AI Citation Monitor

"Did Google AI Overview cite your brand? Know in minutes, not months."

---

🎯 What This Solves

The Problem:
GEO (Generative Engine Optimization) and AEO (Answer Engine Optimization) are the new SEO. But you have no idea if Google AI Overview mentions your company — until a customer tells you they never found you.

The Solution:
Run a batch of natural-language prompts. Get back structured JSON showing:

• Which prompts triggered AI Overviews
• Whether those Overviews cited your brand/domain
• Exact snippets and citation URLs

Pay-per-use. No subscription. Self-hostable.

---

✅ Why Buyers Pay For This

Feature	$99/mo SaaS Competitors	AI Citation Monitor
Pricing	$99–$299/month subscription	$2.50–$4.00 per 1,000 prompts
Commitment	Monthly recurring	No subscription, no lock-in
Interface	Fixed dashboard taxonomy	Any natural-language prompt you write
Self-Hostable	No (SaaS only)	Yes (Docker/Python)
Data Freshness	Cached/historical	Fresh search each invocation
AI Engine Coverage	Fixed set	Google AI Overview, AI Mode, Perplexity, ChatGPT, Copilot, Gemini

Target Customer:

• AI/tech startups tracking AI citation without monthly commitment
• Marketing teams running one-off brand audits
• SEO consultants doing competitor GEO analysis
• Indie hackers building AI-native products

---

📊 How It Works

Input Example

{
  "brand": "NVIDIA",
  "domain": "nvidia.com",
  "prompts": [
    "What are the best AI chips for training large language models?",
    "Which companies are leading in generative AI infrastructure?",
    "Compare NVIDIA H100 vs AMD MI300 for AI workloads"
  ],
  "ai_engines": ["google-overview"],
  "exclude_own_domain": true,
  "country": "US",
  "language": "en"
}

Output Example

{
  "metadata": {
    "brand": "NVIDIA",
    "domain": "nvidia.com",
    "generated_at": "2026-08-24T12:34:56Z",
    "total_prompts": 3,
    "total_citations": 2,
    "engines_used": ["google-overview"]
  },
  "results": [
    {
      "prompt": "What are the best AI chips for training large language models?",
      "ai_engine": "google-overview",
      "cited": true,
      "answer_snippet": "NVIDIA's H100 GPU is widely considered the industry standard for training large language models...",
      "citation_urls": [
        {
          "url": "https://www.nvidia.com/en-us/data-center/h100/",
          "title": "NVIDIA H100 Tensor Core GPU",
          "position": 1
        },
        {
          "url": "https://blogs.nvidia.com/blog/ai-chip-guide/",
          "title": "Guide to AI Chips",
          "position": 3
        }
      ],
      "source": "Google AI Overview",
      "timestamp": "2026-08-24T12:34:56Z"
    }
  ]
}

---

🚀 Getting Started

Quick Start (Local Testing)

# Clone or navigate to the project
cd <your-project-directory>

# Activate virtual environment
source venv/bin/activate

# Install dependencies
pip install -r apify-actor/requirements.txt

# Run smoke test with real data
python smoke_test_actor_mvp.py

# Results saved to: docs/SMOKE_TEST_RESULTS.json

Try the Live Actor

Actor #1 is now LIVE on Apify Store:

• URL: https://apify.com/nextgen.api/spark-agent-commerce-ai-citation-monitor
• PPE Pricing: Active pay-per-event pricing (see pricing table below)

---

📋 Configuration

Required Parameters

Parameter	Type	Description	Example
brand	string	Brand/company name to check	"NVIDIA"
domain	string	Primary domain for URL matching	"nvidia.com"
prompts	array	Natural-language search queries	["What are..."]

Optional Parameters

Parameter	Type	Default	Description
ai_engines	array	["google-overview"]	Which AI search platforms: google-overview, google-ai-mode, perplexity, chatgpt-search, copilot, gemini
exclude_own_domain	boolean	true	Exclude results from your own domain (external citations only)
country	string	"US"	Google search country code
language	string	"en"	Google search language code

---

💰 Pricing

Apify Store PPE (Pay-Per-Event)

Live COGS: $0.0075 per prompt ($7.50 per 1,000)

• Apify google-search-scraper page: $0.0045
• AI Overview add-on: $0.003

Live PPE Pricing Table (as of 2026-09-03):

Event	FREE	BRONZE	SILVER	GOLD	PLATINUM	DIAMOND
Prompt checked	$0.022	$0.019	$0.017	$0.015	$0.015	$0.015
Actor Start	$0.00005 per event (one event per GB of memory, minimum 1)

Note: Pricing is tiered by event type. The "Prompt checked" event uses the tiered pricing above, while "Actor Start" charges $0.00005 per GB of memory used (minimum 1 event).

