Visibility Overview
One score for how often the engines name your brand when buyers ask about your category, tracked over time.
AI Search Intelligence
The AI visibility tool for ChatGPT, Gemini, Perplexity and Claude
ChatGPT, Gemini, Perplexity and Claude now answer the questions your buyers used to type into Google. Metric Vault measures whether those answers mention you, cite you, and describe you correctly, then shows you how to improve.
ChatGPT
Gemini
Perplexity
Claude
Google AI Overviews
The shift
Buyers are asking AI first
A growing share of people ask an assistant before they open a search engine. If the answer does not mention you, you are invisible at the exact moment someone is choosing. Classic SEO tools do not measure this surface. Metric Vault does, right next to your Google rankings, so you are not guessing about half of how people find you now.
Ten blue links
The buyer picks one, and you can count the click.
One AI answer
AI answer
Three tools are usually recommended for this, and the write-ups cite a mix of vendor documentation and community threads.
your domain
rival.com
reddit.com
No impression, no click, no referrer. Either you are named or you are not.
What gets measured
Six signals, taken from answers an engine actually returned
Classic analytics cannot tell you whether an assistant named your brand, because there is no impression to count and no referrer to read. The only way to know is to ask the engines and record what they say.
Together those turn AI visibility tracking into a measurement rather than an impression. Every number comes from an answer an engine actually returned, stored with its date, so a change in the score is traceable to a change in the answers rather than to a change in a model.
What you can see
Nine ways this AI visibility tool measures your brand
Every tool below runs on real prompts sent to the live engines, not a guess or a cached score.
Prompt and Question Research
The real questions buyers ask AI in your space, and whether your brand shows up in the answers.
Brand Performance
How you stack up against named competitors across the engines: who gets mentioned, who gets cited, who gets left out.
Brand Perception and Narrative
Not just whether you appear, but what the engines say about you: the sentiment and the story they tell.
Prompt Tracking
Track a set of prompts over time and get alerted the moment your visibility moves up or down.
Google AI Overview Tracker
See whether your pages show up in Google's AI Overviews, and which competitors are cited instead of you.
AI Citation Tracker
Which engines cite your domain as a source, and on which answers, so you know what is actually working.
Forum and Reddit Visibility
AI engines lean on Reddit and forums. See where your brand is part of those conversations, and where it is missing.
Competitor Research
Pick any rival and see the prompts where they win the answer, so you know exactly where to catch up.
How it runs
AI visibility tracking over time
AI visibility tracking only means anything as a trend, so every pass sends real prompts to the live engines on a schedule and keeps the answers.
Visibility score
stored runs
1
Real prompts, live engines
Each pass generates category questions that do not name you, puts them to ChatGPT, Gemini, Perplexity, Claude and Google AI Overviews, and records what came back: whether you appeared, in what position, in what words, and which sources were cited alongside you. Nothing is read from a cached index.
2
Every run is stored
History is built from your own measurements rather than a vendor estimate. A background pass re-measures saved brands roughly once a day, about every twenty-two hours, taking up to five brands at eight prompts each and starting with the ones you have used most recently. Running a tool yourself moves that brand to the front of the rotation. The trend chart plots up to thirty stored points.
3
Proof, then alerts
Because the verbatim answer is stored with each run, a change in the score arrives with the text that caused it. On the Alerts tab you can set a drop threshold for a brand and be emailed when its visibility falls below it, which is usually how a lost citation gets noticed before a customer is the one to mention it.
Two surfaces
AI search visibility vs Google rankings
These are two different surfaces, and both decide whether you get considered at all.
Google ranking
You can see the impression
Tells you where your page sits once someone types a query and looks at a list of results. You can see the impression, the position and the click, and you can watch all three move.
AI answer
The journey ends before the list
Often ends the journey before that point. The assistant names two or three brands, explains each in a paragraph, and the person never opens a results page. Nothing in your analytics records that you were left out, because there was no impression to lose and no click to miss.
Tracking both side by side is what makes either number actionable. A page can rank third and still never be cited, which points at how it is written and structured rather than at its authority. A page can be cited constantly and rank poorly, which points the other way entirely. Seeing a ranking drop and a lost citation in the same workspace is how you tell those two situations apart, and they call for different fixes. To see where a single page stands before committing to anything, run the free AI visibility checker on it.
