LLM Brand Visibility Tracker – ChatGPT, Perplexity, Gemini (AI Search Rank Tracker)

Does AI recommend your brand? This Actor asks AI answer engines the questions your customers ask, such as "What is the best CRM for small businesses?", and tells you whether your brand appears in the answer, at which rank, your share of voice against competitors, whether your website is cited as a source, and the sentiment of the answer. Use it for GEO (Generative Engine Optimization), AEO and AI SEO monitoring for $0.01 per check.

Engines (all with live web search, called through their official APIs):

Engine	Model (OpenRouter ID)	Web search
Perplexity	perplexity/sonar	built in
OpenAI (ChatGPT model family)	openai/gpt-5-mini	OpenRouter web search
Google Gemini	google/gemini-2.5-flash	OpenRouter web search
Any other model (Claude, Grok, Llama, ...)	your OpenRouter ID in Extra models	added automatically

Honest note: results are API answers with web search. They can differ from what the consumer apps (chatgpt.com, perplexity.ai, gemini.google.com) show a logged-in user, because those apps add personalization, memory, location and their own system prompts. This Actor does not scrape ChatGPT or Google, so it does not break their terms. AI answers also change from run to run, so track trends over many prompts and schedule runs instead of reading too much into one answer.

What you get per prompt x engine

• mentioned: whether your brand or one of its aliases is named
• rank: position in the recommendation list (rankSource: "list"), or order of first mention when the answer has no list
• shareOfVoice: your mentions divided by mentions of you plus your competitors
• competitorsMentioned, competitorRanks, competitorMentionCounts
• domainCited and brandCitationUrls: whether your domain is one of the cited sources
• citationUrls: every source URL the engine cited
• sentiment (positive / neutral / negative / mixed) with a short reason, plus an excerpt around your brand
• answer (full text), model, tokens, llmCostUsd
• changed, previousMentioned, previousRank for scheduled monitoring

There is also one summary row per engine: mention rate, average rank, top-3 rate, share of voice, domain citation rate, sentiment breakdown, competitor share of voice and the domains cited most often. The same summary is saved to the OUTPUT record of the key-value store.

Use cases

• AI search visibility / GEO reporting for your brand or for agency clients
• Competitor benchmarking: who do ChatGPT, Perplexity and Gemini recommend instead of you?
• Content and PR targeting: topCitedDomains shows which sites AI engines trust in your niche, so you know where to get mentioned
• Scheduled monitoring with Only save changed results: get a row only when you enter or drop out of an answer
• Product launches: check whether AI already knows your new product

How to use

1. Enter your brand, optional aliases, your domain and competitors.
2. Add the prompts your customers ask, written without your brand name. You can also leave prompts empty and fill Category + Audience to auto-generate up to 10 prompts such as "best X for Y".
3. Pick the engines and click Start. Each prompt x engine is one check.

Input example

{
  "brand": "Notion",
  "aliases": ["Notion AI"],
  "domain": "notion.com",
  "competitors": ["Obsidian", "Evernote", "Coda", "ClickUp", "Confluence"],
  "prompts": [
    "What are the best note-taking apps for startup teams?",
    "Which wiki or knowledge base tool should a small remote team use?"
  ],
  "engines": ["perplexity", "chatgpt", "gemini"]
}

Output example (one check)

{
  "type": "check",
  "engine": "perplexity",
  "model": "perplexity/sonar",
  "prompt": "What are the best note-taking apps for startup teams?",
  "mentioned": true,
  "rank": 1,
  "rankSource": "list",
  "shareOfVoice": 0.42,
  "competitorsMentioned": ["Obsidian", "Evernote", "ClickUp"],
  "competitorRanks": {"Obsidian": 3, "Evernote": 4, "ClickUp": 2},
  "domainCited": true,
  "brandCitationUrls": ["https://www.notion.com/product"],
  "citationUrls": ["https://www.notion.com/product", "https://zapier.com/blog/best-note-taking-apps/"],
  "sentiment": "positive",
  "sentimentReason": "Recommended as the most flexible all-in-one workspace",
  "excerpt": "1. Notion - an all-in-one workspace for notes, docs and wikis ...",
  "changed": true,
  "error": null
}

Pricing

Pay per event, with no subscription:

Event	Price
visibility-check: one prompt answered by one engine and analyzed	$0.01
summary-report: per-engine summary, once per run	$0.02

Failed checks are free. The run stops cleanly at your maximum total charge per run, and checks that do not fit the budget are not started.

LLM tokens and web search are billed separately to your Apify account by Apify's OpenRouter Actor, so you need no OpenAI, Perplexity or Google key. The typical cost per check on a paid Apify plan is about $0.006 with Perplexity Sonar and about $0.01-0.015 with GPT or Gemini plus web search, mostly the search fee of the provider. Sentiment adds about $0.0001.

Example: 20 prompts x 3 engines = 60 checks = $0.60 + $0.02 summary + about $0.60 tokens, so roughly $1.20 per weekly report.

Compared with alternatives (Apify Store, October 2026)

Tool	Method	Price per check
This Actor	official APIs + web search	$0.01 + tokens (about $0.02 total)
ChatGPT / Perplexity search scrapers	scrapes the consumer UI	$0.20 per result
Other API-based AI rank trackers	APIs	$0.05 to $0.30 per check
SaaS brand-monitoring suites	subscription	from about $199/month

FAQ

Is this the same answer I see in ChatGPT? It is not guaranteed to be. It is the answer of the OpenAI model through the API with web search enabled. The consumer app uses other system prompts, memory and personalization. The API approach is reproducible, legal and cheap, which makes it good for trends and comparisons.

Why are answers different on every run? LLMs are not deterministic and web results change. Use 10 or more prompts and look at rates such as mention rate and share of voice over time.

How is the rank computed? The Actor finds the main list in the answer (numbered headings, numbered items, bullets or bold product names) and reports the position of the item that names your brand. If the answer has no list, rank is the order of first mention among your brand and competitors (rankSource: "mentionOrder").

Should prompts contain my brand name? No. Neutral prompts measure organic visibility. Prompts with your brand name, such as "Is Notion good for startups?", are useful for sentiment only.

Can I monitor daily or weekly? Yes. Create a schedule and enable Only save changed results. The previous state is kept in a named key-value store (State store name). Use one name per brand.

Which other models can I add? Any OpenRouter model ID, for example anthropic/claude-sonnet-4.5 or x-ai/grok-4.7. Web search is added automatically.

Why did a run fail with HTTP 401/402? The run token could not call apify/openrouter, or your Apify account has no credit for tokens. The run ends with a clear error row and nothing is charged.

Use with AI agents (Apify MCP)

This Actor works as a tool in the Apify MCP server: an agent can call llm-brand-visibility-tracker with brand, competitors and prompts and read the summary rows to answer "How visible is my brand in AI search compared with competitors?". You can also call it through the Apify API, or run it on a schedule and send the dataset to Google Sheets, Slack or a webhook.