AI GEO Recommendation Tracker — Audits & Reports

AI GEO Recommendation Tracker is an Apify Actor that audits how ChatGPT, Claude, Gemini, and Perplexity mention and cite your brand, then turns the evidence into prioritized generative engine optimization (GEO) recommendations for agencies and brand teams. It returns mentions, position, share of voice, citations, and sentiment as JSON, CSV, or Excel, plus an HTML report.

Hosted free-plan limit: because every check calls an AI model through OpenRouter, Apify free-plan runs deliver at most 1 result (one prompt-engine check) per run. Paid plans use the configured prompt and engine limits.

What does the AI GEO recommendation tracker do?

• Asks four AI answer engines your buyer questions and records whether, how often, and where your brand is mentioned.
• Ranks your brand against listed competitors by first appearance and computes share of voice.
• Extracts cited URLs and domains, flags whether your domain is cited, and lists cited competitor domains.
• Classifies the sentiment of your first mention as positive, neutral, or negative.
• Compares each run with the previous run for the same trackingId (gained, lost, improved, declined).
• Produces a GEO opportunity score (0–100) and prioritized recommendations for visibility, citation, sentiment, position, and coverage.
• Saves an agency-ready HTML report with score cards and per-engine breakdowns.

Which AI models does it use?

As of October 2026, each engine runs one fixed model through OpenRouter:

Engine	OpenRouter model ID	Web grounding
perplexity	perplexity/sonar	Native search
chatgpt	openai/gpt-6-luna	Web plugin (Exa), up to 3 results
claude	anthropic/claude-haiku-5.5	Web plugin (Exa), up to 3 results
gemini	google/gemini-3.5-flash-lite	Web plugin (Exa), up to 3 results

The Actor owner pays model and web-search costs inside the event price, so you do not need an OpenRouter or OpenAI account or API key. Models are fixed and cannot be changed through the input.

What data does it return?

Each dataset item is one prompt-engine check:

Field	Type	Description
brand	string	Tracked brand
brand_domain	string or null	Brand domain
engine	string	perplexity, chatgpt, claude, or gemini
model_used	string or null	Model that answered; null if the check failed
prompt	string	Prompt sent
mentioned	boolean	Brand appears in the answer
position	integer or null	Rank by first appearance among brand and competitors
share_of_voice	number or null	Brand share (0–100) of tracked-brand mentions
cited	boolean	Your domain is cited or named
citations	array	Source URLs in the answer
brand_cited_url	string or null	First cited URL on your domain
mention_count	integer	Sentences mentioning the brand
mention_context	string or null	First mentioning sentence (max 300 characters)
sentiment	string or null	positive, neutral, or negative
competitors_mentioned	array	Listed competitors in the answer
topic_keywords	array	Topics for content-gap analysis
citation_domains	array	Unique cited domains
competitor_domains_cited	array	Listed competitor domains among citations
previous_mentioned	boolean or null	Mention status in the previous run
previous_position	integer or null	Position in the previous run
position_change	integer or null	Positive means the brand moved up
visibility_status	string	new, gained, lost, improved, declined, unchanged, absent, or error
error	string or null	Failure reason; null on success
checked_at	string	ISO 8601 timestamp

The key-value store also holds SUMMARY (scores, per-engine stats, top cited domains, stop_reason), RECOMMENDATIONS (score breakdown and prioritized actions), and REPORT (HTML).

Use cases

• SEO and GEO agencies producing a repeatable AI visibility audit and client report per account.
• Content teams finding buyer questions where AI answers leave them out, and what to publish next.
• Product marketers comparing position and share of voice against named competitors.
• Digital PR teams finding which domains AI engines cite in their category.
• Consultants running monthly GEO reviews with run-over-run trend labels.

When to use it and when not to

Use it when you want measurement and an action plan in one run. For rank and share-of-voice monitoring only, use AI Visibility Rank Tracker. To inventory cited URLs, use AI Citation Monitor. To score individual pages, use GEO Content Auditor.

It is not a Google rank tracker, site crawler, or backlink index, and it does not read the consumer ChatGPT, Claude, Gemini, or Perplexity apps. Recommendations cannot guarantee that a model changes its answers.

How to use the AI GEO recommendation tracker

1. Open the Actor in Apify Console and click Try for free.
2. Enter your brand name and domain.
3. Add 3–10 real buyer questions, or an industry to auto-generate prompts.
4. List direct competitors and their domains, then click Start.
5. Open the dataset, SUMMARY, RECOMMENDATIONS, and REPORT in the run's storage.

Field	Type	Default	Description
brandName	string	Notion	Brand to track (required)
brandDomain	string	notion.so	Your domain, for citation detection
industry	string	productivity software	Auto-generates prompts when prompts is empty
prompts	array	[] (Console form prefilled with one example question)	Questions; each runs on every selected engine
platforms	array	all four (Console form prefilled with chatgpt)	Any of perplexity, chatgpt, claude, gemini
competitorBrands	array	[]	Competitors for position and share of voice
competitorDomains	array	[]	Competitor domains for citation gaps
generateHtmlReport	boolean	true	Save REPORT
trackingId	string	brand slug	Links runs for trend comparison
useWebSearch	boolean	true	Web grounding for ChatGPT, Claude, Gemini
webSearchResults	integer	3	Web results per check (1–3)
maxPrompts	integer	25	Prompt cap per run (1–200)