The obvious question
Why not just ask ChatGPT yourself?
You can, and you should, once. It tells you what one model said to one person on one day. Everything that makes the answer useful as a measurement is what happens after that.
One answer is one sample
Ask the same question twice and the wording moves, and sometimes the brands do too. A single reply cannot tell you whether you are usually named or got lucky. The same prompts are put to the engines on a schedule so the number is a rate, not an anecdote.
Your own account is the worst test
Your history, your memory and your location all shape what you are shown, and typing your own brand name into the box guarantees it appears in the answer. The prompts here are category questions that never name you, so the count is not flattered by the way it was asked.
A number matters as a trend
Being named in six answers out of ten means nothing until you know it was eight last month. Every run is stored with its date and the verbatim answer, so a change in the score arrives with the text that caused it rather than as a figure you have to take on trust.
One engine is a quarter of the picture
ChatGPT, Gemini, Perplexity, Claude and Google AI Overviews disagree with each other constantly, and they cite different sources to get there. Checking one by hand tells you about one. All five are asked the same questions in the same pass.
What you do with it
A score you cannot act on is a vanity metric
Measurement is the first half. Each of these turns a number into a specific piece of work, and each one names the prompt or the page it came from.
01
Close the citation gaps
The engines cite specific URLs to build an answer. When a rival is quoted on a question and you are not, that gap is a named page on a named prompt, which is a brief rather than a hunch. Sorting by the prompts with the most gaps is the shortest route from the score to a piece of writing.
02
Read the share of voice, not just your own score
The same prompts are scored for every brand in the set, so you can see who is being recommended in the place where you are not. A rising competitor and a falling category look identical from your own number alone and completely different side by side.
03
Approve the prompt set before it runs
The questions are generated from your category and shown to you as a list you edit and approve, because a measurement is only as good as what was asked. Nothing is tracked on your behalf until you have read the prompts it will be tracked on.
04
Set a threshold and stop watching
Set a drop threshold per brand and get an email when visibility falls below it. That is usually how a lost citation gets noticed before a customer is the one to mention it, and it is the difference between a dashboard you have to remember and one that tells you.
Questions
Frequently asked questions
What is an AI visibility tool?
An AI visibility tool measures whether AI assistants name your brand when someone asks about your category. Rather than tracking a position in a list of links, it sends real questions to engines like ChatGPT, Gemini, Perplexity and Claude, then records whether you appeared in the answer, how you were described, and which sources the engine quoted. It exists because a growing share of buying research now finishes inside an assistant, where there is no impression, no click and no referrer for your analytics to capture.
How is AI visibility scored?
Metric Vault reports a visibility score out of 100 next to the raw evidence behind it. The score reflects how many of the prompt and engine combinations named your brand, and the appearances figure is shown as a fraction so you can always see the denominator. Scores are grouped into bands: 80 to 100 is labelled Dominant, meaning you are the default answer in your category, while anything under 30 is flagged as weak. Sentiment and engine coverage sit beside the score rather than being folded into it, because a negative mention with real volume is a different problem from a low count.
Which engines are tracked?
Five: ChatGPT, Gemini, Perplexity, Claude and Google AI Overviews. Coverage is reported as a figure out of five alongside your score, so a low number tells you that engines are not configured rather than that you performed badly on them. Google AI Overviews is counted separately from the three chat assistants on purpose: it is the one surface that sits directly above the ordinary search results, so being absent from it costs you visibility on a page you may already rank well on.
How often is the data refreshed?
Any tool you run measures live at the moment you run it. On top of that, a background pass re-measures saved brands roughly once a day, about every twenty-two hours, covering up to five brands at eight prompts each and choosing the ones used most recently. History accrues from every stored run, so the trend keeps filling in whether the measurement came from you or from the scheduler.
See it on your own site, free
Run the free AI Visibility Checker on any domain and get a 0 to 100 readiness score in under two minutes. No account, no card.
Start measuring your AI visibility
AI Search Intelligence is one section inside Metric Vault, alongside classic SEO, competitor, content and reporting tools.