Will an AI mention us?
It’s the question I get asked most now, and it usually arrives halfway through a conversation about something else entirely. Someone typed the company name into ChatGPT, or asked it to recommend a supplier in their category, and either got nothing back at all or got a competitor.
That’s a jarring thing to watch happen. It’s also partly fixable — and I want to be careful with that word, partly, because the rest of this page depends on it.
What the work consists of
Answer-first structure
A model lifts the passage that answers the question cleanly. If your page spends three paragraphs warming up, the answer gets lifted from whoever got to the point faster. The fix is structural and fairly unglamorous: say the thing, then explain it. Plainer sentences help too — Readability scorer gives you a rough read on where you are.
Questions written as questions
Real FAQ sections, using the wording people actually type rather than the wording your industry prefers, with answers that still make sense if they’re separated from the page around them. AI readiness checker checks for roughly what I check for.
Entities and consistency
The same name, the same description, the same facts about you across every page and anywhere else you appear. Models assemble a picture out of repetition; contradictions make them hedge, or drop you for a source that doesn’t contradict itself.
Schema that matches reality
Organization, Product, FAQ, Article — valid, and describing what’s genuinely on the page. Schema that overstates gets ignored at best, and it’s the sort of thing that ages into a problem.
Comparison tables
Probably the single most quotable format there is, because it arrives pre-structured. It also forces you to be honest about where you don’t win, which is uncomfortable and works considerably better than the alternative.
What you end up with
A page-by-page list of what to change and why, with rewritten openings where a rewrite is what’s needed rather than a note saying “rewrite this”. Schema you can paste in. A structure for the FAQ sections that isn’t just questions nobody asked.
And a baseline: the specific prompts a customer of yours would plausibly type, recorded now and checked again later, so you can watch what moves instead of taking my word for it.
The part every vendor should say out loud
So I’ll say it before you ask.
Nobody controls whether an AI cites you. Not me, not an agency with a bigger tool budget, not whoever is selling “AEO” or “GEO” packages this quarter. The models are closed, they change without notice, and the same prompt can return different sources on different days.
There’s no submission process. You cannot register a site for inclusion in an AI answer, there is no queue to join, and anyone charging you for that is charging you for nothing.
What can be done is real but indirect. Make the page easy to read, easy to parse, factually consistent and genuinely worth quoting, and you improve your odds. You don’t set them.
Anyone quoting you a hard percentage on any of this — including a figure for how many searches now end without a click — is quoting something unverifiable. I’d treat that as useful information about them.
Find out where you currently stand
Generative visibility is one of the four areas in the Diagnostic, so you don’t need to buy this on its own to learn whether it’s your problem or a distraction from one.