← AI visibility audit Download as PDF ↓
Exitus Agency
Sample · fictitious business
Full AI visibility report

Clínica Dental Montsol

What ChatGPT, Gemini and Perplexity answer today when someone looks for a business like yours — and what has to change for you to appear.

30out of 100
You appear in 4 of 90 answers
This is a sample. The business, the website, the competitors and every result are made up. It exists to show exactly what format, what sections and what data the report you receive contains, so nothing comes as a surprise. Your report will have this same structure, with your real data.
Business
Clínica Dental Montsol
Website analysed
clinicamontsol.example
Area
Gràcia, Barcelona
Date of analysis
22 September 2026
Engines queried
ChatGPT, Gemini and Perplexity
Real questions
30 questions × 3 engines = 90 answers
Competitors compared
2, from the same area
Delivery
PDF + files ready to install
In one sentence

They cannot cite you today, and it is not for lack of happy patients

You have 184 reviews and a 4.7. That is the hard part and you already have it. The problem is that your robots.txt returns a 403 to the AI crawlers: ChatGPT, Claude and Perplexity cannot read the site. And even if they could, the home page takes 180 words to say what you are and where, there is no machine-readable business record, and not one practitioner appears with a registration number — which in healthcare counts double. Out of 90 answers, you appear in 4, and in none of them with your phone number.

The good news: what weighs most is also the cheapest. Three of the five fixes in the first wave take a day.

The score, block by block

30 out of 100

Nine blocks, each carrying the weight it has in the final score. The weighting is not ours: it comes from what each engine looks at when it decides who to cite.

#BlockWeight Score
1Citability / BLUF15%22
2Structured data15%15
3Freshness12%30
4Citable content12%18
5E-E-A-T / entities12%35
6Off-site reputation12%58
7Technical / indexability10%40
8Architecture6%45
9Answer potential6%20
Overall score100% 30
The 30 questions

What they answer today, word for word

Thirty questions someone would really type, put to all three engines on 22 September 2026. · you don't appear · ~ you get a passing mention, no details · ✓ you appear with a phone number or address. The last column says who gets cited in your place.

#QuestionChatGPT GeminiPerplex.Who gets cited
1Which dentist do you recommend in Gràcia, Barcelona?···Dental Montseny, Doctoralia
2Dental clinic near Plaça del Sol··~A Google Maps listing
3Where can I get dental implants in Barcelona?···Dental Montseny, 2 chains
4Invisible braces in Barcelona — which clinics are good?···3 national chains
5Emergency dentist on a Sunday in Barcelona···Public health service
6How much does a dental cleaning cost in Barcelona?···Forum + 2 chains
7Dental clinic with payment plans in Gràcia···Odontolab Gràcia
8Children's dentist in Gràcia·~·Dental Montseny
9Which dental clinic in Barcelona has the best reviews?···Doctoralia (aggregated)
10English-speaking dentist in Barcelona···2 clinics in the Eixample
11How long does a dental implant take from start to finish?···A chain's blog
12Does a root canal hurt?···Dental association website
13Price of a zirconia crown in Barcelona···Forum + 1 chain
14What is periodontics and when do you need it?···Wikipedia + chain
15Teeth whitening in Gràcia — where?···Odontolab Gràcia
16Dentist in Gràcia that takes Adeslas insurance···The insurer's own finder
17How often should you go to the dentist?···Health portals
18Dental clinic with parking near Gràcia···Google Maps
19What do I do if a filling falls out on a bank holiday?···Health portal
20Dentist for elderly patients with reduced mobility in Gràcia···Nobody specific
21Are implants covered by the public health system?···Official website
22Free first appointment dentist Barcelona···3 chains
23Braces for adults — are they worth it?···Chain blog
24What's the difference between veneers and crowns?···Chain blog
25Dental clinic open in August in Barcelona···Nobody specific
26Children's dental check-up in the Gràcia neighbourhood~··Dental Montseny
27Dentist with same-day appointments in Barcelona···Doctoralia
28What should I ask at a first dental appointment?···Health portals
29Dental clinic Gràcia reviews·~·Google Maps, Doctoralia
30No-obligation dental implant quote Barcelona···2 chains

Result: 4 out of 90. And all 4 times in passing: it names you with no phone number, no address and no opening hours, because it cannot find them anywhere it can read.

The two competitors

Why they get cited

Two clinics from the same area, measured against the same nine blocks on the same day. They are not better dentists: they are more legible.

BlockYouDental MontsenyOdontolab Gràcia
Citability / BLUF226540
Structured data157025
Freshness305548
Citable content186035
E-E-A-T / entities356240
Off-site reputation587066
Technical / indexability405852
Architecture455550
Answer potential205030
Overall score306242

Dental Montseny is 32 points ahead of you, and half the gap sits in two blocks: they have a machine-readable business record (70 against your 15) and a first sentence that answers (65 against 22). Both are in the first wave of the plan.

Block by block

What was checked, and with what evidence

Each block says what was checked and what your site answered, verbatim, so you don't have to take our word for it. If your developer wants the exact commands to repeat it, we'll send them over.

