AI Search Visibility Tracker: ChatGPT, Perplexity, Gemini & Claude brand mentions

Find out whether AI assistants recommend your brand. Give it the questions your buyers ask ("What is the best CRM for small agencies?"), your brand, your domain and your competitors. The Actor asks OpenAI (the models behind ChatGPT), Perplexity, Google Gemini and Anthropic Claude through their official APIs, with your own API keys (or, for OpenAI, with no key at all), and tells you for every answer:

• is your brand mentioned, and in which position among all brands named,
• is your website cited as a source (and at which rank),
• which competitors were mentioned or cited instead,
• share of voice, simple sentiment, cited domains and the search queries the AI ran.

Plus a visibility SUMMARY: visibility %, average position, citation rate and share of voice per brand and per AI, the most-cited domains, and the prompts where only competitors show up. $10 per 1,000 checks with your own key, or $35 per 1,000 OpenAI checks without a key (AI costs included).

✨ Why this tracker

• ✅ Official APIs only, no ChatGPT scraping. Automating the ChatGPT or Perplexity website breaks their terms of use and can stop working any day. This Actor only calls the documented APIs (OpenAI Responses API with web search, Perplexity Agent API, Gemini API, Claude Messages API with web search).
• 🔑 Bring your own key, see your real costs. You pay the AI provider directly at their list price (for example about $2 per 1,000 answers on Perplexity, $10-15 on OpenAI with web search). No hidden markup on AI usage, and your provider bill shows every call.
• 🚀 Or start without any key. Leave the OpenAI key empty and the Actor uses its own key: one price per check, AI costs included.
• 🧮 Rankings, not just "yes/no". Position of your first mention among all tracked brands, mention count, share of voice and citation rank, per answer and aggregated.
• 🔁 Answers vary, so measure properly. Ask each prompt 1-3 times and get a visibility % instead of a lucky single answer.
• 📈 Monitoring mode. Schedule it weekly: every row says whether your brand was gained or lost, moved up or down, or gained or lost a citation since the last run. "Only changes" mode writes only what changed.
• 💸 You don't pay for failures. Invalid keys, exhausted quota, rate limits, timeouts and refusals are listed with the reason and are not charged. A wrong key stops calls to that provider right away.

Use cases: Brand monitoring · Analyze SEO & marketing (track search rankings) · Research markets & competitors (track competitors) · Developer tools (AI and LLM integrations).

🎯 Use cases

• Brands and marketing teams: do ChatGPT, Perplexity, Claude and Gemini recommend you for your category's buying questions? Which competitors do they name first?
• SEO / GEO / AEO teams (generative engine optimization): track AI visibility like rankings, find the sources AI cites for your topics (outreach and PR targets), and measure whether content changes move the needle.
• Agencies: white-label AI visibility audits per client and market (use one Monitoring ID per client, country and language).
• ChatGPT ads buyers: ChatGPT ads launched in 2026 (US first, Europe from 31 August 2026). Check how ChatGPT's models talk about you organically before and while you buy ads, and compare with competitors who advertise.
• Product and PR teams: see how AI describes your product (answer text, sentiment) after launches or press coverage.

🚀 How to use it

1. Get an API key for at least one provider (each takes 2 minutes, all are pay-as-you-go), or skip this step and run OpenAI without a key (see below):
   • OpenAI: platform.openai.com > API keys (starts with sk-). Add a few dollars of credit under Billing.
   • Perplexity: perplexity.ai > API Platform > API keys (pplx-).
   • Google Gemini: aistudio.google.com > Get API key (AIza...).
   • Anthropic Claude: platform.claude.com > API keys (sk-ant-).
2. Paste your prompts (one per line), your brand name, domains and competitors (Name | domain.com | Alias).
3. Tick the AI assistants and paste the matching API keys. Keys are secret inputs: Apify stores them encrypted, and the Actor never logs them or writes them to the output.
4. Click Start. Each check takes 5-60 seconds (web search included), with 5 running in parallel by default. Download the rows as CSV/Excel/JSON or open SUMMARY for the overview.
5. Optional: turn on Compare with previous run and add a Schedule (e.g. weekly) to monitor changes.

Try it without a key

Click Start on the prefilled example as it is: with no OpenAI key, OpenAI is asked with the Actor's own key. This works for any input, with these limits: OpenAI only (Perplexity, Gemini and Claude still need your own key and are skipped without one), model gpt-6-luna, at most 200 OpenAI checks per run, $0.035 per check with the AI cost included. Rows made this way have hosted: true, and RUN_SUMMARY shows hosted: true. Add your own OpenAI key at any time to pay $0.01 per check instead.

