LLM Citation Tracking: Monitor ChatGPT Sources, URLs and Brand Citations

Sellm captures cited URLs and domains whenever they appear in an AI response. Using the API, you can monitor citations across prompts, providers, countries, and repeated runs to identify meaningful patterns instead of relying on a single response.

Citation tracking differs from brand-mention tracking. A model may mention your brand without citing your website, or cite your content without explicitly recommending your brand. Effective AI visibility monitoring measures both.

What Is LLM Citation Tracking?

LLM citation tracking is the process of collecting and analyzing sources referenced in AI-generated answers. A citation may be a page on your website, a third-party article mentioning your brand, a review profile, an industry publication, or another source supporting an AI recommendation.

Tracking those sources helps answer questions such as:

• Does ChatGPT cite our website?
• Which pages from our domain are cited most often?
• Which third-party sources support competitor recommendations?
• Which domains influence AI answers in our industry?
• Are we gaining or losing citation coverage over time?

citedUrls

citedUrls contains the specific page URLs referenced in the provider response. Use it for page-level analysis, such as identifying which comparison page, review, or article was cited.

{
  "citedUrls": [
    "https://www.g2.com/products/acme-crm/reviews",
    "https://www.acme.com/blog/enterprise-crm-guide",
    "https://techcrunch.com/2026/01/acme-crm-series-b"
  ]
}

citedDomains

citedDomains is a deduplicated list of domains from those URLs. Use it to identify recurring source authority without analyzing every page individually.

{
  "citedDomains": [
    "g2.com",
    "acme.com",
    "techcrunch.com"
  ]
}

Some results also include richer provider metadata under enrichment. The stable fields intended for citation analysis are citedUrls and citedDomains.

How to Track Citation Sources Used by AI Models

A reliable citation-tracking workflow has five steps:

1. Define a stable prompt set. Include the informational and commercial questions customers ask in your market.
2. Run prompts across relevant providers. Compare ChatGPT, Perplexity, Claude, Gemini, and other platforms important to your audience.
3. Use repeated runs. LLM responses are non-deterministic, so one response is not a reliable trend.
4. Collect cited URLs and domains. Store each citation with its prompt, provider, location, and run date.
5. Compare patterns over time. Measure citation frequency by brand, domain, provider, and reporting period.

Avoid drawing conclusions from one answer. A domain cited once may be incidental; a domain cited across several prompts and repeated responses is a stronger signal that the provider relies on it for that topic.

ChatGPT Citation Tracking

A ChatGPT citation tracker records URLs and domains included in ChatGPT responses and connects them to the prompt, brand mentions, and recommendation position. This lets you see whether ChatGPT cites your website, which external pages support competitor recommendations, and how citation patterns change.

ChatGPT does not include sources in every response. Sellm records citation data when sources are present while retaining the surrounding result context. For citation-heavy research, compare ChatGPT with the Perplexity Tracker API.

How to Track Brand Citations in LLMs

Monitor both your own domain and external pages that mention your brand. An AI model might recommend your product while citing a review platform rather than your website. That citation still matters because it identifies a source associated with the recommendation.

Useful brand-citation metrics include:

• Citation frequency by URL and domain
• Third-party pages cited alongside your brand
• Citation rate by provider and prompt
• Competitor citation frequency
• New and lost cited sources

A simple citation rate is the percentage of analyzed responses that cite a given domain:

Citation rate = responses citing the domain ÷ total responses analyzed × 100

AI Citation Analysis for Specific Industries

Citation patterns vary by industry. Software recommendations may rely on review platforms and comparison sites, while healthcare, finance, and legal queries may favor regulators, professional organizations, research, and established publications.

To analyze citation sources in a specific market:

1. Build a representative set of commercial and informational prompts.
2. Include your brand and leading competitors.
3. Run the same prompts across providers and locations.
4. Aggregate cited domains and URLs by frequency.
5. Categorize sources as brand sites, directories, publications, forums, or research.

The resulting report shows which sources recur in AI answers for that market, where competitors have stronger source coverage, and which content or third-party placements may deserve investigation.

Enterprise LLM Citation Tracker with API Access

Sellm provides programmatic access to AI responses, brand mentions, recommendation positions, cited URLs, and cited domains. Organization-scoped API keys can analyze any brand by providing brandName with each request. Project-scoped keys use the analysis brand configured for that project and should omit brandName.

Submit a citation analysis

curl -X POST "https://sellm.io/api/v1/async-analysis" \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "brandName": "Acme CRM",
    "prompts": [
      "best CRM for small businesses",
      "CRM with strong reporting features"
    ],
    "providers": ["chatgpt", "perplexity"],
    "locations": ["US"],
    "replicates": 5
  }'

The API returns an analysis ID. Poll the result endpoint until the run succeeds, or provide a webhook when submitting the analysis.

curl -X GET "https://sellm.io/api/v1/async-analysis/{analysisId}" \
  -H "Authorization: Bearer YOUR_API_KEY"

Analyze the citation fields

{
  "prompt": "best CRM for small businesses",
  "provider": "chatgpt",
  "brandsMentioned": ["Acme CRM", "HubSpot"],
  "position": 2,
  "citedUrls": [
    "https://example.com/crm-comparison"
  ],
  "citedDomains": [
    "example.com"
  ]
}

See the API documentation for authentication, request limits, response fields, and webhook details.

Monitoring Lost and Changed Citations

Citation monitoring becomes more useful when results are compared over time. Run the same prompts with consistent providers, locations, and replicate counts, then compare the current cited URLs and domains with the previous period.

This reveals:

• Newly cited pages and domains
• Sources that stopped appearing
• Competitors gaining citation coverage
• Prompts where your citation rate declined
• Provider-specific changes in source selection

Sellm webhooks notify your application when an analysis succeeds or fails. Your application can compare the completed result with a previous run and trigger an email, Slack message, or internal alert when citation coverage changes.

Can a Tool Predict Whether an LLM Will Quote Your Content?

No tool can reliably guarantee that an LLM will quote or cite a specific page. Responses vary by model, prompt, location, model routing, and time.

Citation-tracking tools measure what models actually cited. Historical evidence can reveal recurring domains, content types, and source patterns, but it should not be presented as a deterministic prediction of future citations.

What to Look For in LLM Citation Tracking Tools

When evaluating an LLM citation tracker, check whether it provides:

• URL-level and domain-level citation data
• Multiple AI providers and geographic locations
• Repeated runs to account for response variability
• Brand and competitor analysis in the same response
• API access to normalized results
• Historical data that can be compared over time
• Webhooks or another way to automate downstream monitoring

Sellm includes citedUrls and citedDomains in completed analysis results whenever citation data is present. Usage is based on the number of prompts, providers, locations, and replicates in an analysis.