What AI says about your company when nobody’s asking you

Someone asked an LLM about your category this morning. Possibly several people did.

It gave them an answer. Named two or three companies, described what each one is good for, maybe ranked them. You were in that answer or you were not, and either way nobody told you.

This is now a normal part of how software gets bought, through AI brand visibility. A buyer forms a shortlist inside a conversation with a model before they visit a single website. By the time they reach you, the comparison already happened.

Most founders have never seen what that answer says about them.

Why there is no dashboard for this

You can see your search rankings. You can see ad impressions, site traffic, which pages people read and where they left. Every other buying surface reports on itself.

This one does not. There is no console, no impressions figure, no notification when a model stops mentioning you. The information is not hidden so much as never collected.

Which means the only way to know is to go and look. Almost nobody does, and that is the actual opportunity here.

What the model (or any Answer Engine) is actually working from

The instinct is to assume it reads your website and repeats it. It does not work that way.

The answer is assembled from whatever the model has absorbed about your category. Review sites. Comparison pages written by other people. Forum threads where someone asked for recommendations. Competitor content that mentions you in passing. Analyst posts. Your own site is one input among many, and often not the loudest.

That is uncomfortable, because it means a large part of what gets said about you was written by other people. But it is also the useful part. You cannot edit a model. You can change what exists for it to read.

The three ways this goes wrong

Founders tend to worry about one failure mode. There are three, and the third is the one that costs the most.

• You are not mentioned. The obvious one. Your category gets an answer, three competitors are named, you are absent. Usually this means the model has very little about you outside your own domain.
• You are mentioned inaccurately. Wrong pricing, a feature you removed two years ago, a market you do not serve. Old content ages into the answer and stays there.
• You are mentioned accurately, as somebody else’s cheaper alternative. This is the quiet one. Everything stated is true, and the framing places you permanently below a competitor. Buyers arrive already thinking of you as the budget option, and you never find out why.

How to check your AI brand visibility

This takes an afternoon and costs nothing.

Write ten prompts a buyer in your category would actually type. Not your product name. Nobody who has not heard of you searches for you. Write the prompts that describe the problem.

Run each one across the models your buyers use. Record which companies get named, in what order, and how each is described. Do it in a fresh session each time so previous questions do not colour the answer.

Then repeat monthly. A single reading tells you where you stand. A series tells you whether anything you did worked, which is the part that matters.

What moves your brand on answer engines

Plain category language. If your homepage says you are a revenue intelligence platform, and buyers say they want software that tells them which deals are stalling, the model has nothing to match. Say the second thing.

Comparison content that names competitors honestly. Pages that say who you are better than and who you are not get cited, because they contain the specific claim a model needs. Pages that avoid naming anyone do not.

Being mentioned somewhere other than your own site. One credible third party mention does more than ten blog posts on your domain, because the model already trusts the source.

What does not move your brand on any LLMs

Keyword density. Publishing more of the same thing. Schema on its own. All of that optimises for a system that ranks pages, and this is a system that retrieves claims. Different problem, different work.

Common questions