🔧 Input

Field	What it does	Default
prompts	Questions to ask, one per line (up to 1,000). Unbranded buyer questions show whether AI recommends you	2 examples
brandName	Your brand as it appears in answers	Apify (example)
brandAliases	Other names that count as your brand (product names, old names)	-
brandDomains	Your domains. Citations of the domain or any subdomain count (docs.apify.com counts for apify.com, notapify.com does not)	-
competitors	Name | domain.com, other.com | Alias 1, Alias 2 per line, or objects {name, domains, aliases} via API (up to 25)	4 examples
providers	openai, perplexity, gemini, anthropic	openai
openaiApiKey, perplexityApiKey, geminiApiKey, anthropicApiKey	Your own keys (secret). Providers without a key are skipped, except OpenAI, which then runs with the Actor's key ($0.035 per check)	-
openaiModel	Any Responses API model, e.g. gpt-6-luna, gpt-6.1-sol, gpt-6-astra (without your own key: gpt-6-luna only)	gpt-6-luna
perplexityModel	Agent API preset (fast, low, medium, high) or provider/model	fast
geminiModel	e.g. gemini-3.8-flash	gemini-3.8-flash
anthropicModel	e.g. claude-opus-5-5, claude-sonnet-5-5, claude-haiku-4-5	claude-opus-5-5
webSearch	Let OpenAI and Claude search the web and cite sources (like ChatGPT search). Off = model knowledge only	true
country	2-letter code sent to the web search as approximate user location (US, GB, DE...)	-
language	Adds "Answer in ..." (e.g. Deutsch)	-
runsPerPrompt	1-3 answers per prompt and provider	1
answerMaxChars	Characters of answer text stored per row (0 = none). Analysis always uses the full answer	4000
caseSensitive	Only the capitalized brand counts (for names like Notion or Box)	false
monitoring / onlyChanges / monitoringId	Compare with the previous run of the same Monitoring ID; optionally output only changes	off
maxConcurrency	Parallel API calls (1-20)	5
requestTimeoutSecs	Timeout per API call (retried up to 3 times)	120

{
  "prompts": ["What are the best CRM tools for small agencies?", "HubSpot alternatives for startups"],
  "brandName": "Acme CRM",
  "brandAliases": ["Acme"],
  "brandDomains": ["acmecrm.com"],
  "competitors": ["HubSpot | hubspot.com", "Pipedrive | pipedrive.com", "Zoho CRM | zoho.com | Zoho"],
  "providers": ["openai", "perplexity", "anthropic"],
  "openaiApiKey": "sk-...",
  "perplexityApiKey": "pplx-...",
  "anthropicApiKey": "sk-ant-...",
  "runsPerPrompt": 2,
  "country": "US"
}

📦 Output

One row per check (prompt x AI provider x run). Example below (illustrative values from a test with a mocked OpenAI response, competitors and answer shortened):

{
  "prompt": "What are the best web scraping tools?",
  "provider": "openai",
  "providerName": "OpenAI",
  "model": "gpt-6-luna",
  "run": 1,
  "brand": "Apify",
  "brandMentioned": true,
  "brandPosition": 2,
  "mentionCount": 2,
  "brandCited": true,
  "brandCitationRank": 2,
  "sentiment": "positive",
  "shareOfVoice": 0.4,
  "mentionedBrands": ["Bright Data", "Apify", "Octoparse"],
  "competitors": [
    { "name": "Bright Data", "mentioned": true, "mentionCount": 2, "position": 1, "cited": true, "citationRank": 1 },
    { "name": "Zyte", "mentioned": false, "mentionCount": 0, "position": null, "cited": false, "citationRank": null }
  ],
  "citedUrls": ["https://brightdata.com/products", "https://docs.apify.com/platform", "https://www.g2.com/categories/web-scraping"],
  "citedDomains": ["brightdata.com", "docs.apify.com", "g2.com"],
  "citationCount": 3,
  "searchQueries": ["best web scraping tools 2026", "web scraping platform comparison"],
  "webSearchUsed": true,
  "answer": "Here are the most popular web scraping tools in 2026:\n\n1. **Bright Data** - ... 2. **Apify** - a reliable, scalable platform ...",
  "answerLength": 503,
  "usage": { "inputTokens": 3120, "outputTokens": 412, "webSearches": 2 },
  "error": null,
  "checkedAt": "2026-10-05T11:10:50Z"
}

• brandPosition: 1 = your brand was the first tracked brand named in the answer.
• shareOfVoice: your mentions / mentions of all tracked brands in that answer (0-1).
• sentiment: positive / neutral / negative from a simple word list around your mentions (a hint, not a judgement), null when not mentioned.
• With monitoring on, each row also has change: {"status": "changed", "details": ["position improved 3 -> 1"], "previous": {...}}.
• Failed checks are listed with error and errorType (auth, quota, rate_limit, timeout, refusal...) and are free.

The SUMMARY record (key-value store) has per-brand visibilityPct, avgPosition, firstPlacePct, citationRatePct, shareOfVoice, a breakdown per AI provider, a leaderboard, topCitedDomains (with the owning brand when it is yours or a competitor's) and promptsWhereOnlyCompetitorsAppear. RUN_SUMMARY has counts, charged checks and errors per provider.

