What is an AI visibility audit — and what should one contain?
A structured test of how AI assistants treat your business today: the questions your buyers ask, run live across ChatGPT, Perplexity, Gemini and Google's AI surfaces, recording who gets named, what gets said about you, which sources are cited, and what is technically blocking you. The output is a gap map and a prioritised plan — not a score. We deliver ours in writing within 24 hours of the first call, and it costs you nothing.
What is an AI visibility audit?
What gets tested
The raw material is prompts — the category questions your customers actually ask, never just your brand name. Each is run several times per platform, because AI answers vary between runs and a single screenshot proves nothing. The audit logs who is named, in what order, with what description, and which web sources each assistant cites for the answer.
Your own presence gets the same treatment: what does each assistant believe your business does, for whom, at what price point, in what territory — and is any of it wrong? Wrong beliefs matter more than absence, because assistants repeat specific errors confidently.
The six gaps a real audit sorts findings into
Findings only become a plan when they are classified. We use six gap types, each with a different fix.
Visibility gap: competitors named where you are absent
Narrative gap: you are named, but described wrongly or thinly
Topic gap: question areas you should own where nothing of yours is retrievable
Format gap: you lack the content types assistants cite — comparisons, guides, video
Mention gap: the lists and directories being cited name competitors, not you
Demand gap: people asking assistants about you, landing on pages that waste the interest
The technical half
Alongside the prompt testing: crawler access verified with real bot user agents, a no-JavaScript fetch of your key pages to see what assistants actually receive, response speed, heading structure, structured data, and whether your analytics can even see AI referrals. This half is mechanical, fast, and frequently where the biggest single win hides — a site that AI crawlers cannot read makes every other effort pointless.
What the plan prioritises
Every finding lands in one of three buckets: fix (update something that exists), build (create what is missing), or influence (earn a mention somewhere you do not control). Quick wins get flagged first — an existing page one meaningful update away from citable, or a single influential list your competitors sit on and you do not. The plan is yours either way; acting on it with us is a separate decision.
How is this different from an SEO audit?
An SEO audit asks how you rank. An AI visibility audit asks what assistants say — different platforms, different levers, different evidence. It tests live answers across multiple AI platforms and maps mentions across the wider web, which no ranking report shows you.
Why do you run each prompt multiple times?
Because AI answers are probabilistic — the same question can cite different sources hours apart. Single-run results are noise. Multiple runs per platform turn noise into a share-of-voice number you can actually track month over month.
What do you need from me to run it?
Twenty minutes: who your best customers are, what they ask when they are looking for you, and who you lose deals to. The rest is our work. You do not need to provide site access for the audit itself — everything we test is what the public web and public assistants see.
Is the 24-hour report really useful, or a sales brochure?
It is the actual audit: named competitors, real prompt transcripts, the gap map, and the prioritised list. We put our findings in writing because that is how we would want to be sold to. If the honest finding is that your AI visibility is fine, the report says so.
Last reviewed 23 August 2026