Free tools / AI Visibility Checker

The main check

Does AI mention your business?

free

Ask ChatGPT or Gemini to recommend a business like yours and it names a few. This tells you whether one of them is you — and, when it is not, exactly who took the slot.

Check your website

up to 5 questions · ChatGPT and Gemini · asked once each

Just the address — yourbusiness.com is enough. Nothing to install.

1

We read your website to work out what you sell and where you are.

2

We write up to five questions a customer of yours would actually type — and show them to you before we go.

3

We ask ChatGPT and Gemini those questions, and read the answers to see if you are in them.

This check runs on our servers, and it has to. A web browser cannot ask ChatGPT a question on your behalf. The four generators further down genuinely do run in your browser, and only those say so.

Northlight

example.com EXAMPLE DATA

4 of 4

Named in

≥1 measured answer

2 of 4

Cited

Gemini checks · ChatGPT was not asked to look

2

Mentioned only

from model memory

1

Not measured

excluded from counts

Example figures on a fictional business. Each question asked once per AI. Gemini is asked to go and look; ChatGPT answers from what it already knows, so a ChatGPT cell here can say named or missing but never cited.
Prompt ChatGPT Gemini Named instead
Best review software for contractors mentioned cited —
Top review management providers mentioned missing competitor-one.com
How much does review software cost? missing mentioned competitor-two.com
Northlight vs alternatives not measured cited —

Four counts, kept apart, never summed. The unmeasured half of the last question is left out of every count rather than scored as a no. Named-in and cited carry different denominators on purpose: naming is counted over both assistants, citation only over the one that was asked to fetch.

What comes back

This is the screen a scan feeds — the real product, on an invented brand. It is a deeper scan than the free one above, and the caption spells out the difference.

app.answermend.com/scan.html EXAMPLE DATA
Four scan tiles: 6 of 8 prompts named this brand, 4 of 8 cited it, 6 of 8 mentioned it, and a dashed tile reading 2 checks not measured of 32, excluded from all three and never counted as a miss. Beside them a next-step card: 7 items, 3 to act on.
A paid scan, not the free one. Eight questions put to four assistants — ChatGPT, Gemini, Claude and Perplexity — once each in this scan, 32 checks in all. The free check above asks up to five questions once each on ChatGPT and Gemini, and asks only Gemini to fetch — so on a free result, “cited” is measured on Gemini’s answers alone, and a ChatGPT cell can read named or missing but never cited. The dashed tile is the part to trust either way: two of the 32 checks here did not return, and they are reported as not measured and excluded from every count, never scored as a no. Northlight is invented; every figure is example data.
app.answermend.com/scan.html EXAMPLE DATA
The not-measured tile, close up: 2 checks not measured of 32 prompt-by-engine checks, excluded from all three denominators and never counted as a miss.
The fourth tile, close up. Two of the 32 checks on this example scan did not return — Perplexity was rate limited — so they land here and are left out of every count: not a zero, not a miss, and never counted against you. The other 30 are what the three tiles above were counted from.
Where we read the answers ChatGPT Gemini Claude Pro plan Perplexity Pro plan Engines this free check does not ask are named, not guessed.

Reading your result

Four things can happen for each question

Each question gets one of these four answers from each AI, and each one means something different for what you do next. The small grey words are the labels we use in the report; the sentence is the part that matters.

AI used your page as a source

The best outcome. The AI actually opened a page on your website and used it to answer the question. This is the only proof it read you rather than remembered you.

cited

AI named you from memory, but didn’t use your page

It knows your name, but it did not go and read your site. Usually that means it cannot get to your pages, or your pages do not answer that particular question. Fixable, and a different job from being unknown.

mentioned

You weren’t in the answer at all

A real no. The useful part is who was named instead — that is the business you are being compared with, on that question, today.

missing

We couldn’t check that one

Our request did not get through — the AI was busy, or slow, or we hit a limit. We tell you which. It is never counted as a no, and it never counts against you anywhere in your result.

not measured
What “missing” looks like from the customer’s side EXAMPLE DATA

whats the best crm for a small law firm

For a small law firm, the options that come up most often are:

1. competitor-one.com

matter management and trust accounting built in

2. competitor-two.com

cheapest of the three, with a client portal

3. competitor-three.com

strongest document automation

Your businessmissing

What you get back

A worked example, start to finish

This is a made-up coffee roaster, so you can see the shape of a real result before you run one. Every number here is invented for illustration.

Fern & Kettle Coffee — fernkettle.example

EXAMPLE DATA

Fern & Kettle was named in 2 of the 3 questions we asked.

We asked Gemini to go and look on 3 questions. On 2 of those it used a page on your website as a source — that is the part that proves it found you rather than remembered you.

ChatGPT answered from what it already knows and was never asked to fetch anything, so whether it read your site is not measured on this check — not a no, not a zero, and not counted in the number above.

1 check could not be completed. That is one question put to one assistant, not a whole question — and it is not a no. We could not get an answer, so it is left out rather than counted against the business.

example figures · each question asked once per AI · not a measurement of any real business

“which coffee subscription is best for a small office?”

ChatGPT named you. Whether it used a page of your site as a source was not measured here — that is not a no.

mentioned

Gemini used your page as a source.

cited

What to do: nothing on your side. One assistant used a page of your site as a source here, which is what you want. The other named you without being asked to go and look, so that half is not measured — not a no, and nothing has been counted against you.

“where can I buy freshly roasted coffee beans near me?”

You weren’t in ChatGPT’s answer. competitor-one.com was.

missing

You weren’t in Gemini’s answer. competitor-one.com and competitor-two.com were named on this question.

missing

What to do: put a page on your site that answers this exact question in your own words — where you are, what you sell, what it costs. Right now competitor-one.com has one and you do not.

