The first UK-wide AI search visibility benchmark for accountancy: how findable and recommendable every firm is when someone asks ChatGPT, Perplexity, Claude or Google AI for an accountant. Every firm measured against the same 153 signals, on the same scale, in the same week.
✓ 153 signals per firm ✓ Same automated pipeline as 850,000+ site benchmark ✓ Replicable on any URL at searchscore.io
When a business owner asks an AI engine for an accountant, it can only recommend firms it can read. We measured how readable the UK profession is and gave every firm a GEO Score: a mark out of 100 for how ready its website is to be found, understood and cited by AI search. Four findings define the market.
The top is empty. One firm in 1,011 reaches AI-Ready. The first practices to clear 80 will own their categories before the rest know there's a race.
The Author Paradox is national. 767 of the 852 firms that display partner credentials to humans label none of them for machines.
Three firms have the complete stack. Five technical signals form a ladder from 46 to 77 average. Only three firms in Britain have climbed every rung.
The market can re-rank fast. 160 firms sit within five points of the Strong tier. Most are one fix away from outranking the firm next door.
Unlike the open web, where a third of sites are functionally absent from AI search, every firm in this benchmark registers. The problem isn't absence. It's indistinguishability.
Nearly half the profession sits in Emerging: cited occasionally, structurally improvable, and interchangeable with the firm next door as far as an AI engine is concerned. The median firm scores 56.3. The leader scores 80.4. That 24-point band is where every recommendation in this market is won or lost, and the work that closes it is technical, not reputational.
The boundaries are crowded. 160 firms sit within five points of the Strong tier, and another 138 within five points of 70. In a market this tightly packed, one technical fix doesn't nudge a ranking. It changes a tier, and the league table re-draws itself around whoever moves first.
Accountancy is a "your money or your life" category: AI engines demand evidence of qualified expertise before recommending a firm. The profession has the credentials. It just shows them to the wrong audience.
Named partners, bylines, qualifications on the page. The human-facing work is done. A visitor can see exactly who they'd be hiring.
Credentials labelled in the site's code so an AI engine can verify that "Jane Smith, FCA" is a chartered accountant, not a testimonial or a copywriter. Think of the difference between a CV written as flowing prose and a CV with proper labelled fields: humans can read both, machines can only rely on the second.
We first observed this pattern in Birmingham, then Manchester, then Greater London. At national scale it holds almost exactly: 767 of 852 firms with visible bylines, 90.0%, have no Author Person schema. To a reader, these firms look qualified. To the systems now answering "who's a good accountant near me", they're unsigned.
Take five technical signals: Organisation schema, Author Person schema, llms.txt, FAQ schema and explicit AI crawler permission. In plain terms: a machine-readable business card, labelled partner credentials, a one-page guide telling AI what your site is about, answers marked up so AI can lift them, and an open front door for AI crawlers. Stack them, and average GEO Score climbs a staircase. Each rung is roughly a tier.
Average GEO Score by number of the five signals deployed. These signals contribute to the score, so part of the climb is direct; but they are five checks of 153, and the staircase chiefly reflects that firms which do these five tend to have done everything else as well. The stack is a reliable proxy for overall AI-readiness.
These are the only three practices in the benchmark with all five signals in place, and it shows: one is the country's sole AI-Ready firm, and all three sit in the top 3% nationally. The complete stack is a few weeks of technical work, and 1,008 firms haven't done it.
Here is where 1,011 firms stand on each rung, plus the deepest gap of all.
Q&A schema: zero firms. Q&A schema is the markup that labels a question and its answer so AI can quote them directly. Of 636 firms with a recorded value, not one has it. An entire profession built on answering questions, and not a single practice has labelled its answers so a machine can lift them. This is the most open competitive space in the benchmark.
More than half of UK practices, 543 firms, provide no Organisation schema at all. They are asking AI engines to guess what they do, where they are and who they serve. Engines don't guess. They cite the firm down the road that told them.
