Free tool · no accountExperimental
What three AI assistants say about your brand when nobody is watching
Your next buyer asks an assistant before they ask you. Brandscope puts the same five questions on your behalf and shows the answers — what they get right, what your own page contradicts, and what nobody can check.
# Your page
—
# ChatGPT
—
# Claude
—
# Gemini
—
# Knows you
—
A free allowance per visitor, no account needed.
Why a brand needs this measured at all
A buyer used to arrive at your site with a question. Now they arrive at an assistant, ask it what you are and how you compare, and reach you — if they reach you — already holding an answer somebody else wrote. That answer is assembled from whatever the model absorbed about you, which may be a press release from two years ago, a competitor's comparison page, or nothing at all.
None of it is visible from your side. Your analytics show the visit that followed, never the sentence that decided it. A brand can rank first for its own name and still be described to every buyer as something cheaper, smaller or older than it is, and nothing in the marketing stack reports that.
This makes the sentence visible. Five questions a buyer actually asks, three assistants, and the answers laid beside what your own page says — so the difference between the two is a thing you can read rather than a thing you suspect.
How to read the result
Three numbers and a verdict, and each one answers a different question.
- Knows you
- How many of the fifteen answers named the brand as something the assistant knows. An answer that repeats your name while saying it has never heard of you does not count — that distinction is the whole reading, and getting it wrong is the easiest way to publish a comforting number.
- Positioning match
- How much of your own homepage language the assistant's description shares. Low is not automatically bad: a short, specific description can share few words with a long one. It is a prompt to read the two sentences side by side, not a grade.
- The verdicts
- Matches means your page says the same thing. Contradicted means your page says otherwise, and only a price can earn it — numbers are the one claim a machine can check. Unverified means we could not check it, which is not the same as false and is never presented as one.
What this is not
- Not our opinion of the brand. Every line is a model's output, dated, beside the question that produced it.
- Not a ranking. The assistants are not ordered best to worst; they are three readers who disagree.
- Not a claim about anything your page does not state. What we cannot check, we call unchecked.
- Asked in English. What an assistant knows about a company is mostly in English, and asking in another language measures the translation as well as the brand.
Questions people ask
Which models are asked?
One model from each of three vendors — OpenAI, Anthropic and Google — kept the same between runs so two readings a month apart are comparable. The version is not shown, because which model we run is our operating decision and a version number on a public page is stale within a month.
Why five questions and not fifty?
Because the same five every time is what makes a second reading mean something. A generated set would make every run a different measurement wearing the same name, and the difference between March and April would be the questions rather than the brand.
Is a score of zero a failure of the tool?
No, it is the reading, and usually the most useful one. Three assistants were asked and none could say what the company is. That is a fact about how the brand is represented, and it is actionable in a way a high score is not.
Why is a wrong answer not called a hallucination?
Because we can only check what your own page states. A price your site contradicts is contradicted and says so. A complaint nobody can verify is unverified. Calling an unchecked claim false would be making a claim about your business that we cannot support.
Do you store the result?
The reading is kept so the same brand asked about again this week is answered from it rather than paid for again, and so a trend exists later. It holds a hashed address and a public company name, never a person.
Asked out loud
spoken, not typedThe same worry in the words people use speaking to an assistant rather than typing into a box. Each one carries the situation it came from, because the situation is what makes it a different question.
Ask it yourself and you get one answer, from one assistant, on one phrasing. This runs five fixed questions past three of them and lays the answers beside what your own page says — so the difference between the two is something you can read rather than something you suspect.
before a board meetingSometimes, and this is the one thing that can be checked mechanically: your page states numbers, the assistant states numbers, and if the two sets do not overlap the answer says contradicted and shows what your page says. Where your site is silent, it says unverified instead of guessing.
after a customer quoted a wrong priceUsually because the material it absorbed is older or thinner than your current positioning — an old release, a comparison page written by a competitor. The positioning column shows how much of your own homepage language the assistant shares, and the two sentences sit side by side so you can see which words it is missing.
when the description feels wrongThat is the most useful reading this produces. Three assistants asked and none could say what the company is means the work is getting known at all, not correcting a detail — and it is a starting number you can measure against later.
early-stage brandThe words on this page
- AI visibility
- Whether an assistant names a brand when asked a question the brand should own.
- Hallucination
- A confident statement a model produces with nothing behind it. Checkable only where a source exists to check it against.
- GEO
- Generative engine optimisation: the work of being described correctly by the systems that answer instead of listing.