“how much should good coffee beans cost per kilo?”

We couldn’t check ChatGPT for this one — it limited how many questions we could ask in a row.

not measured

Gemini used your page as a source.

cited

What to do: nothing on your side. The ChatGPT half did not run, so we have left it out instead of scoring it as a miss.

And what you do about it

The finding is half of it. In the app, a miss becomes an item like this — what to do and why, with the customer questions it came from.

app.answermend.com/fixes.html EXAMPLE DATA

wider than this screen — slide it sideways

One What-to-fix item, chipped gap: questions answered without this brand in the answer. Do this: check that each question has a page answering it in its own words, where an AI crawler can reach it. Because: the brand was neither named nor cited in 2 of the 8 answered questions. Then the two questions, and a button to record a fix.
One item from the what-to-fix list, on the same example scan as the report shot above — a paid scan, on the invented Northlight brand. What to do, why, and the questions it rests on — here, the two of eight answered questions where the brand was neither named nor cited, and the item names both. It hands you the change to make; whether an AI then names you is its decision, and nobody sells that honestly.

What we will not do

No score out of 100, and no promises

Two things you will see on other sites in this category, and will never see on a visibility result. Our free site checks do score a page out of 100 — a page either has a heading or it does not, and that answer does not change between runs. What an assistant says about you does, which is why no number stands in for it here.

We give you a count, not a score

Your result is a plain count: named in 2 of the 5 questions we asked — one snapshot of what these two AIs said on the day.

A score out of a hundred cannot be checked by you, or by us. It is a number we would have made up, and asking an AI the same question twice can give two different answers — so any decimal point in it would be fiction.

The free check asks each question once, so each answer is one sample rather than a rate. We say so on the result instead of dressing it up as a percentage.

Nobody can promise you a mention

Not us, not anyone. We can show you where you are missing, who is being named instead, and hand you the schema, robots and llms.txt files that make your site readable to an AI in the first place. We cannot make an AI say your name, and any company telling you otherwise is selling you something they do not control.

This free check does not cover Google’s AI Overviews — the answers at the top of Google. Google publishes no way to ask for them, so the only route is a paid search-data vendor. That route is on Pro, and it has not yet run against that vendor for real — so no figure from it appears anywhere, on any plan, until it has.

What we print instead — EXAMPLE DATA

2 of 5

named in 2 of the 5 questions we asked

cited ×1 mentioned ×1 missing ×2 not measured ×1

Four counts, kept apart. Adding them together is the number we refuse to print — and the one that would be easiest to sell.

Working today

Seven things you can use right now

All seven are finished and free. None of them measures whether AI mentions you — only asking the AIs does that — but each one fixes or checks something that has to be right before an AI can find and quote you. The four generators run in your browser:

Three more free checks fetch your pages from our server and hand you back what an assistant gets: the AI Crawler Checker (whether you are accidentally blocking AI from reading your site), the Render Check (whether your pages look empty to AI), and the AI Readiness Score (all of the above in one pass).

Fair questions.

cited
A page on your site was fetched and used as a source.
mentioned
Named from model memory. Your site did nothing.
missing
The answer ran and never named you. Someone else got the recommendation.
not measured
The check did not return. Reported with its reason and left out of every count.

What do I have to type in?

Your website address, and nothing else. If you know what your site is called — yourbusiness.com — you have everything you need.

What is the difference between “used your page” and “named you”?

“Named you” means the AI said your business name in its answer. It can do that from memory, from an old article, or from a competitor’s comparison page, without ever opening your website.

“Used your page” means your own web address turned up in the answer’s list of sources — proof that a page of yours was actually read. They are two different situations with two different fixes, so we keep them apart instead of merging them into one number.

Which AI assistants do you check, and how many questions?

The free check asks up to five questions across two assistants, ChatGPT and Gemini, once each — as many questions as your homepage supports, and the result says how many it asked. Repeating the same question several times and averaging the result is on the paid plans. Perplexity is on Pro, asked every third day. Google’s AI Overviews is on Pro. The route to it is a paid search-data vendor and we now buy that data. It has not yet run against that vendor for real, so no figure from it appears anywhere until it has.

Why don’t you give me a score out of 100?

Because we would be making it up. AI answers change from one run to the next, so a score with a decimal point in it would look precise and mean nothing. You get a count instead — named in three of the questions we asked — which is one snapshot rather than a rate. Because the answers move run to run, a single count is not a trend line; watching it move is what repeated runs on a paid plan are for.

Can you promise AI will mention my business?

No, and neither can anyone else. We show you where you are missing, who is being named instead, and what to change. Whether an AI then names you is its decision, not ours.

Who sees my results?

You do, and anyone you give the link to. Your check — your business, your competitors, and where you are losing — is not published, not used in our public data, and not used to train anything, unless you separately say yes. It will never be buried in a terms-of-use box.

The result comes with a link so you can come back to it or send it on. That link is a key rather than a password: it is not indexed and not listed anywhere, but whoever holds it can open the headline without signing in — so share it on purpose. It expires, and the date is printed next to it.

You cannot make an AI name you. You can make yourself readable.

One check is one day. An answer that moves is the reading-and-fixing part done again, and a second run to compare it against — the same questions asked again on a schedule, with the previous period beside them. That is what the plans are for. The pricing page lays out exactly what each one includes.

See what the plans include the check above stays free · no card needed

Shopping around is the sane thing to do here: eight vendors compared, every price read off the vendor’s own page and dated, with a section on when one of them is the better buy. Or keep going for free — the AI Readiness Score takes your own site apart check by check.