project-aeo-monitoring-tools（项目 AEO 监控工具）

AEO Monitoring Tools

Build custom infrastructure for monitoring AI search engine visibility and competitive citation analysis.

Audience: Engineers building custom AEO monitoring systems

For AEO strategy and content optimization: Use marketing-ai-search-optimization instead. For traditional SEO: Use marketing-seo-complete instead.

When to Use This Skill

• Building custom AEO monitoring infrastructure (API pipelines, citation databases, dashboards)
• Evaluating build vs. buy decisions for AI search tracking
• Understanding API vs. scraping trade-offs per platform
• Designing data pipelines for citation analysis
• Estimating costs for multi-platform monitoring

When NOT to Use

• AEO content optimization (improving pages for citation) -> Use marketing-ai-search-optimization
• Traditional SEO (crawlability, indexation, Core Web Vitals) -> Use marketing-seo-complete
• Content planning and editorial strategy -> Use marketing-content-strategy
• Product analytics instrumentation -> Use marketing-product-analytics

Verify Before Committing

AEO tools evolve rapidly (acquisitions, pricing changes, new entrants). Before committing to any tool or API, verify current status via web search:

• "[tool name] pricing [current year]"
• "[platform] API rate limits [current year]"
• "AEO monitoring tools comparison [current year]"

Decision Framework: Build vs. Buy

Before building custom tools, evaluate whether commercial solutions fit your needs.

Commercial tools to evaluate first:

AEO-Native Tools:

Incumbents Adding AEO Features:

See docs/context.md in the reference implementation for 24+ competitors with funding data.

Platform Access Overview

Each AI platform requires different access approaches.

Note: DeepSeek and Grok are listed for completeness. The reference implementation currently supports Perplexity, Gemini, Claude, and OpenAI. DeepSeek and Grok collectors are not yet implemented.

Key insight: Perplexity Sonar API is the most AEO-friendly - it returns citations natively in the response.

See: references/platform-access-methods.md

Architecture Tiers

Tier 1: API-First (Recommended)

Use official APIs where available. Lowest risk, most maintainable.

Query Bank -> API Orchestrator -> Response Store -> Analysis Layer
   (30-100 quick start;     (rate limiting,    (PostgreSQL/    (citation extraction,
    queries)     retry logic)       BigQuery)       brand detection)

Platforms covered: Perplexity, Gemini, Claude, OpenAI (baseline; use official web-search tooling if available) Cost: $15-300/mo depending on volume Risk: Low

Tier 2: Hybrid (API + Commercial Scraping)

Add commercial scraping services for platforms without good APIs.

Additional coverage: ChatGPT web interface, Google AI Overviews Cost: $500-1,500/mo (adds commercial scraper fees) Risk: Medium (dependent on scraper provider)

Tier 3: Full Custom Scraping (Not Recommended)

DIY web scraping of AI platforms.

Why to avoid:

• High ToS violation risk
• Aggressive bot detection (especially Google, ChatGPT)
• Maintenance burden (UI changes break scrapers)
• Potential legal liability

See: assets/technical/architecture-diagrams.md

Risk Assessment Matrix

Legal developments to monitor:

• Publisher lawsuits and data sourcing disputes (example: Reddit v. Perplexity AI (2024))
• Platform ToS enforcement and liquidated damages policies (example: X ToS changes)
• Rising use of crawler blocks and WAF rules (GPTBot, ClaudeBot, etc.)

See: references/legal-compliance.md

Cost Estimation

See: references/cost-estimation.md

Implementation Timeline

See: assets/setup/minimal-setup-guide.md

What to Load (Progressive Disclosure)

Load additional references based on your needs:

Quick Validation (First API Call)

Test Perplexity Sonar to confirm citations work:

curl -X POST "https://api.perplexity.ai/chat/completions" \
  -H "Authorization: Bearer $PERPLEXITY_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "sonar",
    "messages": [{"role": "user", "content": "What is [YOUR_BRAND]?"}]
  }'

Expected: JSON response with citations array containing source URLs. If your brand appears with citations, monitoring is viable for that platform.

