A sample audit
This is the kind of technical read that sits behind our work: schema, structure, technical foundations and AI visibility, measured together, with evidence and a plan. The business below is an illustrative example, so you can see the depth before you start.
This is a sample. Riverside Dental Practice, its competitors and all figures are illustrative, used to show what a real report looks like. A real read is built from your website, your searches and your real answers from each engine. Nothing here describes a real clinic.
Riverside Dental Practice
The verdict, in one screen
Modern search has three jobs: rank on Google (SEO), be the answer when someone asks a question (AEO), and get named by generative AI (GEO). We score all three, plus the technical, entity and content foundations they stand on.
- You rank, but you are not the answer. SEO is your strongest pillar at 52, yet AEO (31) and GEO (24) lag badly. Being found is no longer the same as being chosen, and the gap to the category leader is almost entirely on the answer and AI side.
- Accuracy is the emergency, not visibility. Three engines state you do not offer implants. You do. That single error is quietly routing your most valuable enquiries to competitors.
- The causes are fixable and mostly technical. No structured data, thin answer content, inconsistent business facts across the web, and a weak entity footprint. None of this needs a rebuild.
Who the machines actually name
Across the money questions, this is how often each name appears in an AI answer, pooled over every engine and run. You are being out-named by two local competitors and a national directory.
Share of voice is the share of AI answers in which a business is named for the tracked questions. It is the single number that best predicts AI-driven enquiries, and the one we move first.
What each engine says about you
We asked the questions your patients ask, five times per engine, in their words. Green is the share of runs you were named. Amber means named with a factual error. Grey means not named at all.
| Patient question | ChatGPT | Gemini | Perplexity | Copilot | Google AI |
|---|---|---|---|---|---|
| Best dental clinic in Bristol | Not named | 2/5 | Not named | Not named | 1/5 |
| Where to get Invisalign in Bristol | 3/5 | 2/5 | 2/5 | Not named | 2/5 |
| Dental implants near me, Bristol | Wrong info | Wrong info | Not named | Wrong info | Not named |
| Good cosmetic dentist in Bristol | 1/5 | Not named | 1/5 | Not named | Not named |
| Emergency dentist Bristol Saturday | Not named | Not named | Not named | Not named | Wrong hours |
When you are named, you appear on average 3rd in the list. Clifton Smile Studio appears 1st or 2nd in 8 of 10 answers where both of you are mentioned.
A real answer, annotated
This is the kind of answer a patient sees when they ask about implants. It is the single most expensive moment in this report.
Where the AI gets its facts
Generative engines do not invent answers, they assemble them from sources they trust. For your category, these are the pages they lean on. You appear on far fewer of them than the businesses being named.
| Cited source type | Cited in category | You present | Read |
|---|---|---|---|
| Google Business Profile | Very high | Thin | Live but under-filled and missing services |
| Independent review platforms | High | Partial | Reviews exist but are not marked up or surfaced |
| Health and clinic directories | High | Mostly absent | Listed on 3 of 11 that engines cite for Bristol dentistry |
| Your own website | Medium | Weak | Cited rarely because pages are hard to extract cleanly |
| Local press and blogs | Medium | Absent | No earned mentions in the last 12 months |
Does the web know who you are?
Engines recommend entities they can resolve with confidence: a single, consistent organisation with a clear identity across the web. Yours is fragmented, which lowers trust before a single answer is written.
Are you answer-ready?
AEO is about being the clean, quotable answer to a real question. Right now there is very little on your site for an engine to lift, so it quotes someone else.
| Buyer question | Answered on site? | Extractable? | Owns snippet? |
|---|---|---|---|
| How much do implants cost? | No | No | Competitor |
| Does Invisalign hurt? | Buried | Weak | Competitor |
| Do you offer finance? | Yes | Weak | None |
| Are you taking NHS patients? | No | No | Competitor |
| Emergency dentist opening hours? | Outdated | Weak | None |
Extractability measures how cleanly an engine can lift a correct answer: clear question-shaped headings, a direct answer in the first sentence, and supporting structure. Most of your pages fail it.
Where you rank, and how healthy the site is
Most engines read the same open web Google indexes, so weak SEO quietly caps your AI ceiling. The fundamentals here are mid-table and fixable.
| Search | Monthly volume | Position | Trend |
|---|---|---|---|
| invisalign bristol | 1,300 | 8 | Up 2 |
| dental implants bristol | 880 | 14 | Down 3 |
| cosmetic dentist bristol | 590 | 17 | Flat |
| dentist bristol | 4,400 | 22 | Down 1 |
Can Google and AI actually read you?
AI crawlers do not run JavaScript, so anything that only appears after the page renders is invisible to them. This is where most sites quietly fail.
| Check | Status | Detail |
|---|---|---|
| AI crawler access | Pass | GPTBot, ClaudeBot, PerplexityBot and Google are all allowed in robots.txt |
| LocalBusiness / Dentist schema | Missing | No structured identity, services, hours or geo in the raw HTML |
| FAQ schema | Missing | No question-and-answer markup on any page |
| Review / AggregateRating schema | Missing | Real reviews exist but are invisible as a structured signal |
| Raw vs rendered HTML | At risk | Key service copy is injected client-side, so AI crawlers see an emptier page than users do |
| llms.txt | Missing | No curated summary file for language models. Low impact today, free to fix |
| Sitemap and canonicals | Pass | Present and correct |
Fix these first, whatever else you do
A prioritised 90 days
On the monthly plan we do this for you. On the audit alone, this is your map: sequenced by impact, tagged by discipline, highest return first.
What we would stake our name on
We do not sell guarantees, we make predictions and grade ourselves against them. For this example, on the plan above:
- The implant error is corrected across the three engines within 4 to 8 weeks of the fix reaching their next crawl.
- Share of voice for the money questions moves from 24% towards the low-to-mid 40s over the first quarter.
- The largest single gain comes not from ranking higher, but from being named correctly where you are currently mentioned and dismissed.
How we measured this, and the honest limits
Every question was run five times per engine to sample past the randomness these systems build in, and we report shares with a 95% confidence interval rather than a single lucky or unlucky result. Google rankings are measured for the location that matters, and technical checks are run against the raw HTML that AI crawlers actually see, not the rendered page. Share of voice pools every named mention across engines and runs for the tracked question set.
An audit is a point-in-time read. AI answers shift with phrasing, the user and the day, which is exactly why we sample repeatedly and re-measure monthly if you act on the findings. We will never guarantee a ranking or a place in any AI answer, because nobody can control what these systems say. What we guarantee is that the numbers are real, every error is evidenced, and when a month is flat we tell you plainly.
This was a sample. Want this on your own website?
This is the level of technical and AI-visibility detail we work with on the monthly plan: your actual website, your searches, your real answers across every engine. One flat price, £100 a month or £1,000 a year, no lock-in. Read more about the Search and AI Audit.
Get started