A signal the city benchmarks never surfaced. AI engines check when content was published and last updated before deciding whether to trust it. In a profession governed by tax years, most firms don't say.
Published and last-updated dates on articles and guidance. Among firms scoring 70+, dating content is near-universal: 85% carry both, and 90% show content-freshness signals.
Across the whole market only 381 of 1,011 firms, 37.7%, publish dates at all. Nearly two-thirds of UK accountancy content is undated.
At a 51-point gap between leaders and the field, dating behaviour separates the visible from the invisible as sharply as schema does. The logic is unforgiving: thresholds, allowances and rates change every April. A guide to dividend tax with no date on it is a guide an AI engine cannot safely cite, because it cannot tell which tax year it describes. Undated expertise reads as stale expertise, however current it actually is.
The most AI-visible accountancy firms in Britain are not where the money is. They're where the technical work has been done. Rotherham and Wigan outrank most of the City.
| # | Firm | City | GEO Score |
|---|---|---|---|
| 1 | Acumen Accountants & Tax AdvisersAI-Ready acumenagc.com Full stack: Organisation, Person and Author Person schema, llms.txt, FAQ schema, content hub, LinkedIn | London | 80.4 |
| 2 | Golden Tree Accounting goldentreeconsulting.co.uk Organisation, Person and Author Person schema, llms.txt, FAQ schema, answer-first content, content hub | London | 78.9 |
| 3 | Number Crunchers number-crunchers.co.uk Organisation and Person schema, llms.txt, content hub | Rotherham | 78.4 |
| 4 | Elver E-Commerce Accountants elverecommerceaccountants.co.uk Organisation, Person and Author Person schema, llms.txt, FAQ schema, content hub | Wigan | 78.4 |
| 5 | AA Accountants aaaccountants.uk Organisation, Person and Author Person schema, LinkedIn presence | Bristol | 78.2 |
| 6 | Elman Wall elmanwall.co.uk Organisation, Person and Author Person schema, llms.txt, content hub | London | 77.4 |
| 7 | Clarke Bell clarkebell.com Organisation, Person and Author Person schema, FAQ schema, content hub | Manchester | 77.4 |
| 8 | Thames Williams thameswilliams.com Organisation and Person schema, llms.txt, FAQ schema, content hub | Birmingham | 77.1 |
| 9 | Sibbalds sibbald.co.uk Organisation, Person and Author Person schema, content hub | Derby | 77.0 |
| 10 | Clear Cut Accounting clearcutaccounting.co.uk Organisation, Person and Author Person schema, llms.txt, FAQ schema, content hub | Manchester | 76.9 |
Top 10 named; all firms notified prior to publication. Firms outside the top 10 are anonymised in the underlying dataset and not identified publicly. Scores reflect each site at time of audit.
London holds three of the top ten. The single biggest accountancy market in Europe, out-produced seven-to-three by Rotherham, Wigan, Derby, Bristol, Birmingham and Manchester combined. Brand, postcode and headcount don't move a GEO Score. Schema, structure and crawler access do, and those are available to a two-partner practice in Rotherham on the same terms as a fifty-partner firm in EC2.
Why this decides who gets the next client. Buyers increasingly start with an AI assistant rather than a search results page, and the assistant names two or three firms it can read and verify. 73% of B2B buyers now use AI during their research stage before making contact. A referral gets checked with AI before the call is made; if the engine can't describe the firm, the referral converts less reliably than it used to. None of this loss appears in analytics. There is no dashboard for the calls that didn't come.
Forrester B2B Buying Survey, 2025 · cross-industry, UK + US respondents (n=2,341)
Every one of these is a technical website change, not a reputation project. A well-resourced firm can close all five in four to six weeks.
543 firms have none. It's the machine-readable answer to "who are you, what do you do, where are you". Hours of work, and the biggest single win in the dataset.
