AI Visibility Tracker

Measure how often Google AI Overview, Perplexity, ChatGPT mention, recommend, and cite your brand.

Instead of treating one generated answer as a reliable ranking, AI Visibility Tracker repeats each question and reports a rate with a confidence range. It also keeps recommendations separate from mentions and citations, so you can see whether your brand was merely named, actively endorsed, or used as a source.

Use cases

• Measure brand mentions and explicit endorsements across AI answers.
• Compare a fixed question panel over time.
• Identify questions where competitors appear and your brand is absent.

Who it is for

• Brand and content teams tracking visibility in AI answers
• SEO, GEO, and AEO agencies monitoring clients
• Product marketers comparing share of voice against competitors
• Teams that want weekly change tracking instead of one-off screenshots
• Automated workflows that consume structured results through Apify

What you get

For every question and AI surface, the output includes:

• Brand mention rate across repeated checks
• Confidence range and a plain-language uncertainty label
• Recommendation rate, measured separately from mentions
• Citation rate for your domain and the cited domains found
• Average mention position when the brand appears
• Frequently mentioned competitors
• Change since the last comparable run
• Optional monthly search demand, missed searches, and estimated gap value
• Optional run-level summary report

Quick start

The default input uses demoMode: true, which returns sample output without live checks or charges.

To measure your brand:

{
  "brand": "Acme",
  "queries": [
    "what is the best project management tool for startups",
    "acme vs asana"
  ],
  "demoMode": false
}

Or start with a topic and let the Actor find relevant questions:

{
  "brand": "Acme",
  "seedTopic": "project management software",
  "maxDiscoveredPrompts": 1,
  "demoMode": false
}

GEO → AI visibility workflow

Use GEO Prompt Research to build a focused question panel, then pass that panel into this Actor for repeated measurement.

{
  "brand": "Acme",
  "promptPanel": {
    "panelId": "acme-core-questions",
    "panelVersion": "research-2026-09",
    "prompts": [
      { "text": "What are the best project management tools for startups?", "intent": "commercial" },
      { "text": "Which project management tool is easiest for a small team?", "intent": "comparison" }
    ]
  },
  "samples": 1,
  "trackChanges": true,
  "demoMode": false
}

The source panel ID and version remain visible in the results. The tracker also identifies the exact question content measured, so edited panels are not treated as the same measurement.

Inputs

Input	Type	Default	Description
brand	string	required	Brand name to find in AI answers. Matching is case-insensitive.
brandDomain	string	—	Domain to track in citations, such as acme.com.
queries	string[]	—	Questions to check directly.
seedTopic	string	—	Topic used to discover relevant questions.
maxDiscoveredPrompts	integer	1	Maximum discovered questions, from 1 to 25.
promptPanel	object	—	Inline named question panel.
promptPanelRunId	string	—	Accessible GEO Prompt Research run ID.
promptPanelDatasetId	string	—	Accessible GEO Prompt Research dataset ID.
surfaces	string[]	Perplexity	AI answer surfaces to check.
samples	integer	1	Repeats per question and surface, from 1 to 10.
country	string	us	Two-letter country code.
language	string	en	Two-letter language code.
device	string	desktop	desktop or mobile.
weightByDemand	boolean	false	Add search demand, CPC, missed searches, and gap value.
trackChanges	boolean	true	Compare with the last compatible complete run.
report	boolean	false	Produce a run-level summary report.
assumedCtrPercent	integer	20	Click-through assumption used only for estimated gap value.
demoMode	boolean	true	Return sample output without live checks or charges.

Use only one prompt-panel source at a time: promptPanel, promptPanelRunId, or promptPanelDatasetId. Direct queries and seedTopic can be combined with a panel.

Output

Measurement rows use recordType: "measurement". A requested summary uses recordType: "report".

Important measurement fields:

Field	Meaning
mentionRate	Share of usable checks that named the brand.
confidenceLow, confidenceHigh	Confidence range around the mention rate.
recommendationRate	Share of evaluable checks that actively recommended the brand.
recommendationSamples	Number of checks that could evaluate recommendation.
sourceReliability	Whether repeated checks returned consistent citation sources.
domainCitedIn	Number of checks that cited your domain.
averagePosition	Average order of the brand when mentioned, or null.
competitors	Other brands frequently present in the answers.
mentionRateChange	Change from the last comparable complete run, or null.
comparisonStatus	Why a comparison was or was not available.
monthlySearches	Search demand when demand weighting is enabled.
missedSearchesPerMonth	Estimated searches where the brand was absent.
gapValueUsd	Estimated monthly equivalent ad value, or null.
panelId, panelVersion	Identity of the exact question set measured.
sourcePanelId, sourcePanelVersion	Original GEO panel labels when supplied.