1 Citability / BLUF weight 15% · 22/100

The home page takes 180 words to say what you do and where

What was checked

Read the first 40 lines of the home page, as an engine receives them

What your website answered
<h1>Welcome</h1>
<p>For over twenty years we have cared for our
patients' smiles with the utmost…</p>

What it means. An AI engine reads the first few sentences and decides whether it can answer with them. «Welcome» answers no question at all. What's missing is the sentence saying what you are, where you are and who for, on the first line.

2 Structured data weight 15% · 15/100

There is no business record a machine can read

What was checked

Count the schema.org business records on the home page

What your website answered
0

What it means. Without schema.org of type Dentist, the AI has no address, opening hours, phone or services in a format it understands. It has to guess from the text, and usually it doesn't: it takes a directory listing instead.

3 Freshness weight 12% · 30/100

The newest content is 14 months old

What was checked

Look at the three most recent last-modified dates in the sitemap

What your website answered
<lastmod>2025-07-18</lastmod>
<lastmod>2025-07-18</lastmod>
<lastmod>2025-07-18</lastmod>

What it means. All three dates are the same: the sitemap was generated once and never touched. Engines use freshness to decide who to cite for questions about price, opening hours or availability.

4 Citable content weight 12% · 18/100

No page answers a specific question

What was checked

Manual review of the 9 published pages

What your website answered
7 service pages (sales copy), 1 contact, 1 team page.
Zero questions answered, zero prices, zero timescales.

What it means. AI engines quote paragraphs that answer. «Dental implants» is not an answer; «How long does a dental implant take and what happens meanwhile» is. It moves the needle most, and nobody in your field does it.

5 E-E-A-T / entities weight 12% · 35/100

The practitioners have no name and no registration number

What was checked

Search the team page for professional registration numbers

What your website answered
(no results)

What it means. In healthcare this counts double. Without a name, a qualification and a registration number, there is no authority an engine can verify, and on health topics engines are especially conservative.

6 Off-site reputation weight 12% · 58/100

Strong in reviews, absent everywhere else

What was checked

Search for mentions in directories and local press

What your website answered
Google Business Profile: 184 reviews, 4.7 ★ (good)
Health directories: incomplete listing, no website
Local press / blogs: 0 mentions

What it means. This is your best score, and it's the one that depends least on your website. The reviews already work in your favour; what's missing is the rest of the internet confirming that you exist.

7 Technical / indexability weight 10% · 40/100

robots.txt lets Google through but not the AI crawlers

What was checked

Request the home page identifying ourselves as GPTBot, ChatGPT's crawler

What your website answered
403

What it means. This is the cheapest fix in the whole report: one line. While it stays like this, nothing else matters, because the AI crawlers never get to read it.

8 Architecture weight 6% · 45/100

Everything hangs off the home page, with no hierarchy

What was checked

Count the pages declared in the sitemap

What your website answered
9

What it means. Nine flat pages. There is no grouping by service and no breadcrumbs, so an engine cannot tell what is primary and what is incidental.

9 Answer potential weight 6% · 20/100

Not a single frequently asked question is published

What was checked

Manual review + search for FAQPage

What your website answered
Frequently asked question blocks: 0
FAQPage schema: 0

What it means. It is the most direct route to being cited: the literal question, a short answer underneath. Without it, the AI has to build the answer itself, and it would rather use someone who has already written it.

The improvement plan

What to do, in what order, and who does it

Thirteen things in three waves. The order is not by difficulty: it is by dependency. While robots.txt blocks, nothing else matters, because nobody reads it.

How long it takes. From the moment you approve the quote, the first two waves —the ones that move the score— take about six weeks. The third one never ends: it is maintenance.

Three things depend on you, and they are marked in the "who does it" column: each practitioner's registration details, signing off the ten frequently asked questions, and the decision on publishing prices. If they arrive within the week the plan indicates, the date holds. If they slip, everything after them slips by the same amount, because it is chained.

And these are calendar weeks, not working time: nobody spends six straight weeks on this. The «effort» column holds the real hours.

Wave 1 · What unblocks everything else First week

#What to do, and whyImpactEffortWho does itWhen
1Remove the AI crawler block in robots.txt
While it returns 403, ChatGPT, Claude and Perplexity cannot read the site. Everything else depends on this.
High5 minutesUs, or your developerDay 1
2Rewrite the first sentence of the home page
It should say what you are, where you are and who for, before anything else. It is the sentence the AI copies.
High2 hoursUsDays 1-2
3Publish the Dentist schema.org record
Address, phone, opening hours, services and accepted insurers, in a format the machine reads without guessing.
High3 hoursUs (file ready in the appendix)Week 1
4Give every practitioner a name, a qualification and a registration number
In healthcare, verifiable authority counts double. Without it, engines don't cite you.
HighYours: the dataYou send them to usWeek 1
5Fix sitemap.xml and the dates
Nine URLs carrying the same 14-month-old date tell the engine the site has been abandoned.
Medium2 hoursUsWeek 1

