Skip to main content Scroll Top

AI Visibility Checker: Does ChatGPT Name Your Business?

Updated September 2026 · Written and maintained by the Progression Agency strategy team

A free AI visibility checker with no API cost and no credit limit. It writes the prompts, you run them in your own assistant account, and it reports whether you were named, linked, hedged — and who was named instead of you.

On this page · 10 sections
  1. What an AI visibility checker actually measures
  2. Why this asks you to run the prompts yourself
  3. The prompts matter more than the tool
  4. Being named is not the same as being cited
  5. Hedged mentions, and why they are a source problem
  6. What the competitor list is telling you
  7. Why the number moves even when nothing changed
  8. What to do with a result of zero
  9. How this differs from a paid AI visibility platform
  10. Where the sources an assistant reads actually come from

The short answerNobody can measure how often an assistant names you across everyone’s conversations, because nobody observes those conversations. What can be measured honestly is a sample. This checker generates six buying-intent prompts from what you do and where, you run them in ChatGPT, Gemini, Claude or Perplexity on your own account, and it analyses the pasted answers in your browser: named or not, position in the list, whether your domain was cited, whether the language was hedged, and which competitors were named instead. Being named in none of them is the usual result and it is fixable.

Assistant answers are not deterministic. The same prompt can return a different answer in a different session, account or region, so one reading is a sample rather than a score. Re-run monthly with identical wording and watch the direction. Search volumes and cost-per-click figures quoted are Ubersuggest, United States, September 2026.

What an AI visibility checker actually measures

It measures whether an assistant names your business when somebody describes what you do and asks who to use. That is a different question from where you rank on Google, and for a growing share of buyers it is the only question.

There is no public API that reports how often you appear across everyone’s conversations, and no vendor observes that either. What can be measured honestly is a sample: run real buying questions, read the answers, and count. That is what this does.

AI answer visibility checker

Find out whether ChatGPT, Gemini, Claude or Perplexity name your business when somebody asks for what you do. It writes the prompts; you run them on your own account and paste the answers back.

Results appear here.

Runs entirely in your browser. Nothing you paste is uploaded, logged or stored — which is also why it takes pasted text rather than a URL: a browser cannot read another site’s files without that site’s permission.

How the check worksHow the check works
The analysis is the valuable half and it runs locally. That is what keeps this free rather than metered per check.

Why this asks you to run the prompts yourself

Every automated checker calls a paid API behind the scenes and passes the cost on, either as a subscription or as a credit limit.

This one does not, which is why it is free with no cap. You run six prompts on your own account and paste the answers back. It takes about four minutes and it removes the one thing people reasonably distrust about automated scores: you saw the raw answer yourself.

What you get from doing it manually

  • You read the actual wording, which is usually more informative than the score.
  • You see which competitors the assistant reached for and in what order.
  • You notice the hedging, which no numeric score conveys.
  • There is no metering, so you can run it as often as you like.
  • Nothing about your business is sent anywhere.

Which assistants to test

Run the same prompts in at least two. ChatGPT and Gemini have the largest consumer reach; Perplexity cites sources most explicitly, which makes it the most useful for seeing where an answer came from; Claude is worth adding if your buyers are technical. Answers differ meaningfully between them, and a business invisible in one is sometimes prominent in another.

The prompts matter more than the tool

The single biggest mistake in checking AI visibility is asking the assistant about your brand by name.

“What do you know about Acme Plumbing?” will almost always produce something, because you have handed it the answer. It tells you nothing about whether you would have been discovered. The prompts that matter describe the need, not the supplier.

Ask the way a customer would

  • Describe the job, not the company: “who does bathroom remodels in Queens”.
  • Include the qualifier a real buyer would use: budget, urgency, speciality, area.
  • Ask for a shortlist, because that is what forces a naming decision.
  • Ask the comparison question, because that is where reasons get attached to names.
  • Keep one branded prompt at the end, as a check on what it knows once prompted.

Hold the wording constant when you re-test

Changing the phrasing changes the result, so a month-on-month comparison using different wording measures your rewriting rather than your visibility. The tool generates the same prompts from the same inputs for exactly this reason. Save them.

Two ways of asking, and what each one actually tests
Prompt styleExampleWhat it testsWorth tracking?
Branded lookupWhat do you know about Acme Plumbing?Whether it can recall you once namedWeakly
Category discoveryWho are the best plumbers in Queens?Whether you are discoverable at allYes
Qualified needI need an emergency plumber in Queens tonightWhether you fit a specific buying momentYes
ComparisonCompare the top plumbers in QueensWhether reasons get attached to your nameYes
Shortlist with linksGive me plumbers in Queens with their websitesWhether you earn a citationYes

Being named is not the same as being cited

A mention without a link rarely produces a visit. The assistant answered the question, and the person never needed to open a website.

