ChatGPT Brand Visibility & Citation Tracker

Track whether a brand appears in current ChatGPT-style answers and whether its official website is cited. The Actor uses a fixed internal openai/gpt-4o-mini route through OpenRouter with the openrouter:web_search server tool, so users cannot change the provider model or increase cost through input.

This is a directional GEO/AEO benchmark, not a guarantee of what every ChatGPT user will see. Results vary by model, prompt, search index, time, location, and provider settings.

Setup

Add OPENROUTER_API_KEY as a private Actor environment secret. Do not put the key in Actor input. OPENROUTER_HTTP_REFERER is optional and can be used to set your OpenRouter attribution URL.

Inputs

• brandName — required tracked company, product, or brand.
• brandDomain — optional official domain used for citation detection.
• industry and location — optional context for three generated discovery prompts.
• queries — optional custom prompt list, up to 5. If supplied, it replaces generated prompts.
• competitors — optional names to detect in the same answers.
• useWebSearch — defaults to true; set false only when testing model-only behavior.

The Actor makes at most five sequential OpenRouter requests per run. The provider model is fixed internally to openai/gpt-4o-mini, which keeps the model-driven cost envelope predictable; OpenRouter input/output and web-search charges still depend on the provider's current rates.

Apify pricing is pay-per-event: $0.20 for one complete brand-visibility-report after at least one prompt succeeds and the report is written. Failed all-request runs do not emit or charge a report event. The current pricing configuration also passes Apify platform usage through to the user; OpenRouter provider charges are separate.

Output

The dataset contains one report record with:

• visibilityScore from 0–100: 70% successful-answer mention rate plus 30% official-domain citation rate when a domain is provided.
• queryResults with the exact prompt, answer, mention count, context snippets, citations, competitor mentions, fixed model used, and error state.
• mentionRate, citationRate, queriesCompleted, competitorMentions, recommendations, and warnings.

The Actor output schema exposes this report through the default dataset link in the run's Output tab and API output object.

Failed provider requests are not treated as missing brand mentions. If every request fails, the Actor fails honestly with a diagnostic instead of emitting a misleading empty report. A missing API key or invalid runtime input exits with a warning so configuration mistakes do not inflate the Actor's failure rate.

Example input

{
  "brandName": "Notion",
  "brandDomain": "notion.so",
  "industry": "project management software",
  "location": "United States",
  "competitors": ["Asana", "Trello", "ClickUp"]
}

For repeatable monitoring, keep the same location and custom prompt list across scheduled runs and compare the underlying queryResults as well as the score.

Responsible use

Use the Actor for legitimate brand research, content measurement, and competitive analysis. Do not infer market share or factual truth from a single generated answer. Validate material decisions against primary sources and respect OpenRouter, model-provider, and source-site terms.

Local development

npm install
npm test

The local tests cover mention parsing and scoring. A live API test requires an authorized OPENROUTER_API_KEY and incurs provider costs.