AI Visibility Tracker: brand mentions and citations in ChatGPT, Gemini, Perplexity and Google AI Overviews

Find out whether your brand shows up in AI answers. This AI visibility tracker asks ChatGPT, Gemini, Google AI Mode, Google AI Overviews and Perplexity the prompts your buyers ask and reports, for every answer, whether your brand is mentioned, your position versus competitors, whether your site is cited as a source and which domains the AI relies on, with a share-of-voice summary per engine. Built for generative engine optimization (GEO/AEO) and SEO teams; pay per answer, no subscription.

Built for:

• SEO and GEO agencies: AI-visibility reports for clients, a baseline before GEO work, and proof of progress after it. Schedule it weekly and chart the trend.
• Brand and content marketers: see which competitors AI recommends instead of you, and which sites (reviews, listicles, forums) feed those answers, so you know where to get covered.
• Founders and product marketers: check "best X for Y" and "X vs Y" prompts in your category in minutes.

You pay per answer. Failed requests are free, and so are prompts where an engine gives no answer (for example, no AI Overview shown).

What you get

	Per answer (type: "answer")	Per run (type: "summary", free)
Brand	brandMentioned, brandMentions, brandPosition (1 = mentioned before every tracked competitor), brandRankAmongAllBrands (ChatGPT), excerpt around your first mention	brandMentionRate, brandCitationRate, avgBrandPosition per engine and for all engines
Citations	brandCited, brandCitationRank, brandCitedUrls, sources (rank, url, domain, title), topCitedDomains, competitorsCited	citationShare per brand, topCitedDomains across answers
Competitors	competitorsMentioned (in order of appearance), competitorMentions	shareOfVoice and mentionsByBrand
Answer	answerText (markdown), model, fanOutQueries (ChatGPT), checkUrl, scrapedAt
Google extra	aiOverviewShown, googleOrganicRank and googleOrganicUrl (your classic top-10 rank for the same prompt)

Engines and data source

Engine	Data source
ChatGPT	answer as shown on chatgpt.com, with web search
Gemini	answer as shown on gemini.google.com
Google AI Mode	AI Mode answer and its references
Google AI Overview	AI Overview on the normal results page, plus the organic top 10
Perplexity	Sonar model with web search

All answers come through DataForSEO's licensed API.

Input examples

Basic check (one prompt, all five engines):

{ "brandName": "HubSpot", "brandDomains": ["hubspot.com"], "prompts": ["best crm for small business"] }

Agency report: 20 prompts, competitors, each prompt asked twice:

{
  "brandName": "Notion", "brandDomains": ["notion.so", "notion.com"], "brandAliases": ["Notion AI"],
  "competitors": ["Confluence | atlassian.com", "Coda | coda.io", "ClickUp | clickup.com", "Obsidian | obsidian.md"],
  "prompts": ["best note taking app for teams", "notion alternatives", "best wiki software for startups"],
  "engines": ["chatgpt", "gemini", "google_ai_mode", "google_ai_overview", "perplexity"],
  "country": "us", "repeats": 2
}

UK, Google only (AI Mode and AI Overviews):

{ "brandName": "Monzo", "brandDomains": ["monzo.com"], "competitors": ["Revolut | revolut.com", "Starling | starlingbank.com"],
  "prompts": ["best bank account for freelancers uk"], "engines": ["google_ai_mode", "google_ai_overview"], "country": "gb" }

Tips

• Use the prompts your buyers really type: "best X for Y", "X vs Y", "X alternatives", "is X worth it", "how to do Y".
• Add your domains so citations are detected. Subdomains (blog., help.) count.
• Answers vary between runs. Use repeats: 2–3 for a reliable mention rate. A weekly schedule shows trends.

Output example (one answer, shortened)

{
  "type": "answer", "status": "ok", "prompt": "best crm for small business", "engine": "chatgpt",
  "engineLabel": "ChatGPT", "model": "gpt-6", "repeat": 1, "brand": "HubSpot", "country": "United States",
  "brandMentioned": true, "brandMentions": 4, "brandPosition": 1, "brandRankAmongAllBrands": 1,
  "brandCited": true, "brandCitationRank": 1, "brandCitedUrls": ["https://www.hubspot.com/products/crm"],
  "competitorsMentioned": ["Zoho CRM", "Pipedrive", "Freshsales"], "competitorsCited": ["Zoho CRM", "Pipedrive"],
  "excerpt": "…My recommendation: Start by comparing HubSpot, Zoho CRM, and Pipedrive…",
  "topCitedDomains": ["hubspot.com", "spotsaas.com", "help.zoho.com"],
  "sources": [{ "rank": 1, "url": "https://www.hubspot.com/products/crm", "domain": "hubspot.com", "title": "..." }],
  "answerText": "…", "checkUrl": "https://chatgpt.com/?q=best%20crm%20for%20small%20business&hints=search",
  "charged": true, "scrapedAt": "2026-10-17T01:48:40Z"
}