This is the part most visibility reporting skips. The tool separates the two because the fixes are different: being named is a reputation and corroboration problem, while being cited is a source-quality and extractability problem.

What earns a citation rather than a mention

  • Being the original source of a specific fact rather than a restatement of a common one.
  • Publishing something that is not available elsewhere: your own measurements, your own pricing logic, your own data.
  • A page that answers the exact question in its first sentence, so it is quotable without editing.
  • A stable URL, because a citation to a moved page is a citation lost.
  • Scarcity: assistants cite when few sources cover a question well, and paraphrase when many do.

Why generic content never gets cited

If forty sites explain the same thing equally well, an assistant has no reason to attribute the answer to any of them. It states the fact and moves on. Citation follows source scarcity, not effort.

What this page's keywords cost to buyWhat this page's keywords cost to buy
Cost per click above thirty dollars on an informational keyword means vendors are bidding hard. The demand is real and it is commercial.

Hedged mentions, and why they are a source problem

When an assistant says a business “may” be worth contacting, or that it “could not verify” details, it is telling you something specific about its sources.

Hedging appears when the model has thin, inconsistent or stale information. It is not a judgement on your quality. It usually means your details disagree across the places it read, or that only one weak source mentions you at all.

The usual causes, in order of frequency

  1. Inconsistent name, address or phone across directories and your own site.
  2. A single thin source, so there is nothing to corroborate against.
  3. Stale information that contradicts your current offering.
  4. A generic description that could apply to any competitor.
  5. No independent mention anywhere that is not controlled by you.

What actually resolves it

Consistency and corroboration. The same business name, address, phone and description everywhere, plus at least a few independent sources saying the same thing. That is unglamorous work and it moves hedging more reliably than any content change.

Named — You appear. The first gate.
Position — Where in the list. Order carries weight.
Linked — A citation to you. Where visits come from.
Hedged — Qualified language. A source problem.
Absent — Not mentioned. The usual result.
Competitors — Who got it instead. Your real gap.
The six prompts the tool generates, and what each one is for
#Shape of the promptWhat it tests
1Who are the best [what you do] in [where]?Plain discovery: are you a candidate at all
2I need a [what you do]. Which companies would you recommend?Buying intent with a request for reasons
3What should I look for, and who does it well?Whether you are attached to quality criteria
4Compare the top options. Who is best for a small business?Whether you survive a direct comparison
5Give me a shortlist with their websitesWhether you earn a citation, not just a mention
6What do you know about [your business]?Recall once prompted, as a control

What the competitor list is telling you

The businesses named instead of you are the most useful output on the page, and the one people skip past.

They are not necessarily better. They are better represented in the sources these systems read. Looking at what those businesses have that you do not is a faster route to a fix than any amount of guessing.

What to look at, for each name on that list

  • Where are they listed that you are not?
  • Do they have independent coverage, or only their own site?
  • Do they publish anything specific enough to be quoted?
  • Is their category description more precise than yours?
  • Do they have review volume that corroborates a claim?

A caveat about the automatic list

If you leave the competitor field blank, the tool detects names by looking for capitalised phrases. That catches place names and product names as well as businesses. Entering your real competitors gives a clean comparison, and the tool says so in the output rather than presenting noise as data.

Named, cited, or absent — and what each one means
ResultWhat it meansWhat fixes it
AbsentNothing it read considers you a credible answerPresence and corroboration in third-party sources
Named, hedgedThin or conflicting information about youConsistency across listings, then independent mentions
Named, no linkYou are known but not the sourcePublish something specific enough to be worth citing
Named and linkedYou are a source it trusts for this questionHold the URL stable and keep the page current
Named firstYou are the default answerProtect it; watch monthly for displacement
Directories — Third-party lists. Heavily quoted.
Reviews — What customers wrote. Corroboration.
Your site — Extractable answers. The part you own.
News — Coverage and mentions. Independent proof.
Forums — Reddit and Q&A. Frequently cited.
Profiles — Consistent details. Entity clarity.

Want this done for your site?We build and maintain the search, content and paid programmes described on this page.

Get a free proposal

Why the number moves even when nothing changed

Assistant answers are not deterministic. The same prompt, the same account, an hour apart, can return a different list.

