for agentic shopping visibility: use references/commerce-protocol-ucp.md.

5) Add proof and trust hooks (citation fuel)

• Prefer primary sources and verifiable numbers; attribute claims clearly.
• Show authorship, review, and freshness (dateModified / "Last updated") where appropriate.
• Avoid "LLM bait": prioritize user value and factual accuracy.

6) Measure, iterate, and defend against regressions

• Track "share of model" / citation share using your query bank, not vanity rankings. For automated tracking, see project-aeo-monitoring-tools (custom infrastructure) or commercial alternatives in references/llm-tracking-tools.md.
• Re-test after shipping changes; keep snapshots of answers and citations.
• Separate SEO wins vs assistant visibility wins; avoid false attribution.

Implementation Examples

Query Bank Construction

Quick start (30-100 queries):

Problems:     "how to [solve X]", "why does [Y happen]"
Comparisons:  "[product] vs [competitor]", "best [category] for [use case]"
Integrations: "[product] [integration] setup", "does [product] work with [tool]"
Pricing:      "[product] pricing", "[product] free plan"

Advanced (250-500 queries): Expand with persona variants, regional variations, long-tail variations, and seasonal queries. See project-aeo-monitoring-tools for full query bank methodology.

Content Structure Patterns

Apply these patterns to high-intent pages:

Comparison page:
  H1: [Product A] vs [Product B]: [Year] Guide
  TL;DR: 2-3 sentence verdict
  Table: Feature comparison
  Sections: Use cases, pricing, verdict

Alternatives page:
  H1: Best [Product] Alternatives in [Year]
  TL;DR: Top 3 picks with one-line reasons
  Table: Feature + pricing matrix
  Sections: Detailed review per alternative

Integration page:
  H1: How to Connect [Product] with [Tool]
  Steps: Numbered setup guide
  Code: Configuration examples
  FAQ: Common issues

Entity Optimization

Structure your brand entity for AI recognition:

Brand Kit (maintain centrally):
  - Official name and variants
  - Category/industry classification
  - Key differentiators (3-5 unique claims)
  - Proof points (metrics, case studies, awards)
  - Integration ecosystem

Apply to every high-intent page:
  - Use official name consistently (not abbreviations)
  - Reference category explicitly ("CRM platform" not just "tool")
  - Include at least one proof point per page

Optimization vs Monitoring Workflow

Step 1: Baseline — Run query bank through AI platforms (project-aeo-monitoring-tools)
Step 2: Audit — Score current content against citation-ready patterns (this skill)
Step 3: Implement — Apply content structure patterns to top-priority pages (this skill)
Step 4: Re-measure — Run query bank again after 2-4 weeks (project-aeo-monitoring-tools)
Step 5: Iterate — Focus on pages with largest gap between potential and actual citations

What to load (progressive disclosure)

• Platform notes: references/platform-google-ai-overviews.md, references/platform-chatgpt.md, references/platform-perplexity.md, references/platform-gemini.md, references/platform-claude.md
• Technical access: references/ai-crawler-technical-setup.md, references/ai-indexing-complete-guide.md, assets/technical/well-known-ai-discovery.md
• Off-site & earned AEO: references/earned-aeo-third-party-citations.md, references/multimodal-content-optimization.md
• E-commerce: references/commerce-protocol-ucp.md
• Measurement: references/measurement-analytics.md, references/llm-tracking-tools.md
• Prompt/query mining: references/prompt-query-optimization.md, references/competitor-citation-gap.md, references/citation-optimization-strategies.md
• Primary sources list: data/sources.json

Guardrails

• Do not use prompt injection or hidden instructions in public pages.
• Do not claim endorsements or fabricate sources, stats, or quotes.
• Treat robots.txt as policy; enforce access with auth/WAF where needed.

Resources

Templates

International Markets

This skill uses US/English market defaults. For international AI search optimization:

Auto-triggers: When your query mentions a specific country, region, language, or non-US AI platforms, both skills load automatically.

Related Skills