---

🔧 Technical Details

Architecture

┌─────────────────────────────────────────────────────┐
│              Apify Actor (main.py)                   │
│  - Validates INPUT_SCHEMA                            │
│  - Calls worker.check_ai_citations()                 │
│  - Pushes results to dataset                         │
└─────────────────────────────────────────────────────┘
                        ↓
┌─────────────────────────────────────────────────────┐
│         Worker Core (worker/core.py)                 │
│  - Calls apify/google-search-scraper                 │
│  - Parses AI Overview fields                         │
│  - Matches citations against brand/domain            │
│  - Formats output per WEDGE.md schema                │
└─────────────────────────────────────────────────────┘

Dependencies

• apify (async client)
• pydantic (input/output validation)
• ddgs (DuckDuckGo for free testing)

Supported AI Engines

Engine	Status	Notes
google-overview	⚠️ Paid	Requires Apify billing unlock
google-ai-mode	⚠️ Paid	Requires Apify billing unlock
perplexity	🚧 Planned	Phase B
chatgpt-search	🚧 Planned	Phase B
copilot	🚧 Planned	Phase B
gemini	🚧 Planned	Phase B

Free Testing: DuckDuckGo search available for local testing without Apify billing.

---

📊 Sample Output

Example 1: Brand Cited (cited: true)

{
  "prompt": "What are the best AI chips for training large language models?",
  "ai_engine": "google-overview",
  "cited": true,
  "answer_snippet": "NVIDIA's H100 GPU is widely considered the industry standard for training large language models. The H100 Tensor Core GPU delivers exceptional performance for AI workloads, with 80GB HBM3 memory and 3.35 TB/s memory bandwidth.",
  "citation_urls": [
    {
      "url": "https://www.nvidia.com/en-us/data-center/h100/",
      "title": "NVIDIA H100 Tensor Core GPU",
      "position": 1,
      "match_type": "exact_domain"
    },
    {
      "url": "https://blogs.nvidia.com/blog/ai-chip-guide-2026/",
      "title": "Guide to AI Chips in 2026",
      "position": 2,
      "match_type": "exact_domain"
    }
  ],
  "source": "Google AI Overview",
  "timestamp": "2026-08-24T14:23:45Z",
  "snippet_length": 287,
  "num_citations": 2
}

Interpretation:

• cited: true — Your brand/domain appears in the AI Overview
• citation_urls — Array of URLs cited, with positions and match types
• answer_snippet — The full AI-generated text (for context)

Example 2: Brand NOT Cited (cited: false)

{
  "prompt": "What GPU does OpenAI use for ChatGPT?",
  "ai_engine": "google-overview",
  "cited": false,
  "answer_snippet": "OpenAI primarily uses NVIDIA GPUs for training and inference of ChatGPT. The company has partnered with NVIDIA for H100 and A100 GPUs, and also develops custom infrastructure in collaboration with Microsoft Azure.",
  "citation_urls": [],
  "source": "Google AI Overview",
  "timestamp": "2026-08-24T14:24:08Z",
  "snippet_length": 267,
  "num_citations": 0,
  "note": "Brand mentioned in snippet but no direct URL citations"
}

Interpretation:

• cited: false — AI Overview exists but doesn't cite your URLs
• answer_snippet — AI still generated text (may mention your brand without linking)
• citation_urls: [] — Empty array means no direct URL citations
• Opportunity: This is a GEO optimization target — your content should be cited here

Output Metadata

{
  "metadata": {
    "brand": "NVIDIA",
    "domain": "nvidia.com",
    "generated_at": "2026-08-24T14:23:45Z",
    "total_prompts": 5,
    "total_citations": 4,
    "citation_rate": 0.8,
    "engines_used": ["google-overview"],
    "data_mode": "live"
  }
}