{
  "brandName": "Notion",
  "brandDomain": "notion.so",
  "industry": "productivity software",
  "prompts": [
    "What is the best note-taking and productivity app?",
    "Recommend the top productivity software for teams."
  ],
  "platforms": ["perplexity", "chatgpt", "claude", "gemini"],
  "competitorBrands": ["Evernote", "Obsidian", "Coda"],
  "competitorDomains": ["evernote.com", "obsidian.md", "coda.io"],
  "trackingId": "notion-geo-audit",
  "generateHtmlReport": true,
  "useWebSearch": true,
  "webSearchResults": 3,
  "maxPrompts": 25
}

Output example

{
  "brand": "Acme Notes",
  "brand_domain": "example.com",
  "engine": "chatgpt",
  "model_used": "openai/gpt-6-luna",
  "prompt": "What is the best note-taking app for small teams?",
  "mentioned": true,
  "position": 2,
  "share_of_voice": 50,
  "cited": true,
  "citations": ["https://example.com/features"],
  "brand_cited_url": "https://example.com/features",
  "mention_count": 1,
  "mention_context": "Acme Notes is a flexible option for small teams.",
  "sentiment": "positive",
  "competitors_mentioned": ["Globex Docs"],
  "topic_keywords": ["collaboration", "pricing"],
  "citation_domains": ["example.com"],
  "competitor_domains_cited": [],
  "previous_mentioned": false,
  "previous_position": null,
  "position_change": null,
  "visibility_status": "gained",
  "error": null,
  "checked_at": "2026-10-20T12:00:00.000Z"
}

Use it via API, MCP, and AI agents

curl -X POST "https://api.apify.com/v2/acts/muhammadafzal~ai-geo-rec-tracker/run-sync-get-dataset-items" \
  -H "Authorization: Bearer $APIFY_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{"brandName":"Notion","brandDomain":"notion.so","prompts":["What is the best note-taking app?"],"platforms":["chatgpt","perplexity"]}'

import { ApifyClient } from 'apify-client';

const client = new ApifyClient({ token: process.env.APIFY_TOKEN });
const run = await client.actor('muhammadafzal/ai-geo-rec-tracker').call({
  brandName: 'Notion',
  prompts: ['What is the best workspace tool for startups?'],
  competitorBrands: ['Evernote', 'Obsidian'],
});
const { items } = await client.dataset(run.defaultDatasetId).listItems();
const recs = await client.keyValueStore(run.defaultKeyValueStoreId).getRecord('RECOMMENDATIONS');

MCP clients such as Claude, ChatGPT, and Cursor can call it through https://mcp.apify.com?tools=muhammadafzal/ai-geo-rec-tracker. Apify schedules, webhooks, and integrations with Make, Zapier, n8n, and Google Sheets can run audits on a cadence and forward results.

How much does it cost?

Pay-per-event prices, active since September 20, 2026 (no change scheduled):

Event	FREE	BRONZE	SILVER	GOLD	PLATINUM	DIAMOND
prompt-engine-check	$0.050001	$0.04875	$0.0475	$0.04	$0.04	$0.04
actor-start	$0.006251	$0.006094	$0.005938	$0.005	$0.005	$0.005

actor-start is charged once per run after the input is validated, regardless of memory. A prompt-engine-check is charged only after its result is saved. Failed checks and rejected input are not charged. Model costs, web search, and Apify platform usage at the default 1,024 MB memory are included.

• Free plan, prefill input: one check, $0.006251 + $0.050001 = $0.056252.
• BRONZE, 4 prompts × 4 engines (16 checks): $0.006094 + 16 × $0.04875 = $0.786094.

Set a maximum total charge to cap spend; the Actor stops before a check it cannot charge.

Limits and reliability

• Free-plan runs deliver one result. Paid runs check up to 200 prompts on four engines.
• Checks run one at a time and usually take a few seconds each; raise the run timeout above the default 300 seconds for audits of more than about 40 checks.
• Answers come from model APIs, not signed-in consumer apps; location and personalization are not simulated.
• AI answers vary between runs and days. Compare trends across runs, not single answers.
• Mention detection is case-insensitive text matching; sentiment uses a keyword lexicon.
• If the model provider rejects requests, the run stops, keeps delivered results, and states the reason.

FAQ

Do I need an OpenRouter or OpenAI API key? No. Model access is included in the event price.

How often should I run it? Weekly or monthly with the same prompts and trackingId.

Why do results differ between runs? Models sample answers and web results change, so compare repeated runs.

Can I schedule it? Yes. Use an Apify schedule with a fixed trackingId.

Are the recommendations written by an AI model? No. Rules generate them from the collected metrics, with no extra charge.

How is it different from AI Visibility Rank Tracker? That Actor monitors rank and share of voice; this one adds the GEO score, recommendations, and HTML report.

Related Actors

• AI Visibility Rank Tracker: AI brand monitor with rank and share of voice.
• AI Citation Monitor: which sources ChatGPT and Perplexity cite.
• GEO Content Auditor: page-level AI answerability score.
• AEO Tracker: answer engine optimization tracking.
• ChatGPT Brand Visibility Tracker: ChatGPT-only mentions and citations.
• AI Brand Reach Checker: quick check of AI brand mentions.

Legal and responsible use

The Actor sends your prompts to AI model APIs and analyzes the answers; it does not access private accounts. Use it for lawful research, keep personal data out of prompts, and do not promise clients guaranteed AI placement.

Support

Report problems or request features in the Actor's Issues tab. Feedback is welcome.