A Daviesis Guide for Accountancy Firms
Clients are starting to ask ChatGPT, Google AI Overviews and Perplexity to recommend an accountant before they ever search Google the old way. This guide explains how that works, and what your firm needs in place to be the one they’re given.
AI Search
GEO — Generative Engine Optimisation
AEO — Answer Engine Optimisation
For Accountancy Firms
01 – THE BASICS
For twenty years, “search” meant typing a few words into Google and clicking a blue link. That’s changing. A growing share of people now ask a question in plain English to ChatGPT, Google’s AI Overviews, Perplexity or Gemini and get a direct, written answer, often with two or three businesses named inside it.
That answer isn’t written by a person. It’s generated by a large language model that has read enormous amounts of web content and learned, in effect, who and what is worth citing on a given topic. When someone asks “who’s a good accountant for a small limited company in Manchester”, the model isn’t ranking ten blue links. It’s picking, at most, a small handful of names to mention by name.
A ranked list of links. The user clicks through, compares several sites themselves, and forms their own opinion before making contact.
A single written answer naming a handful of firms. The user’s opinion is already partly formed by what the AI says about each one before they visit any website at all.
The two skills behind getting found this way have their own names. GEO (Generative Engine Optimisation) is the practice of structuring your content and your wider online presence so AI models are more likely to find, understand and cite your firm. AEO (Answer Engine Optimisation) is the closely related discipline of writing content as direct, well-structured answers to the actual questions people ask, rather than as marketing copy. In practice the two overlap heavily and most firms need both.
02 – Why it matters here
Choosing an accountant is a research-heavy decision. People searching for “best accountant for a limited company near me” or “do I need an accountant for self assessment” are, by definition, doing exactly the kind of question-led research that AI assistants are built to answer. They want a clear explanation, not ten competing adverts.
That makes accountancy one of the professional services sectors most exposed to this shift, alongside law and financial advice. A firm with strong technical knowledge and a loyal client base can still be invisible at the exact moment a prospective client is forming their shortlist, simply because the AI doing the recommending has never encountered clear, citable content from that firm.
When ChatGPT or an AI Overview names a firm directly in answer to a question, it carries an implicit endorsement. The prospective client arrives already half persuaded, having been told by a source they trust that this firm knows what it's talking about. That's a different, and often shorter, route to a first enquiry than ranking eighth on a Google results page.
03 – Under the bonnet
AI models don’t have opinions of their own. They draw on patterns in the content they’ve read, and increasingly, on live web pages they retrieve at the moment of answering. A few factors consistently make a firm more likely to be the one mentioned.
Content that states the answer plainly near the top of the page, in language close to how the question was actually asked, is easier for a model to lift and cite confidently. A page that buries the answer under three paragraphs of firm history is much harder for a model to use.
A firm that only talks about itself, on its own website, looks self-promotional to both search engines and AI models. A firm that’s mentioned in local press, industry directories, partner sites, review platforms and third-party guides builds a pattern of independent corroboration. Models weigh this distributed presence heavily.
“We offer excellent service” tells a model nothing useful. “We handle Companies House filings for over 80 limited companies in Stockport and the surrounding boroughs” is a specific, checkable claim a model can confidently attribute to your firm when answering a specific question.
If an AI crawler can’t read your site cleanly, fast, with clear headings and without being blocked, it simply won’t be able to use your content at all, however good it is.
04 – Reality check
You can get a useful first read on your own visibility in about twenty minutes, using a fresh, signed-out browser session for each check so previous searches don’t skew the answer.
"Who is a good accountant for a small limited company in [your town]?"
"What should I look for in an accountant for a sole trader business?"
"Recommend a firm that handles self assessment tax returns in [your region]."
Ask each question in ChatGPT, Perplexity, Google (checking the AI Overview), and Gemini. Note whether your firm is named at all, whether it’s named alongside competitors, and whether the description given is accurate. Most firms doing this for the first time are surprised by how often they’re absent entirely, even when they rank reasonably on traditional Google search.
05 – What to do about it
Not keyword guesses, the actual questions: “do I need an accountant if I’m self-employed”, “how much does a limited company accountant cost”, “what’s the deadline for my first tax return”. Build content around these, one clear answer per page.
Open each page with a direct, two or three sentence answer to the question in the title. Then expand with detail, examples and context underneath. This structure is what makes a page easy for an AI model to cite accurately.
Replace generic claims with concrete detail: client numbers, sectors served, software you work in (Xero, QuickBooks, FreeAgent), turnover ranges you specialise in, and the geographic areas you actually cover.
Local business directories, accountancy body listings, client reviews on Google, contributed articles in trade press and local business publications all add the third-party corroboration AI models look for.
Fast load times, a clean site structure, no blocks on AI crawlers, and proper schema markup for your business and services. None of this is glamorous, but it’s the gatekeeper for everything else working.
Re-run the same set of prompts every month using a fresh browser session, and track which firms appear for each one. Visibility shifts as models update, so this needs to be a habit, not a one-off check.
06 – Pitfalls
AI visibility isn’t a website rebuild you do once. Models retrain and re-crawl continuously, so a firm’s standing can move up or down over months.
Stuffing pages with keywords reads as unnatural to both AI models and to human readers, and tends to bury rather than answer the actual question.
A beautifully built website that exists in isolation, with no independent mentions anywhere else, gives an AI model very little reason for confidence.
Standard keyword position tools weren’t built for this and miss most of what’s happening in AI answers. Checking manually, prompt by prompt, engine by engine, currently gives a more accurate picture for most firms.
The firms that show up when AI assistants are asked “who should I use” aren’t necessarily the biggest. They’re the ones whose content gives a clear, specific, well-corroborated answer to the question being asked.
See where your firm stands today
We’ll run your firm through the same prompt set used in this guide across ChatGPT, Perplexity, Gemini and Google AI Overviews, and send you a short summary of what’s found.
Professional services specialists