// AI Visibility Audit · prepared by Renownly
A real audit:
"an independent small-business accountant in a major UK city"
How visible is this firm when prospects ask AI assistants for an accountant in their area, and what would move the needle? Below is the audit we actually ran, with the firm anonymised. Run a contractor practice? See an IR35-specific example audit instead.
// headline score · proxy estimate
0/8
Proxy estimate across the engines' shared web index: the firm did not appear for 0 of 8 unbranded, high-intent queries. On a branded query (its own name) the firm was found and described accurately. Not a verbatim per-engine read.
// How to read this score
This sample is an early proxy read, not a verbatim per-engine read. We used live web search (the same web index ChatGPT, Perplexity, Gemini and Claude draw on) plus one labelled AI model as a proxy for "who an assistant would name". The full paid audit goes further: it reads all four engines directly (ChatGPT, Perplexity, Gemini and Claude), eight queries, five runs of each, and reports the consensus plus the variance. Results in this sample are US-weighted, so UK conditions are approximated, and this is a point-in-time snapshot. We don't control these tools and don't guarantee rankings.
// Section 1
What we tested
Eight unbranded, high-intent queries, plus one accuracy check.
We test the questions a prospect would actually ask: local, service-specific and high-intent, not vanity terms. The queries below are generalised versions of the real set (the firm's city and exact wording have been removed to anonymise it). For the same foundations check run across 93 firms, see our UK Accountancy AI-Visibility Report 2026.
- Q1"Best accountant for a small business in [the firm's city]"
- Q2"Best accountant for contractors in [the firm's city]"
- Q3"Best accountant for landlords / property in [the firm's city]"
- Q4"Accountant for self-employed / freelancers in [the firm's city]"
- Q5"Affordable accountant for a limited company in [the firm's city]"
- + 3 further unbranded service/location variants, and 1 branded accuracy check (the firm's own name)
Methodology note
- Engines (the full audit)
- ChatGPT, Perplexity, Gemini and Claude. The full paid audit reads all four directly. In this early sample none were queried verbatim; all four are approximated by proxy.
- Method
- Live web search per query (the shared index the assistants draw on) plus one labelled AI model as a proxy for "who gets named".
- Runs
- Single pass per query in this sample. The full paid audit runs each query five times per engine and reports the consensus and variance; this early sample does not, noted honestly below.
- Dated & weighted
- All results stamped 03 Jun 2026. AI answers are non-deterministic and change without notice; the web index is US-weighted, so UK conditions are approximated.
// Section 2
AI visibility results
The visibility matrix.
Every query and what the proxy returned. Named means the firm was recommended; Not named means a rival or a directory appeared instead. The runs column is honest: this early sample is a single run per query, and a full audit fills it with five runs and the variance per engine.
| Query | Proxy result (who gets named) | Firm named? | Runs / variance | Surfaced instead |
|---|---|---|---|---|
| Best accountant, small business | Directories + a recurring set of local rival firms |
Not named |
1 run · n/a (v1) |
Aggregator directories dominate |
| Best accountant, contractors | Contractor-specialist firms + directories |
Not named |
1 run · n/a (v1) |
A find-an-accountant directory + rivals |
| Best accountant, landlords / property | Property-tax specialists rank directly |
Not named |
1 run · n/a (v1) |
Rival firm pages; the firm absent |
| Accountant, self-employed / freelancers | A cluster of local rivals + a software directory |
Not named |
1 run · n/a (v1) |
Rival firms with query-shaped pages |
| Affordable accountant, limited company | Budget-positioned rivals + directories |
Not named |
1 run · n/a (v1) |
Local-firm pages and aggregators |
| + 3 further unbranded variants | Sector specialists and review directories |
Not named |
1 run each · n/a (v1) |
Same pattern: directories + rivals |
| Branded: the firm's own name | The firm's own site, testimonials, its directory + company-registry profiles |
Named · accurate |
1 run · facts correct |
Location, founder & focus all correct, with no hallucinated facts |
Named: firm recommended / described accurately
Not named: a rival or directory appeared instead
runs = repeat runs (v1 is a single pass; a full audit reports ≥3 + variance)
// Section 3
The gap
Strong on its own name, invisible on discovery.
The pattern is clean and consistent: when someone already knows the firm, AI finds it and describes it correctly. When a prospect asks an open question like "best accountant for contractors near me", the firm does not surface. Those slots go to aggregator directories and a recurring set of rival firms.
// where the enquiries go
Directories win the unbranded queries
The "best accountant in [city]" answers are dominated by aggregator directories (find-an-accountant sites, review platforms). The firm has a profile on one of them, but is thin or absent on the others that keep getting cited. That's how rivals keep appearing where the firm doesn't.
// what rivals have
Rivals have query-shaped pages
Competitors rank with pages literally titled for the question: "Accountants for Landlords", "Startup Accountants", "Accountants for Freelancers". This firm's content is more brand-led than query-led, so the AIs can't match it to the question being asked.