Metrics:

• citation_rate — Percentage of prompts where your brand was cited
• total_citations — Total number of URL citations found
• data_mode: live — Each invocation runs fresh searches (no caching)

---

📁 File Structure

apify-actor/
├── main.py                 # Actor entry point
├── INPUT_SCHEMA.json       # Apify input form schema
├── actor.json              # Apify metadata + categories
├── requirements.txt        # Python dependencies
└── README.md              # This file

worker/
├── __init__.py            # Worker exports
└── core.py                # Citation checking logic

docs/
├── WEDGE.md               # Product spec + pricing strategy
├── WORKER_SPEC.md         # I/O schema specification
├── MARKETPLACE.md         # Marketplace listing pack
└── SMOKE_TEST_RESULTS.json # Live test output

ops/
└── LOGAN_UNLOCKS.md       # Apify billing unlock steps

---

⚠️ Important Constraints

Honest Limitations

✅ Apify billing required for production — Google AI Overview scraping requires Apify paid billing. This is not a bug — it's the business model.

✅ Live PPE pricing active — See pricing table above for current rates by event type and tier.

✅ AI Overview add-on may cost extra — Capturing AI Overviews successfully may incur additional charges per query.

✅ Requires Google AI Overview capture — If Google does not render an AI Overview for a query, the cited status is unknown (not a false). This is a limitation of Google's AI Overview feature, not a bug in the actor.

What This Is NOT

❌ Not a DuckDuckGo product — DDG has no AI Overviews. This is specifically about Google AI citations.

❌ Not a full web scraper — We only need AI Overview extraction, not full SERP scraping.

❌ Not a monthly subscription — This is PPE (pay-per-event) on Apify Store, not SaaS.

❌ Not a historical database — Fresh search each invocation. No stored past results.

---

🛠️ Support & Documentation

Documentation

• docs/WEDGE.md — Product spec + market validation
• docs/WORKER_SPEC.md — I/O schema specification
• docs/MARKETPLACE.md — Marketplace listing pack
• ops/LOGAN_UNLOCKS.md — Apify billing unlock steps

Common Issues

Issue	Solution
"No input provided"	Add brand, domain, and prompts to Actor input
"Import error: apify"	Install with pip install apify
Empty results (Google)	Apify billing unlock required for Google AI Overview
Empty results (DDG)	Try different brand/prompt; DDG may not find matches
Need Google AI data	See billing unlock steps in ops/LOGAN_UNLOCKS.md

Testing

# Run smoke test with DuckDuckGo (free, no billing needed)
python smoke_test_actor_mvp.py

# Check results
cat docs/SMOKE_TEST_RESULTS.json

---

📈 Roadmap

Phase A (Current)

• ✅ MVP implementation complete
• ✅ INPUT_SCHEMA aligned with WEDGE.md
• ✅ Worker integration complete
• ✅ Smoke test produces real results (DuckDuckGo)
• ✅ Actor #1 LIVE on Apify Store — Published 2026-09-03
• ✅ PPE Active — Pay-per-event pricing enabled

Phase B (Planned)

• Enable Google AI Overview + AI Mode engines
• Add Perplexity, ChatGPT Search, Copilot, Gemini
• RapidAPI listing (secondary marketplace)
• Scheduled runs (daily/weekly monitoring)
• Webhook notifications for new citations
• Historical comparison (detect citation changes over time)

---

🤝 Related Products

• ../../actor2/README.md — See which domains Google AI Overview cites for your target queries. Turn AI sources into a deduped lead table for competitor intel and link prospecting.

Both products share the same COGS model ($0.0075/prompt) and billing infrastructure.

---

📄 License

See LICENSE.txt in the project root.

---

🎯 Ready to Track Your AI Citations?

1. For testing: Run python smoke_test_actor_mvp.py locally
2. For production: Try the live actor on Apify Store
3. Deploy: cd apify-actor && apify push (for your own instance)
4. Run: Configure input in Apify Store and execute

Current version: 0.1.0 — MVP complete with spec-aligned output
Status: ✅ LIVE — Published to Apify Store on 2026-09-03 with active PPE pricing

---

Questions?
Review the docs/FAQ.md or check the docs/WEDGE.md for product strategy.