Summary row (shortened):

{ "type": "summary", "engine": "all", "answers": 5, "brandMentionRate": 1.0, "brandCitationRate": 0.2,
  "avgBrandPosition": 1.4, "shareOfVoice": { "HubSpot": 0.278, "Zoho CRM": 0.278, "Pipedrive": 0.222, "Salesforce": 0.111, "Freshsales": 0.111 } }

Ready dataset views: AI answers, Citations and Share of voice. Export to CSV, Excel or JSON, or use the API. Plug it into Looker Studio or Sheets for client reports.

Pricing (pay per event)

Event	When it is charged
Actor start	once per run
ai-answer	one engine's answer to one prompt, delivered and analysed

Each prompt × engine × repeat is one answer. Example: 20 prompts × 5 engines = 100 answers.

You are not charged for:

• prompts where an engine returned no answer, or Google showed no AI Overview (the row is still delivered, with your organic rank);
• failed requests;
• skipped inputs (duplicates, invalid domains, unknown engines);
• summary rows;
• anything beyond your run's maximum charge. The run stops cleanly and lists what it didn't process.

Apify free plan: 1 prompt across all selected engines, no repeats, per run. Any paid Apify plan unlocks up to 200 prompts, 3 repeats and 1,000 answers per run.

Use from AI agents (MCP)

AI assistants such as Claude, ChatGPT and Cursor can run this Actor through Apify's MCP server (https://mcp.apify.com). The agent can look up the Actor, read its input schema, run it and read the results. Runs use your own Apify account and are charged at the prices on this page.

• Actor name: vncntlabs/ai-search-visibility-tracker
• Server URL with only this tool: https://mcp.apify.com?tools=vncntlabs/ai-search-visibility-tracker. The default https://mcp.apify.com also works; the agent can then search Apify Store and call the Actor by name.
• Claude (desktop app or claude.ai): Settings > Connectors > Add custom connector, paste the server URL, then sign in to Apify.
• ChatGPT: turn on Developer mode, then Settings > Apps & Connectors > Create, paste the server URL and choose OAuth.
• Cursor: run /add-plugin apify in Cursor chat and connect the Apify MCP server under Cursor Settings > Tools & MCPs, or add the server to mcp.json:

{ "mcpServers": { "apify": { "url": "https://mcp.apify.com?tools=vncntlabs/ai-search-visibility-tracker" } } }

Example prompt: "Use the Apify Actor vncntlabs/ai-search-visibility-tracker to check whether my brand Acme (acme.com) is mentioned or cited for the prompt "best project management tool for small teams"."

To cap the cost of an agent's run, it can pass maxTotalChargeUsd in the run options.

FAQ

How is "position" calculated? It is the order of first mention among your brand and the competitors you list. 1 means the AI names you before every listed competitor. For ChatGPT, brandRankAmongAllBrands also ranks you among every brand ChatGPT itself tagged.

What about prompts that name my brand ("HubSpot vs Salesforce", "HubSpot alternatives")? The answer naturally repeats your name, so such rows carry brandInPrompt: true. Filter them out for a pure "is AI recommending us" rate, or keep them to see how AI describes you.

What counts as a mention? Your brand name or aliases as a whole word (case-insensitive), or your domain written in the text. A link alone doesn't count as a mention; it counts as a citation.

Why do results differ from what I see? AI engines personalise and vary their answers. These are logged-out answers for the chosen country and language at scrapedAt. Use repeats and schedules for trends.

Which countries? 24, including the US, UK, Canada, Australia, India, Germany, France, Spain, Italy, the Netherlands, Brazil, Mexico and Japan. Perplexity answers aren't location-targeted.

How long does a run take? ChatGPT and Gemini take about 30 seconds per answer; Google and Perplexity take a few seconds. Answers run in parallel, so 100 answers usually finish in a few minutes.

Is this affiliated with OpenAI, Google or Perplexity? No. Data comes from DataForSEO's licensed API. Not affiliated with or endorsed by OpenAI, Google or Perplexity.