Branded visibility is the firm's strength
On the branded query the proxy named the firm and got every fact right: its city, its founder, its small-business focus. That's a genuine asset. The reputation is there, it just isn't reachable from open, unbranded questions yet.
The biggest single signal is fixable
The sharpest firm-specific finding is a location signal conflict (see the scorecard): the firm markets itself in one city, but its structured address sits in another, and more than one phone number appears across sources. Mixed signals make it harder for an AI to tie the firm confidently to its city.
// Section 4
Foundations scorecard
The signals AIs read, line by line.
The structured, public signals AI assistants use to decide who to recommend. Each is marked done, partial or missing, with a one-line note. Lines tagged tool come straight from our automated check; lines tagged analyst are web-search observations, kept separate so the two are never conflated.
80/100
- Homepage reachable & readable : donetoolReturns readable content (HTTP 200), with a title and meta description present. No bot-challenge or WAF interstitial blocking AI crawlers.
- Schema.org markup : done (with one gap)toolClean JSON-LD present: AccountingService, Organization, WebSite. A solid foundation that many firms lack.
- FAQ schema (FAQPage) : missingtoolNo FAQPage markup. An easy, cheap way to hand AI directly extractable Q&A, and the winning rivals have it.
- Location / NAP consistency (positioning vs reality) : conflicttool + analystThe sharpest finding. The firm markets itself in one city, but its structured (schema) address sits in a different city, and a directory lists a second phone number. A genuine cross-source inconsistency that dilutes its local relevance.
- NAP consistency within the website : donetoolThe visible contact details on the site match the firm's own structured data. The conflict is across sources, not within the site itself.
- AI-crawler access (robots.txt) : donetoolrobots.txt only blocks an admin path (universal best practice) and declares a sitemap. No AI crawler is meaningfully blocked.
- llms.txt : absenttoolNot present. Emerging and optional, a cheap edge rather than a problem, worth adding while it's still rare.
- Sitemap : donetoolA valid sitemap is present and declared in robots.txt. Crawlers can discover the firm's pages.
- Directory & registry presence : partialanalystActive on the company register and on one find-an-accountant directory, but thin or absent across the aggregators that dominate the unbranded answers. (See the directory checklist below.)
Directory presence: where the unbranded answers come from
AI assistants lean on these aggregators for "best accountant" answers. Marked present / partial / absent.
- National find-an-adviser directory: absent
- Reviews / vouching directory: present
- Professional-body "find an accountant": sparse
- Google Business Profile: present, thin
- Major review platform: absent / unclaimed
- Local / B2B business directories: thin
- Company register: active & correct
- Niche directory for the firm's specialism: absent
// Section 5
Your three fixes
Prioritised, firm-specific, ready to action.
Quick win, medium and strategic. Each in plain English with the "why", ready to hand to your web person. No jargon, no 25-page matrix.
// ~1 hour · in-house
Fix 1 · Quick win
Resolve the location signal.
Pick one genuine, client-facing city address and one canonical phone number, then make the website schema, the visible footer, the Google Business Profile and the main directories all say the same thing. Why: AI ranks firms it can confidently tie to a place. Right now the firm markets itself in one city while its structured address points to another, and that conflict dilutes its local relevance and undercuts the whole local positioning.
// ~half a day · web person
Fix 2 · Medium
Add FAQ schema and create query-shaped pages.
Add a FAQPage JSON-LD block answering the exact questions prospects ask ("How much does an accountant cost for a small business?", "Do you work with contractors / landlords?") to the relevant service pages. Why: the firm already has clean AccountingService schema (good) but no FAQ markup. FAQ schema gives AI directly extractable Q&A, and the winning rivals rank with pages literally matching these questions. Want to do it yourself? Our copy-paste FAQ schema guide and AccountingService schema guide have ready-to-fill blocks.
// ~2–3 weeks · content + outreach
Fix 3 · Strategic
Build presence in the directories AI actually cites.
The unbranded answers are dominated by a handful of aggregator directories. The firm is on one; claim and complete strong, review-rich profiles on the others (and the niche one for its specialism). Why: assistants lean on these aggregators for unbranded discovery queries, so being inside them is exactly how the rival firms keep surfacing where this firm currently doesn't.
What's your Foundations Score?
This firm scored 80/100. Check your own AI-Visibility Foundations Score free in under a minute: the same real number, scored on your live homepage. Or go straight to the flat £149 audit for the full picture across all four engines in 48 hours.
Based on a real Renownly audit, firm anonymised. The findings above are genuine, from an audit we actually ran on an independent UK accountancy firm; we've removed the firm's name and generalised its location and the exact query wording so it can't be identified. This was an early proxy audit: a point-in-time read via the engines' shared (US-weighted) web index plus one labelled AI model, not a verbatim per-engine read, and a single pass per query rather than the five runs across all four engines a full paid audit reports. Renownly reports the current state of third-party AI tools and gives best-practice recommendations. We do not control those tools and do not guarantee rankings, recommendations, traffic or revenue. AI answers are non-deterministic and change over time; all findings are dated at the time of testing (03 Jun 2026). Foundations findings are drawn from public data only. AI visibility is not the same as SEO ranking.