Example:

{
  "recordType": "measurement",
  "brand": "Acme",
  "query": "what is the best project management tool for startups",
  "aiAssistant": "aiOverview",
  "device": "desktop",
  "samples": 5,
  "mentionedIn": 3,
  "mentionRate": 0.6,
  "confidenceLow": 0.23,
  "confidenceHigh": 0.88,
  "recommendedIn": 1,
  "recommendationRate": 0.2,
  "domainCitedIn": 2,
  "averagePosition": 3,
  "comparisonStatus": "compared",
  "mentionRateChange": -0.2,
  "monthlySearches": 1600,
  "missedSearchesPerMonth": 640,
  "gapValueUsd": 2762.24
}

Pricing

AI Visibility Tracker uses Pay-Per-Event pricing. Current prices are shown on the Actor's Pricing tab in Apify Console.

Required configured prices:

Item	Configured price
Standard visibility check, including up to 5 samples	$0.12
Perplexity visibility check, including up to 5 samples	$0.18
Additional standard sample	$0.03
Additional Perplexity sample	$0.04
Demand weighting per priced question	$0.02
Summary report	$1.25
Discovered question	$0.05

Required prices: visibility-check $0.12; premium-visibility-check $0.18; extra-sample $0.03; premium-extra-sample $0.04; demand-weighting $0.02; prompt-discovery $0.05; visibility-report $1.25.

The Actor checks the run's spending cap before starting paid work. If the selected scope does not fit, reduce the number of questions, surfaces, samples, or optional features, or raise the cap.

Practical limits

• One sample is the quick-start default and is only indicative. For repeated measurement, select five samples and increase the run timeout.
• More questions, surfaces, and samples increase both price and runtime.
• ChatGPT can vary significantly between identical checks. Their rows include uncertainty guidance.
• Slow AI surfaces can make large runs exceed the run timeout. Split large panels into smaller runs when needed.
• Empty or unusable answers are excluded rather than counted as brand absences.
• Partial runs are returned, but they do not become the baseline for future comparisons.

The quick start defaults to Perplexity and one sample per question. Two questions have a conservative 29-minute allowance, fitting a 30-minute run timeout. One sample cannot measure repeatability. For five samples across both AI Overview and Perplexity, those two questions require a 352-minute allowance and a longer timeout. Each sequential AI Overview sample allows 20 minutes for retrieval; each other surface allows 12 minutes, plus 1.5 minutes per sample for extraction and orchestration. Optional discovery/demand adds time. Large panels (including 100 AI Overview questions) are rejected before vendor work if they exceed the remaining runtime. These allowances are not completion guarantees.

Privacy

• You provide only your brand, questions, and measurement settings. No customer API key is required.
• Change history and generated reports are stored on the Apify account running the Actor.
• Your measurement data is not used to create a shared cross-customer history.
• Public logs show progress counts and generic status messages, not brand names, questions, panel references, report links, or spend details.

Report rows contain reportStorage: {storeId, recordKey} (or null if rendering fails), never a generated public URL. Open the record through your authenticated Apify Console or API. Keys include run, snapshot, panel, locale, device and a unique artifact identifier, so subsequent runs do not overwrite earlier reports. Access and retention follow your Apify storage and sharing settings; anyone you grant access to the storage can read its contents.

Mention rates have uncertainty from small samples, correlated answers and extraction errors. Wilson intervals describe sampling uncertainty under an independence assumption; citation overlap does not prove factual accuracy.

FAQ

Is a mention the same as a recommendation?

No. A brand can appear in a list without being endorsed. Mention and recommendation rates are reported separately.

Why repeat each question?

AI answers vary. Repetition shows how often the result occurs and gives a confidence range instead of presenting one answer as a stable ranking.

Why can recommendation fields be null?

Recommendation is reported only when the answer could be evaluated by the extractor. null means it was not measured for that row, not that the recommendation rate was zero.

What happens when I edit the question panel?

The edited set is treated as a different measurement. The next run will not claim a before-and-after comparison against different questions.

Does device choice affect history?

Yes. Desktop and mobile checks are tracked separately.

Can I schedule this weekly?

Yes. Keep the brand, country, language, device, surfaces, and question panel stable to get useful changes over time.

Do I need API keys?

No. Buyers do not provide provider credentials.

Related Actors

• GEO Prompt Research — build and prioritize a reusable question panel
• Google Rank Tracker — track traditional organic search positions
• Keyword Fanout — discover long-tail questions and search phrases
• Keyword Volume — retrieve search volume and CPC