Ask an AI assistant to recommend an accountant in a specific town and it will usually name a couple of firms without hesitating. Almost nobody at the practices it did not mention has ever asked the question. That is the real problem, and it is not a technical one: most firms have simply never checked what an assistant says about them, because it has never occurred to anyone to look.
This guide covers what AI visibility actually is, why so few practices have checked theirs, what the assistants appear to be reading when they answer, and what to do about it. It was written after running our own checker across a spread of accountancy websites, so the patterns below come out of real results rather than theory.
What is AI visibility, and why does it matter for an accountancy practice?
AI visibility is whether an AI assistant names your firm when somebody asks it for a recommendation. Someone looking for an accountant in Stockport increasingly opens ChatGPT rather than Google, types something like "who are the good accountants in Stockport", and works with whatever comes back. If your practice is not in that answer, you were never in the running.
This matters more for accountancy than for most trades. Choosing an accountant is a considered decision, people rarely switch, and a single new client can be worth years of work. Being absent from the shortlist at the moment somebody is actively looking is expensive in a way that is very hard to see afterwards, because the enquiry that never arrived leaves no trace.
It also compounds. Assistants tend to name the same firms repeatedly, so a practice that is already described clearly keeps getting mentioned while a practice that is not stays invisible, regardless of which is actually better at the work.
Why have most practices never checked?
A practice that has slipped down Google's rankings usually finds out fairly quickly. Enquiries drop, somebody checks the analytics, and there is a familiar report to look at. AI visibility does not work that way yet.
There is no dashboard, no weekly email, and no ranking position to watch. Checking means deliberately opening a chatbot, phrasing a question the way a stranger would, and reading the answer properly. Nobody has that on a to-do list. So a firm can be entirely absent from every assistant's answers for a year with nobody inside the practice any wiser.
The second reason is that it sounds like somebody else's problem. AI gets discussed at practice level as a bookkeeping tool or a drafting tool, something that changes how the work gets done. The idea that it also quietly decides who gets recommended has not really landed yet.
How do AI assistants decide which firms to name?
Mostly they are not searching the live web at the moment they answer. They are recalling firms that were described clearly and specifically somewhere they could read: the practice's own website, a directory listing, a review, local press coverage.
That has a direct consequence. The assistant can only repeat back a description that exists in words somewhere. A firm described in the same generic language as every other local practice gives it nothing distinctive to recall. A firm that states plainly who it works with, which software it runs, and which town it sits in has given the assistant something specific to name.
This is why the practices that do well here are not usually the ones doing anything clever. They are the ones whose website happens to say clearly what they do and for whom, often because somebody wrote it properly years ago.
What we found running the checker
We built a small tool that runs the question for you. It asks an AI assistant to recommend a firm in a given town, then reports whether your practice appears in the answer. You can check your own practice with the AI visibility checker in about a minute.
Run across a spread of accountancy websites, the pattern was consistent. Firms with a plain, specific description of their clients and their software were named far more often than firms with an identical service list wrapped in stock photography. Several practices with genuinely strong reputations locally did not appear at all, because their websites said almost nothing an assistant could use.
The most common result by a distance was no mention whatsoever. Not a poor description, not an outdated one, simply absence. For most practices running it, that is the first time anyone there has seen where the firm stands.
What makes a practice visible to AI assistants
Say who you actually work with
"We work with sole traders and small limited companies across Greater Manchester" is a sentence an assistant can quote directly. "We serve businesses of all sizes" is not, because every other firm in the county could say exactly the same thing and many of them do.
Naming the client you want is the single highest-value change on most accountancy websites. It feels like turning work away. In practice it is what makes you findable, because specific is the only thing that is memorable.
Name your software in plain text
A practice that works with Xero, QuickBooks or Dext should say so in words on the page. Logos in an image do not count. Nothing reading that page can see what a logo says, so a site that shows its software purely as graphics has effectively kept it secret from every assistant.
Make the site readable at all
A site that loads slowly, or hides most of its text behind a slider, a video or an animation, gives an assistant very little to work with even when the practice behind it is excellent. The same is true of text baked into images, which is more common on accountancy websites than you would expect.
Answer the obvious questions in writing
Every practice gets asked the same handful of things on a first call: what software do you use, how quickly do you reply, what happens in the first month, do you handle payroll. Putting those answers on the page in plain language serves two purposes. It stops the questions arriving by email, and it gives assistants exactly the sort of specific, quotable material they draw on.
Does this replace Google SEO?
No, and treating it as a separate discipline is a mistake. Almost everything that makes a practice visible to an AI assistant is the same work that has always made a website rank: clear specific pages, fast loading, plain text rather than pictures of text, and content that answers a real question.
What has changed is the cost of skipping that work, and how visible the gap now is. A vague website used to just underperform quietly. Now there is a simple way to check, and the result comes back as a straight yes or no rather than a ranking that can be explained away.
So the honest framing is that this is a website development problem wearing new clothes. The fix is not an AI trick. It is writing pages that say something definite, and building them so a machine can actually read them.
How to check your own practice today
Read your homepage as a stranger would, ignoring everything you already know about the firm. Ask whether the words alone tell you which town you are in, what kind of client this practice wants, and what software it runs. Most practices find at least one of those is missing entirely.
Then run the checker on your own firm and see what comes back. Whatever the result, it is information you did not have before, and it takes about a minute.
Why choose Dawn Studio
Dawn Studio is run by a qualified accountant who has worked in practice, so the advice above comes out of the same questions we asked about our own work before building anything. We hand-build every site in modern code, with no WordPress and no page builders, which is how sites load quickly and stay readable to both people and machines.
We do not sell AI visibility as a separate product, because it is not one. It is a consequence of a clear, fast, well-written website, which is the thing we actually build.