Personalisation, region, model updates and source churn all move the result. Anyone presenting a single AI visibility score as a stable measurement is either sampling much more heavily than they disclose, or overstating what one reading means.

How to get a reading you can act on

  • Run all six prompts rather than one.
  • Run them in at least two assistants.
  • Keep the wording identical between runs.
  • Record the date, because the answer is only true for that day.
  • Compare months, not sessions.
  • Treat a change of one mention as noise, not a trend.

What a good cadence looks like

Monthly is enough for most businesses. Weekly produces movement you cannot act on and invites over-reaction to noise. If you have just made a substantial change to your listings or published something significant, wait a few weeks before re-testing.

Same prompt — Different answer. Not deterministic.
Different account — Different result. Personalisation.
Different region — Different list. Geography.
Different day — Different sources. Index churn.
One reading — Not a trend. Sample of one.
Monthly — Direction. What to watch.
What this tool can and cannot tell youWhat this tool can and cannot tell you
The four it cannot do are the four most commonly sold as certainties. Assistant answers are not deterministic and no vendor observes all of them.

What to do with a result of zero

Being named in none of the answers is the most common outcome, especially for local service businesses, and it is not a verdict on the business.

It means the sources these systems read do not contain enough about you to make naming you a safe answer. That is a fixable problem and the order of the fix matters.

The sequence that works

  1. Make your name, address, phone and category identical everywhere they appear.
  2. Get listed accurately in the directories your category actually uses.
  3. Ensure your own pages answer the buying questions directly, in the first sentence.
  4. Publish at least one thing specific enough that nobody else has it.
  5. Earn independent mentions, because self-published claims corroborate nothing.
  6. Re-test after a month and compare against the same six prompts.

What not to do

  • Do not write pages addressed to the model rather than to a customer; it reads the same sources either way.
  • Do not add markup describing content that is not on the page.
  • Do not buy listings in bulk on low-quality directories; inconsistency is worse than absence.
  • Do not judge the result from a single prompt on a single day.

How this differs from a paid AI visibility platform

Paid platforms run the prompts for you on a schedule and store the history. That is a real convenience and it is what you are paying for.

What they cannot do is see more of the truth than you can. They sample the same non-deterministic systems from their own accounts, then present the sample as a score. Knowing that is worth more than the score itself.

What you actually buy with a subscription

  • Automation: the prompts run without you.
  • History: a stored record you did not have to keep yourself.
  • Breadth: more prompts and more assistants than you would run by hand.
  • Alerting: being told when something changes.
  • None of which is the same as observing all conversations, which nothing does.

When paying is the right call

If you are tracking dozens of terms across several markets, or you need the history for reporting to someone else, automation earns its fee. For a single business checking a handful of buying questions monthly, four minutes by hand gives you the same information and a clearer view of the wording.

The question to ask any vendor

Ask how many prompts, how often, from which accounts and regions, and whether the score is a sample or a census. Every honest answer is “a sample”. A vendor who will not answer plainly is selling certainty that does not exist.

Automated platforms against checking by hand
Paid platformThis checker
CostSubscription or per-creditFree, no cap
Who runs the promptsThe vendor, from their accountsYou, from your own account
What is measuredA sampleA sample
You see the raw wordingSometimesAlways
History kept for youYesNo, you record it
Sees every conversationNoNo

Where the sources an assistant reads actually come from

Assistants do not read your website and stop. They assemble an answer from whatever they consider reliable about your category, and your own site is only one input among several.

This matters because it changes where the work goes. A business can have an excellent website and be invisible, because nothing outside that website corroborates it.

The inputs that tend to carry weight

  • Directories and listings specific to your category.
  • Review platforms, for both volume and consistency of detail.
  • Independent coverage: news, local press, industry write-ups.
  • Community sources such as forums and question sites, which are quoted more than most people expect.
  • Your own pages, where they answer a question directly and are extractable.
  • Structured data, which removes ambiguity about who and where you are.

Self-published claims corroborate nothing

Saying you are the best on your own site is not evidence to a system weighing sources. Independent mentions are the difference between a claim and a fact, and they are the slowest part of this work, which is why they are worth starting first.

Want this tracked properly rather than spot-checked?

We run the same prompt set every month across several assistants, record what changes and work the sources behind it, so the number moves for a reason you can point at.

Get a free proposal

Not sure which of these applies to you?Tell us the situation and we will say plainly what we would do first, and what we would not.

