What ChatGPT says about the top 10 UK accountants
We put 200 client-style questions about small-business accountancy to ChatGPT, Claude, Gemini and Perplexity. These are the firms the assistants name — and the facts they fumble.
41%
of answers naming an independent firm contained at least one factual error
What we found
Accountancy is a hedging sector for the assistants: unless a firm's own content is strong, they retreat to naming national brands. The independent firms that do get named win an outsized share of recommendations — but the assistants routinely misstate specialisms, quoting Companies House filings and directory pages years out of date.
The top 10, as the assistants see them
Mention rate is the share of our customer-style questions in which at least one assistant named the business. Quotes are representative of the assistants' answers.
63%
mention rate
“Fenwick & Moore is often recommended to limited-company contractors for fixed-fee packages.”— ChatGPT
What the AI gets wrong: Claude names a founding partner who retired in 2023 as the current point of contact.
58%
mention rate
“A cloud-first firm praised for responsive Xero support and startup know-how.”— Gemini
No factual errors found this month — a clean sheet for this accountancy firm.
54%
mention rate
“Long-established practice with strong reviews from family-run businesses.”— Claude
What the AI gets wrong: ChatGPT says they offer statutory audit. They deregistered from audit in 2024.
49%
mention rate
“Recommended for hospitality clients; reviewers mention proactive VAT advice.”— Perplexity
What the AI gets wrong: Gemini quotes a “free initial consultation” offer that ended in 2025.
45%
mention rate
“A practical, plain-English firm popular with tradespeople and small landlords.”— ChatGPT
What the AI gets wrong: Perplexity lists an office in Durham that closed during the pandemic.
Wondering what they say about your accountancy firm?
We'll run the same analysis on your business and send you a sample report — free.
Analyse my business41%
mention rate
“Frequently named for R&D tax relief work with tech startups.”— Claude
No factual errors found this month — a clean sheet for this accountancy firm.
36%
mention rate
“Well regarded for payroll and CIS returns across South Wales.”— Gemini
What the AI gets wrong: ChatGPT attributes a 3.2-star rating from a review page that belongs to a similarly named firm in Bridgend.
33%
mention rate
“A growing firm; recent reviews praise fast year-end turnaround.”— Perplexity
No factual errors found this month — a clean sheet for this accountancy firm.
29%
mention rate
“Named for e-commerce specialisms, particularly Amazon sellers.”— ChatGPT
What the AI gets wrong: Claude says they're accepting new clients. Their own site says the waiting list is closed until autumn.
24%
mention rate
“A boutique practice recommended for owner-managed businesses in Kent.”— Gemini
What the AI gets wrong: Perplexity cites their 2022 fee schedule as current pricing.
The error pattern this month
The dominant error class is specialisms: assistants confidently attribute services (audit, R&D claims, payroll) that firms either never offered or quietly dropped. Second most common: partner names that left the firm years ago, lifted from archived team pages.
How we ran it
We asked ChatGPT, Claude, Gemini and Perplexity 200 natural-language questions a real customer would ask — the kind people now type instead of a Google search — then scored every answer for who was named, how warmly, and whether the facts were right. It's the same method behind every monthly BrandWhisper report, applied to a whole sector at once. Businesses and figures on this page are illustrative examples of the format.
Is AI lying about your business?
7of the businesses above didn't know what the assistants were telling their customers. Get your own report — sentiment, share of voice against named rivals, and every factual error, with a ranked fix-list — delivered monthly as a branded PDF.