The 90% gap. Attach name, role and qualification to every partner bio so engines can verify the expertise you already display. Only 86 firms in Britain do this. Joining them puts you in the top decile on the profession's weakest signal.
30 firms have FAQ schema. Zero have Q&A schema. You answer client questions all day; labelling those answers makes them liftable into AI responses. This is the emptiest competitive space in the benchmark.
Crawlers are the programs AI engines send to read the web, and robots.txt is the file that tells them what they may read. Only 25 firms explicitly welcome them. Add allow rules for GPTBot, ClaudeBot and PerplexityBot: a crawler that can't read you can't cite you.
An llms.txt is a plain-text file at yourdomain.com/llms.txt that tells AI engines what your firm does and which pages matter. One in four firms has one. Answer-first means the first sentence of a service page answers the question directly; one in five firms does it. Both take an afternoon, and both directly feed what engines extract and cite.
UK accountancy practices is AI-Ready. The top of this market is effectively unclaimed.
of firms that show credentials to humans show nothing to machines. The Author Paradox, at national scale.
firms have Q&A schema deployed, in a profession that exists to answer questions.
in Britain have the complete five-signal stack. The ladder from 46 to 77 average is open to anyone, and almost no one has climbed it.
firms sit within five points of the Strong tier. One technical fix re-draws the league table.
of UK accountancy firms publish undated content, in a profession governed by the tax year.
top firms are outside London: Rotherham, Wigan, Bristol, Derby, Birmingham, Manchester. AI visibility has no postcode.
average GEO Score across 1,011 firms. The median scores 56.3; the leader scores 80.4.
UK-based independent accountancy practices. The benchmark excludes software vendors, fintechs, payroll-only services, international networks and Top-50 national firms, and counts one domain per firm. Every exclusion is logged with a stated reason in the underlying dataset.
Each firm was scored across 153 GEO signals covering AI citability, brand authority, content quality and E-E-A-T, technical foundations, structured data, platform optimisation, topical authority and AI platform readiness, combined into a single GEO Score from 0 to 100. Tiers run in 20-point bands: Invisible (0–19.9), Low Visibility (20–39.9), Emerging (40–59.9), Strong (60–79.9), AI-Ready (80+). One previously benchmarked firm, Intertax of Manchester, could not be rescanned because its website was returning errors at audit time, and is not in this edition.
Scores in this edition are not directly comparable with the May 2026 city benchmarks (Birmingham, Manchester, Greater London). Since those editions the audit pipeline gained full JavaScript rendering, correcting systematic under-scoring of script-heavy sites, and runs an expanded signal set. Like-for-like rescans of 29 firms named in the city benchmarks moved +4.0 points on average, with 28 of 29 rising. Per-firm scores in this edition supersede previously published figures.
This sector edition sits alongside the Q2 2026 global SAVI Report (866,000+ sites) and the quarterly index at searchscore.io/savi. This is the first accountancy edition, covering H1 2026. Sector editions run half-yearly alongside the quarterly SAVI index; the H2 2026 edition will be the first to show movement. There is no trend line yet, by definition: this edition draws it. Every figure here, the 54.8 average, the single AI-Ready firm, the three complete stacks, the 90% Author Paradox, is the line against which every future edition of this benchmark will be measured. Movement, in either direction, will be public.
Anyone can run the same audit on any URL, free, at searchscore.io. The top 10 firms were notified prior to publication and are named; the remaining firms are anonymised to protect them from unsolicited outreach. Full signal documentation is at searchscore.io/methodology.
Run the same 153-signal audit we ran on every firm in this report. Free, 60 seconds, no email. You'll get your GEO Score, SEO Score and CRO Score, plus the specific list of what's broken.
Check your firm's score → Founding programme for accountantsThe AI Visibility Programme ships these fixes for you. Limited to five UK accountancy firms at founding rates.
Already scoring well? Tracker tells you when the league table moves, weekly, across six AI engines.