LLM Visibility Checklist

Brand name Exactly as you want it written. If you cannot answer that in one spelling, that is the first finding.

Entity clarity Everything here is on your own property, which means all of it is fixable this month.

The name is written identically everywhereOne spelling, one capitalisation, one decision about the Ltd or Inc. Three variants is three weaker entities instead of one. An About page states plainly what you areIn one sentence, near the top: what you sell, to whom, and what you are not. Not the founding story. Organization schema on the home pagename, url, logo, description. It states in machine-readable form what your design already says to humans. sameAs points at your real profilesLinkedIn, Crunchbase, GitHub, X, the app stores. This is the line that ties the scattered mentions of you together. Name, address and phone agree wherever they appearOld local-SEO advice that turns out to matter for the same reason it always did: a contradiction is a reason to trust none of the copies. A Wikidata item exists for the companyWikidata is open to any entity that can be sourced. Wikipedia is not, and writing your own article is the reliable way to have it deleted.

Third-party presence None of this is on your site, which is exactly why it carries the most weight. A model that has only ever read your own copy knows what you claim, not what you are.

You are named in listicles for your categoryThe "best X for Y" pages. These are where a model gets its list of who exists, and being absent from them is the most common reason a brand never comes up. Listed in the directories your buyers actually useCategory directories and marketplaces, not link farms. The test is whether a buyer would ever open it. Reviews on a platform models have readG2, Capterra, Trustpilot, Clutch, the app stores. Reviews are third-party sentences about you, in volume, in a predictable format. Mentioned on sites with their own audienceTrade press, podcasts with show notes, guest posts, conference pages. The oldest form of visibility there is, and still the one that moves most. Your people are quoted by name off-siteA named human tied to the company is a second entity pointing at the first. Someone else compares you to a named competitorA comparison page on a site that is not yours puts you in the same sentence as the brands you want to be considered alongside.

Consistency of the facts Contradictions do not average out. They make every version of the fact less certain.

What you sell is described the same way on and off your siteIf your directory listing says agency and your site says platform, both are now uncertain. Pricing, founding year, headcount and location agree everywhereCheck the profiles you set up years ago and forgot. They are still being read. Old names and rebrands are explained somewhere findableOne page saying "formerly X" resolves every stale mention on the web into the current entity instead of a second, ghostly one. No two of your own pages contradict each otherOld landing pages with superseded pricing are the usual culprit.

0

Weighted score

—

Now go and measure it

This score is a reasoned model of how a language model ends up with a stable idea of a brand — from repeated, consistent, corroborated mentions across sources. It is not a measurement of what any model currently knows about you, and no tool that has not asked the model can tell you that. The prompts above are the measurement; this list is the hypothesis.

Nothing you type here is uploaded, stored or sent anywhere. It all runs in your browser.

A model does not look you up. It has already read about you, or it has not

When an assistant answers "what is Acme Analytics", two quite different things may be happening. It may be recalling what it absorbed during training, in which case what matters is how often and how consistently you were described across the web that was read. Or it may be fetching pages right now, in which case what matters is what ranks today and whether the crawler can reach it. Most products do both. The practical consequence is the same either way: what other people say about you carries more than what you say about yourself, and consistency across sources carries more than any single source.

Why entity clarity is worth the boring work

Everything in the first group exists to make it obvious that the scattered mentions of you on the web are about one thing. A model working out who "Acme" is has to reconcile a LinkedIn page, six directory listings, two old press releases and a Reddit thread. sameAs, a stable name and an About page that says plainly what you are do not make you popular. They make you resolvable, and an entity that cannot be resolved cannot be recommended.

Why the third-party group is weighted highest

Because it is the only group a model cannot get from you. Your own site tells it what you claim. A listicle, a review platform and a comparison page written by someone else tell it what category you are in, who you compete with and whether anyone rates you — and those are precisely the questions an assistant is being asked when it recommends somebody. This is also the group you cannot buy your way out of quickly, which is why most "AI visibility" advice focuses on the other two.

Go and ask, and write down what you get

The four prompts this tool generates are a crude measurement, and a crude measurement beats a confident guess. Run them in a fresh chat with memory and personalisation off, in each assistant your buyers use, and keep the answers in a document with the date on it. What you are looking for is not a ranking but the specific wrong thing: the competitor you get confused with, the product you stopped selling in 2023, the "I do not have information about" that tells you nothing outside your own domain has ever described you. Each of those is a content brief.

Not this list, but the page in front of you?

This one is about the brand. For a single page, the AI Search Readiness Checker audits whether the page can be quoted; the Citation Readiness Checker goes sentence by sentence through the prose; and the GEO Content Checklist is what to keep open while you write the next one.

Questions people ask

Can any tool tell me what ChatGPT thinks of my brand?

Only by asking it and reading the answer, which is what the prompts here are for. Anything presenting a brand-visibility number without querying a model is estimating from inputs, exactly as this page does — the difference is whether it says so.

Do I need a Wikipedia page?

No, and you probably cannot have one: Wikipedia requires notability established by independent coverage, and writing about yourself there tends to end in deletion and a bad mark against the name. Wikidata is a different matter — it is open to any entity that can be sourced, it is structured, and it is widely ingested.

Does schema markup make a model understand my brand?

It helps the machines that build entity graphs resolve you to one thing, and those graphs feed a lot downstream. On its own it moves very little. Consider it hygiene: cheap, harmless, and no substitute for anyone else mentioning you.

How often should I redo this?

Quarterly is plenty for the checklist. Run the prompts more often if you are actively working on it — they are the only part that reflects reality rather than intent, and the answers drift as models are updated.

Is my brand name stored or sent anywhere?

No. Everything here runs in your browser, nothing is saved between visits, and the prompts are assembled locally from what you typed.

Related tools

Most of this list is other people talking about you

The entity half you can fix this week. The third-party half is earned, one placement at a time, and that is the work I am hired for.

Book a Call More free tools