AI Visibility

Nobody is finding you. They're only looking you up.

There are two ways to show up in an AI answer, and only one of them is worth anything. I ask ChatGPT, Gemini and Claude the questions your buyers actually ask — thirty to forty of them, three runs each — and hand you a scored record of what came back, which competitors got named instead of you, and a ranked list of what to fix first.

Start with the free snapshot

The snapshot is one page and takes 48 hours. The full report is what this page is about.

The choosing already happened somewhere you can't see

Your buyer used to open Google, scan ten blue links, and decide for themselves which one to trust. You could win that with content, and you could measure it with a rank.

Now they open ChatGPT and get one answer. Three vendors named, a paragraph of reasoning, and no visible list of who didn't make it.

Most companies have never checked what that answer says. Not because they don't care — because there's no rank to look up and no dashboard that reports it. So the question goes unasked, and the answer keeps getting given.

Here's the uncomfortable part: if you've never run it, you don't know whether the engine is describing you, hedging about you, or leaving you out of your own category entirely. And your competitor's sales team is not going to tell you.

The two kinds of appearance

Not every mention counts. This is the distinction the whole report is built on.

Brand-direct

Someone types your company name. The engine describes you. Nice, but they already knew you existed. This appearance is worth almost nothing — it's a lookup, not a discovery.

Category discovery

Someone types the problem, the use case, or "best vendors for," with no brand name anywhere. The engine picks who to name. This is the only appearance that grows anything.

Most companies that "show up in ChatGPT" are showing up in the first column and calling it visibility.

This month I ran an unsolicited audit on a certified manufacturer with decades in its category — not a client, just a company in a category I wanted to measure. It was named in 16 of 30 answers. Twelve of those sixteen came from prompts that already contained its name. Found on merit: four times out of eighteen. On category-discovery prompts it was named a third of the time.

That gap is invisible until someone runs the prompts and writes down what came back.

22% ON MERIT

Named in 53% of answers — and found on merit in 22%. The outer ring is what a dashboard would have shown. The lime arc is the number the pipeline actually feels.

Read the full field report →

What I audit

What you get

  1. 01

    The scorecard

    Appearance rate, share of answer, and citation share, broken out by engine and by intent category.

  2. 02

    The verbatim record

    Every answer, dated, with the prompt shown, so any finding can be checked.

  3. 03

    Competitive share of answer

    Who owns which questions, and which ones nobody owns yet.

  4. 04

    Gap analysis

    The questions where no vendor is winning. Usually the fastest wins on the list.

  5. 05

    A ranked remediation plan

    Impact over effort, with owners and a 90-day sequence.

  6. 06

    A re-measurement plan

    Exactly what to re-run in 90 days and what better looks like.

Every prompt, engine, date and run count is written into the method section. You could run this yourself — the method is in the report, not in my head. Hand it to your team or to another consultant and they can re-run it and check my work.

Reproducibility is the credibility.

The receipt

I grew a regional broadcaster's site from 500 to 50,000 monthly visitors in six months. That was the search era, and the mechanics were the same then as they are now: figure out the question, structure the answer so a machine can extract it, get cited somewhere that isn't your own domain.

What changed is that the machine stopped showing its work.

What it isn't

It isn't AI SEO. Nobody controls what a model says. I don't sell placements, rankings or guaranteed appearances, and you should walk away from anyone who does. What I sell is measurement, structure, and better odds — stated plainly.

It isn't a tool subscription. No dashboard, no scraper, no monthly number that drifts. Prompts run in fresh sessions, by hand, with the answers recorded verbatim. That's slower and it's the reason the findings hold up.

It isn't a manufactured crisis. If you're already visible, the report says so and names the one gap. A clean finding is a real finding, and it gets reported as one.

Timeline and investment

Flat project fee, agreed before anything starts. No hourly billing. Exact pricing comes with your proposal — it's scoped to the number of competitors and the size of the prompt set.

The free snapshot is the place to start. It's a tenth of the scope. It usually decides the question for you.

FAQ

{faq(av_faq)}

Find out what the machines say about you.

One page. Forty-eight hours. The prompts your buyers actually ask, and exactly what came back — including who got named instead of you.

  • 01One page, inside 48 hours. Not a note saying it's in progress.
  • 02Three engines, the prompts your buyers actually type.
  • 03No call required, and no pitch inside the report.
  • 04Free. There is no version of it that costs money later.

I run a limited number each month, so there may be a short wait.

Tell me where to look.

Takes about a minute. The more precise the last two answers, the sharper the report.

The form isn't loading.
Email me at [email protected] with your company name, website, how a buyer would describe what you sell, and two competitors you lose to. Same 48 hours.