Talk it through

Getting found in search

AI, AEO and what is changing

Paid media and lead generation

Websites and design

Choosing and working with an agency

Social, content and brand

By industry and by situation

Frequently asked questions

What is an AI visibility checker?
A tool that tests whether AI assistants name your business when somebody asks for what you do. This one generates the prompts, you run them in your own assistant account, and it analyses the pasted answers locally: whether you were named, in what position, whether you were linked, whether the mention was hedged, and who was named instead.
Is this really free, with no limits?
Yes. It makes no API calls, so there is nothing to meter. You run the prompts on your own account, which is why there is no cost to pass on and no credit cap.
Why do I have to run the prompts myself?
Because checking automatically means calling a paid API on every check, which is what every metered tool is charging you for. Running them yourself keeps it free and lets you read the raw answer rather than trusting a score.
Which AI assistants should I test?
At least two. ChatGPT and Gemini have the widest consumer reach, Perplexity cites its sources most explicitly which helps you see where an answer came from, and Claude is worth adding if your buyers are technical. Results differ meaningfully between them.
Why should I not just ask the assistant about my own business?
Because you have handed it the answer. A branded lookup almost always returns something and tells you nothing about whether you would have been discovered. The prompts that matter describe the need, not the supplier.
How many prompts do I need to run?
The tool generates six, which is enough for a usable sample. One prompt is not a measurement; assistant answers vary between sessions.
Why does the same prompt give different answers?
Assistant responses are not deterministic, and personalisation, region, model updates and source churn all move the result. The same prompt an hour later can return a different list.
How often should I re-run this?
Monthly. Weekly produces movement you cannot act on. Keep the wording identical between runs, record the date, and compare months rather than sessions.
What is a good AI visibility rate?
There is no published benchmark, and anyone quoting one is guessing. What matters is your own direction over time and how you compare against the specific competitors named in your results.
What does it mean if I am named but not linked?
You are known but you are not the source. The assistant answered from general knowledge and the person never needed to visit your site. Earning a citation requires publishing something specific enough that paraphrasing it is not sufficient.
What does a hedged mention mean?
Language like “may”, “appears to” or “I could not verify” signals thin, inconsistent or stale sources. It is a source problem rather than a judgement on quality — usually your details disagree across the places it read.
How do I fix a hedged mention?
Consistency and corroboration. Identical name, address, phone and description everywhere, then at least a few independent sources saying the same thing. It is unglamorous and it works better than content changes.
What should I do if I am named in none of the answers?
That is the most common result for local service businesses and it is fixable. In order: fix consistency across listings, get accurately listed where your category is actually covered, make your own pages answer buying questions directly, publish something nobody else has, then earn independent mentions.
Why does the competitor list include things that are not businesses?
If you leave the competitor field blank, names are detected by looking for capitalised phrases, which catches place names and product names too. Entering your real competitors gives a clean comparison, and the output says which method was used.
Does this tell me my share of all AI answers?
No, and no tool can. Nobody observes every conversation. Any product presenting a total share figure is extrapolating from its own sample without saying so.
Is this the same as checking Google AI Overviews?
No. AI Overviews are generated from Google’s search index and appear inside search results. Assistant answers are a separate surface with different sources. A business can appear in one and not the other.
Does blocking AI crawlers affect this?
It can. If you block the retrieval crawlers, you remove your own site as a source an assistant can read, while leaving every third-party mention of you in place. There is a separate free tool on this site for checking which crawlers your robots.txt actually allows.
Will better rankings on Google fix this?
Partly and indirectly. Assistants read many sources beyond search results, including directories, reviews and forums, so a page ranking well is neither necessary nor sufficient for being named.
Can I game this by writing content aimed at the model?
No. The assistant reads the same sources either way, and content written at a model rather than a customer tends to be generic, which is the one thing that reliably prevents citation.
Is anything I paste into this tool uploaded?
No. The analysis runs entirely in your browser and nothing you paste is stored. The only thing sent to us is the contact detail you enter to unlock the tools.

Want this done for your site?We build and maintain the search, content and paid programmes described on this page.

Get a free proposal

Get a free marketing proposal

Tell us what you are trying to grow and we will come back with a plan, not a pitch deck. Same-day reply on weekdays.

Our website uses cookies from third party services to improve your browsing experience. Read more about this and how you can control cookies by clicking "Privacy Preferences".
Privacy Preferences
When you visit our website, it may store information through your browser from specific services, usually in form of cookies. Here you can change your privacy preferences. Please note that blocking some types of cookies may impact your experience on our website and the services we offer.
Privacy Policy
You have read and agreed to our privacy policy
Required
Contact Us