💵 Pricing: $10 per 1,000 checks with your own key, $35 without

A check = one prompt answered by one AI provider once (one row). Platform compute is included, plus Apify's tiny $0.00005 per-run start fee. Failed checks (invalid key, no quota, rate limit, timeout, refusal) are free.

Mode	Price per successful check	AI provider cost
Your own key (any provider)	$0.01 (event check)	Paid by you to the provider
No OpenAI key (OpenAI, gpt-6-luna, max 200 per run)	$0.035 (event hosted-check)	Included

Both modes can be mixed in one run: for example OpenAI without a key plus Perplexity with your own key.

AI provider costs are separate and paid by you to the provider with your own key. Rough list prices per 1,000 answers with the defaults (October 2026, check the providers' pricing pages): Perplexity fast preset about $1.5-3; OpenAI gpt-6-luna with web search about $10-15 ($10 per 1,000 web search calls + tokens); Gemini gemini-3.8-flash (no web search) about $2-4; Claude Opus 5.5 with web search about $50-100 ($10 per 1,000 searches, up to 5 searches per answer, + tokens; Sonnet 5.5 or Haiku 4.5 cost much less). Turning web search off makes OpenAI and Claude much cheaper.

Example: 50 prompts x 3 providers x 2 runs = 300 checks = $3 on Apify per run. Cap spending with Max charge per run: the Actor checks the remaining budget before every API call and stops before making a call your limit can't pay for.

❓ FAQ

Is this allowed? Do you scrape ChatGPT? No scraping. The Actor only calls the providers' official, documented APIs with your own key, which is exactly what those APIs are for. It never automates chatgpt.com, perplexity.ai or gemini.google.com. Your use of the answers is governed by your agreement with each provider (for example, OpenAI and Anthropic ask you to show citations when you display web-search answers to end users, and Perplexity asks you to credit it when you publish its output).

Is the API answer the same as what ChatGPT shows? Close, not identical. The ChatGPT app adds its own system prompt, memory, personalization and model routing, which no official API exposes. API answers with web search are the closest compliant proxy, and they are consistent run to run, which is what matters for tracking trends.

Why are there no citations for Gemini? Google's Gemini API terms do not allow storing, analysing or monitoring answers grounded with Google Search, or collecting their links. So Gemini is asked without Google Search grounding: it tells you what the model itself knows and recommends. OpenAI, Perplexity and Claude return citations.

Why do results change between runs? AI answers are not deterministic and web results change. Use Runs per prompt = 2-3 and look at the visibility % in SUMMARY rather than a single answer.

What about privacy and my keys? Keys are secret inputs (encrypted by Apify), sent only to the provider's own API endpoint in a request header, never logged, and never written to the dataset, SUMMARY or RUN_SUMMARY. Prompts go to the providers you select; OpenAI and Gemini requests are sent with storage turned off (store: false). On Google's free tier, Google may use prompts to improve its products. Use prompts without personal data.

Which questions should I track? Unbranded buying questions ("best X for Y", "X alternatives", "X vs Y", "how to do Z") show whether AI recommends you. Branded questions ("Is Acme CRM good?") show how AI describes you.

My brand name is a common word. Add precise aliases, turn on Case-sensitive matching, or use a distinctive form (e.g. "Notion app"). Matching is whole-word; spaces, hyphens and dots are interchangeable, and when two names overlap the longer one wins ("Google Cloud" is not counted as "Google").

Something is wrong or missing? Open an issue on the Issues tab with the prompt, provider and model. Fixes usually land within 24-48 hours.

📝 Changelog

• 1.1 (2026-10-05): OpenAI without your own key ($0.035 per check, AI cost included, max 200 per run). Rows and RUN_SUMMARY have hosted.
• 1.0 (2026-10-05): First release. OpenAI Responses API (web search), Perplexity Agent API, Gemini Interactions API, Claude Messages API (web search). Mentions, position, citations, share of voice, sentiment, SUMMARY, monitoring mode, failed checks free.

This Actor is not affiliated with or endorsed by OpenAI, Perplexity, Google or Anthropic. ChatGPT, Perplexity, Gemini and Claude are trademarks of their owners.

🤝 Need a custom version?

I build custom scrapers, scheduled data feeds and integrations (CSV, Excel, Google Sheets, API, MCP for AI agents). Tell me the sites and fields you need:

• Email: bence.kadi@gmail.com
• Apify profile: https://apify.com/kadi_bence

Related Actors

More low-cost Actors by the same developer, built on official APIs and public data:

• SEO Site Audit — 0-100 SEO score per page, broken links and fix hints
• PageSpeed Insights Bulk Checker — Core Web Vitals and Lighthouse scores in bulk
• Website Tech Stack Detector — CMS, ecommerce, analytics and frameworks of any website
• Document to Markdown — PDF, Word, PowerPoint and Excel to clean Markdown, with OCR
• Website Screenshot & PDF API — full-page screenshots and PDFs of any URL
• Bulk WHOIS & RDAP Domain Lookup — registrar, dates and DNS for many domains
• Apple App Store Scraper — top charts, app details, developers and reviews

All my Actors: apify.com/kadi_bence