AI Search Visibility Tracker — ChatGPT, Perplexity, Gemini & Google AI Overviews

Find out whether AI assistants recommend your brand. Give it the questions your customers ask ("best CRM for a small agency", "Notion vs Confluence") plus your brand and competitors. It asks ChatGPT, Perplexity, Gemini and Google AI Overviews and tells you, for every answer:

• Is your brand mentioned, how often, and in which position against your competitors (1 = named first)
• Is your website cited as a source, and at which position
• Which competitors are named instead, and which sites the AI relies on
• Per run: visibility %, share of voice, average position per engine, and the prompts where you are missing (your content gaps)

It's built for GEO / AEO (generative engine optimisation) tracking, brand monitoring and competitor research. Schedule it weekly to follow your AI visibility over time.

• All four engines in one run, one dataset, one report.
• $0.004 per answer with your own API keys for ChatGPT, Perplexity and Gemini. That's about 50× less than single-engine AI answer scrapers at $0.20, and you pay the AI provider directly, usually 1–4¢ per answer.
• Google AI Overviews need no key.
• Live web search answers. ChatGPT uses OpenAI's web search, Gemini uses Grounding with Google Search, and Perplexity uses Sonar. These are the same search-backed models behind the assistants.
• Failed answers are free. You pay only for answers delivered.

No API key? Run it with Google AI Overviews only, or get a free Gemini key at aistudio.google.com (1,500 free answers a day). Answers on our own keys (no key needed, from $0.02) are coming soon.

Quick start

{
  "prompts": [
    "best project management software for small teams",
    "what is the best free alternative to Trello"
  ],
  "brand": "Notion | notion.so | notion.com",
  "competitors": ["Asana | asana.com", "Trello | trello.com", "ClickUp | clickup.com", "Monday.com | monday.com | Monday"],
  "engines": ["chatgpt", "perplexity", "gemini", "google_ai_overview"],
  "repetitions": 2
}

Brand format: Name | website | other spellings. The website is how the actor knows whether the AI cites you. Add several domains if you have them. Other spellings (e.g. Monday for Monday.com) also count as mentions. Matching is case-insensitive and on whole words, so "Notion" doesn't match "notionally".

What you get

1. A report page (REPORT in the run's key-value store, linked on the Output tab): headline numbers, a brand × engine visibility table, share of voice, the prompts where you're missing with the competitors named instead, and a line per answer.

2. One dataset row per answer:

{
  "prompt": "best project management software for small teams",
  "engine": "perplexity",
  "engineName": "Perplexity",
  "model": "sonar",
  "repetition": 1,
  "brand": "Notion",
  "brandMentioned": true,
  "brandMentionCount": 1,
  "brandRank": 3,
  "brandCited": true,
  "brandCitationPosition": 3,
  "competitorsMentioned": ["Monday.com", "Asana"],
  "competitors": [
    {"name": "Asana", "mentioned": true, "mentionCount": 1, "rank": 2, "cited": true, "citationPosition": 2}
  ],
  "trackedBrandsInOrder": ["Monday.com", "Asana", "Notion"],
  "answer": "Popular choices include monday.com for visual workflows [1], Asana [2] and Notion...",
  "sourceCount": 3,
  "citedDomains": ["monday.com", "asana.com", "notion.com"],
  "sources": [{"position": 1, "url": "https://monday.com/blog/pm", "title": "Best PM tools", "domain": "monday.com", "cited": true}],
  "searchQueries": [],
  "usedOwnApiKey": false,
  "checkedAt": "2026-10-23T18:00:00+00:00",
  "error": null
}

• brandRank: your position among the tracked brands the answer names, by first mention. null means not mentioned.
• brandCited: true or false when you gave a website, null when you didn't.
• sources: the cited sources first, then sources the engine consulted without citing them (cited: false; ChatGPT and Perplexity report these).
• searchQueries: the web searches the engine ran to answer (ChatGPT, Gemini).

3. SUMMARY (JSON in the key-value store): the same numbers as the report, overall and per engine, ready for dashboards.

Pricing

Pay per answer delivered. There is no subscription, and failed or empty answers cost nothing.

Event	Price
ChatGPT, Perplexity or Gemini answer made with your own API key	$0.004
Google AI Overview analysed	$0.002
Coming soon: answer on our keys (ChatGPT $0.03, Perplexity $0.02, Gemini $0.04)	–

Google AI Overviews are fetched by Apify's official Google AI Overviews Scraper, which runs in your own Apify account. Apify bills that scraper separately at its own price, $0.001–0.003 per overview depending on your plan. Searches that show no AI Overview cost nothing here.

Example: 20 prompts × ChatGPT + Perplexity + Gemini × 2 repetitions with your own keys, plus AI Overviews = 120 × $0.004 + 20 × $0.002 = $0.52, plus your AI providers' fees and Apify's AI Overview fee. A single-engine scraper at $0.20 per answer would charge $24 for the same 120 answers.

Set a maximum cost per run in the run options. The actor stops asking before it would go over.

Input

Field	Default	Description
prompts	required	Questions, one per line. Up to 200 per run.
brand	required	Name | website | other spellings
competitors	–	One per line, same format. Up to 25.
engines	all four	chatgpt, perplexity, gemini, google_ai_overview
repetitions	1	Ask each prompt 1–5 times per engine. AI answers vary, so 2–3 gives a steadier visibility %.
country	US	Location for ChatGPT's web search (two-letter code).
includeAnswerText	true	Keep the full answer text.
includeSources	true	Keep the list of sources.
openaiApiKey, perplexityApiKey, geminiApiKey	–	Your own keys (stored encrypted), needed for ChatGPT, Perplexity and Gemini. Answers cost $0.004.
chatgptModel, perplexityModel, geminiModel	gpt-5-mini, sonar, gemini-2.5-flash	Advanced: any model of that provider with web search.

Good to know

• API answers vs the chat apps. ChatGPT, Perplexity and Gemini are asked through their official APIs with web search on. Answers come from the same models and search backends, but they can differ from what one logged-in user sees in the app, which is also personalised. Track the trend across runs and repetitions rather than a single answer.
• AI Overviews don't show for every search. Those prompts get a row with noAiOverview: true and aren't charged.
• Incomplete AI Overviews are retried. Google sometimes loads an overview's list or table separately, and the capture then misses the brands in it. The actor retries those searches once. Overviews that stay incomplete get aiOverviewPartial: true, are left out of the scores and aren't charged.
• Country applies to ChatGPT's search location. Gemini and Perplexity search without a set location, and AI Overviews use Apify's default Google settings.
• Prompts in any language work. Write them the way your customers would.

FAQ

How is this different from rank tracking? Search engines rank pages. AI assistants name a few brands and cite a few sources. This actor measures whether you are named and cited, and who is named instead.

Can I monitor this every week? Yes. Create a task with your prompts and schedule it. Each run's SUMMARY makes a dated data point; send it to Google Sheets or a webhook.

Do you store my API keys? No. Keys you enter are encrypted by Apify as secret input and used only for that run's requests.