Quick Start Checklist

[ ] Define query bank (30-100 for quick start; 250-500 for advanced)
[ ] Choose platforms to monitor (prioritize by ICP usage)
[ ] Evaluate build vs. buy decision
[ ] If building: Set up API accounts (Perplexity, Gemini, Claude/OpenAI)
[ ] Run quick validation call above to confirm API access
[ ] Create database schema (PostgreSQL recommended)
[ ] Build API orchestrator with rate limiting
[ ] Implement citation extraction
[ ] Set up scheduled runs (daily/weekly)
[ ] Create Share of Model dashboard
[ ] Document maintenance procedures
[ ] (Optional) Monitor access to discovery files (/llms.txt, /.well-known/*.json)

Key Metrics

Primary metric: Share of Model (SoM)

SoM = (Your brand mentions / Total responses) * 100

Track SoM:

• Per platform (ChatGPT, Perplexity, Gemini, Claude)
• Per query intent (informational, commercial, transactional)
• Over time (weekly/monthly trends)
• vs. competitors

Secondary metrics:

• Brand mention rate (% of responses where your brand is named in answer text — 3.2x more frequent than citations per BrightEdge)
• Citation rate (% of responses with your URL)
• Position in citations (1st, 2nd, 3rd mention)
• Third-party vs owned citation ratio (what % of citations come from G2, Reddit, YouTube vs your site)
• Sentiment of brand mentions
• Query coverage (% of target queries where you appear)

Advanced metrics (reference implementation features):

• Bot ingestion rate (% of pages crawled by AI bots from server logs)
• Page health score (composite: freshness + structure + citation-readiness)
• Citation network depth (how many hops from your cited page to the AI response)
• AI referral tracking (traffic from known AI assistant domains)
• Persona visibility (brand appearance segmented by user demographic/persona)
• Content optimization score (gap between current content and ideal citation-ready structure)

Advanced Features (Beyond Basic Monitoring)

The reference implementation extends basic monitoring with these advanced capabilities:

Bot Analytics and Crawler Intelligence

Track which AI crawlers access your content and how they process it.

• Server log analysis for GPTBot, ClaudeBot, PerplexityBot, GoogleOther
• Crawl frequency and depth patterns per bot
• Content type preferences (which pages bots visit most)
• Ingestion-to-citation correlation (does being crawled lead to being cited?)
• Discovery file access tracking: monitor requests to /llms.txt, /.well-known/llmprofiles.json, /.well-known/mcp.json, /.well-known/agents.json to measure AI agent adoption of emerging standards

Citation Network Analysis

Map how citations flow between your content and AI responses.

• Citation graph: track which of your pages are cited, by which platforms, for which queries
• Citation co-occurrence: which competitor pages appear alongside yours
• Citation depth: direct citation vs. derived/summarized mentions
• Temporal patterns: how citation freshness decays over time
• Third-party citation tracking: monitor when third-party sources (G2, Reddit, YouTube, listicles) cite your brand in AI responses vs. your owned pages. See marketing-ai-search-optimization for the earned AEO strategy that feeds this data

Content Optimization Engine

Automated recommendations for improving citation probability.

• Gap analysis: compare your content structure against top-cited pages
• Recommendation engine: specific suggestions (add TL;DR, add comparison table, cite primary sources)
• A/B tracking: measure citation rate changes after content updates
• Priority scoring: which pages have highest citation improvement potential

Personas and Demographics

Understand how different user segments discover your brand through AI.

• Persona-based query segmentation (technical buyer, executive, end user)
• Platform preference by persona (developers prefer Perplexity, executives prefer ChatGPT)
• Visibility gaps by segment: where you're strong vs. weak per persona
• Brand hub: centralized brand identity data for consistent AI representation

Related Skills

Disclaimer

This guidance is for educational purposes. Users must:

• Conduct their own legal review
• Ensure compliance with applicable terms of service
• Respect robots.txt directives
• Follow laws and regulations in their jurisdiction

Building monitoring tools that violate platform ToS may result in account termination, legal action, or both.