You do not need a six-month project to learn whether your firm shows up in AI search. You need ninety minutes, the questions your clients actually ask, and the discipline to write down what comes back. By this afternoon you will know where you stand — which is where every real improvement starts.
Why this is different from a Google ranking
There is no single position to check. AI answers are generated fresh each time, so visibility is something you observe, not something you look up. The only practical method is to run the questions a prospect would ask and record what each tool returns. That is the whole audit: structured observation, not a dashboard.
Set up in five minutes
Open a simple sheet with three columns: the question, the platform, and the result — absent, mentioned, cited or recommended. Add three more if you have time: competitors named, sources cited, anything inaccurate. You are collecting evidence, not writing a board report. Keep it plain.
Run the questions your clients are already asking
Most firms make the same first mistake: they test their own name. “What is Acme Consulting?” is reassuring but it is not where the risk sits. The risk is in the unbranded question, because prospects start with their problem, not with you. Run five.
The problem question — “how should a mid-sized UK professional-services firm improve its visibility in AI search?” Does AI name firms, frameworks or experts, and are you among them?
The category question — “which UK consultancies specialise in [your niche]?” The straight shortlist.
The comparison question — “compare [your firm] with [competitor] for [client goal].” Does AI understand your positioning, or repeat something out of date?
The proof question — “which firms have credible evidence or thought leadership on [your topic]?” This tests whether AI can find signals of authority.
The accuracy question — “what does [your firm] do, and would it suit [specific problem]?” Now check for missing services, old positioning and anything simply wrong.
Test more than one tool, more than once
ChatGPT, Claude, Gemini, Perplexity and Google’s AI overviews do not agree with each other. A firm can be visible in one and absent from another in the same week. Run each question in at least three, and run the ones that matter twice — a single answer is a sample, not the truth.
Read the result honestly
Sort each answer into four levels. Absent means you are not there at all — common, and not a verdict on your quality, only your footprint. Mentioned means you appear as a name in a list. Cited means a tool used your content or profile as a source. Recommended means AI actively makes the case for you. They are not equal, and the gap between mentioned and recommended is exactly where the commercial value sits.
Why the audit is the easy bit
You can run the audit alone this afternoon. You cannot build visibility alone over the following months — that takes consistent content, a cleaner footprint and a habit of checking what AI is learning about you. This is where the Expert Circle programme comes in. AI visibility is one of its themes, worked through with a monthly structure, two seats per firm so leader and marketer share the same context, and an accountability meeting that turns “we should fix this” into “we did.” Bring your audit results in and work the gap, rather than filing them away. You can see how it works at Expert Circle.
Three things to do this afte
- Run the five questions across three tools and record where you land on the absent-to-recommended scale.
- Note the pattern — which competitors keep appearing, and what sources AI leans on to recommend them.
- Choose your first three fixes from what you find: usually one page to restructure, one profile to correct, one piece of proof to publish.
A good audit does not end with “AI recommends us everywhere.” It ends with clarity — what matters, who is ahead, what is missing and what to fix first. That is more than enough to begin.