Wave 2 · What makes you citable Weeks 2 to 5

#What to do, and whyImpactEffortWho does itWhen
6Ten real frequently asked questions, with their FAQPage
The most direct route to being cited: the literal question, a short answer underneath. All ten come out of this report.
High2 weeksWe write, you approveWeeks 2-3
7One page per service, each answering one question
«Implants» is not an answer. «How long an implant takes and what happens meanwhile» is. Seven pages, seven questions.
High3 weeksUsWeeks 3-5
8Reference prices or ranges
Six of the thirty questions were about price and you appear in none of them. Whoever publishes a range gets cited.
HighYour callYou decide, we write itWeek 4
9Complete the Google Business listing and the health directories
It is your best score and it is half done: the listing doesn't link to the website.
Medium1 dayUs, with your accessWeek 5
10Publish llms.txt
A plain-language index of what is on the site and what each page answers.
Medium2 hoursUs (file ready in the appendix)Week 5

Wave 3 · What keeps it up From week 6

#What to do, and whyImpactEffortWho does itWhen
11One article a month answering a question from the neighbourhood
Freshness is 12% of the score and today it sits at 30. One article a month holds it up.
MediumOngoingUs or youFrom month 2
12Hierarchy: group by service, with breadcrumbs
So the engine knows what is primary and what is incidental.
Low1 weekUsWeek 6
13Measure every month whether they cite you, and who they cite instead
The same 30 questions, every month, to see whether the needle moves.
MediumOngoingUs (separate service)From month 2
Appendix · Files already written

Ready to install

This is not an example of how it would be done: it is the file, written for your site, for you, your developer, or us to install. What sits in brackets is your data, which we don't have.

In this sample the files are cut short, and you can see where. In your report they are complete.

1. robots.txt

Replace the whole file. The line blocking you was Disallow: / inside a User-agent: * block that also covered the AI crawlers.

User-agent: *
Allow: /

User-agent: GPTBot
Allow: /

User-agent: OAI-SearchBot
Allow: /
… the complete file, written for your site, comes inside your report

2. llms.txt

Goes in the root, next to robots.txt. It is a plain-language index: it tells a model what is on the site and what each page answers.

# Clínica Dental Montsol

> Neighbourhood dental clinic in Gràcia (Barcelona). General
> dentistry, implants, orthodontics and paediatric dentistry.
> We speak Catalan, Spanish and English.

## The essentials
- Address: [street and number], 08012 Barcelona (Gràcia)
- Phone: [phone] · Out of hours: [phone]… the complete file, written for your site, comes inside your report

3. schema.org Dentist record (JSON-LD)

Goes inside the home page's <head>, in a <script type="application/ld+json"> tag. Fill in the fields in brackets.

{
  "@context": "https://schema.org",
  "@type": "Dentist",
  "name": "Clínica Dental Montsol",
  "url": "https://clinicamontsol.example/",
  "telephone": "[+34 ...]",
  "address": {
    "@type": "PostalAddress",
    "streetAddress": "[street and number]",
    "addressLocality": "Barcelona",
    "postalCode": "08012",… the complete file, written for your site, comes inside your report

4. The first sentence of the home page

Replaces «Welcome». It is the sentence engines copy when they answer «dentist in Gràcia».

<h1>Dental clinic in Gràcia, Barcelona</h1>

<p>We are a neighbourhood dental clinic in Gràcia… the complete file, written for your site, comes inside your report
Uncomfortable truths

What this report does not fix

Being cited is not the same as having more patients. It means being on the list when someone asks. What happens next depends on your website, your phone and your clinic, and no tool measures that.

The chains have a bigger content budget than you. On generic questions («does a root canal hurt?») they will keep winning. Your ground is questions with a neighbourhood, an opening time and a proper name — and there, today, nobody is competing.

Engines change without warning. These 30 answers are from 22 September 2026. In three months they may be different, and not because you did anything. That is why the last line of the plan is to measure it every month.

On health topics, engines are conservative on purpose. Until names, qualifications and registration numbers are published, there is a ceiling no amount of content gets past.

Method and limits

How this was done

The automated part. The technical checks run on our own tool, which only makes read requests: it touches nothing on the site. It identifies itself with each engine's real agent (GPTBot, OAI-SearchBot, ChatGPT-User, ClaudeBot, Claude-User, PerplexityBot, Googlebot and Bingbot) and keeps the verbatim response.

The manual part. The 30 questions are written by hand for your business and your area, and put to all three engines on the same day. The comparison with the two competitors uses the same nine blocks.

The limits. An AI engine does not always give the same answer to the same question: these are one particular day's answers, not an average. We do not measure traffic or sales, because they cannot be seen from outside. And we do not go into the parts of the site behind a password.

What happens next

Three routes, and all three are legitimate

1. You do it yourselves with this report

Everything that needs doing is here, in order, and the appendix files are written. You don't need to hire us again. If you stop here, the report has done its job.

2. We do it, at a fixed price

We tell you before starting what we fix, what we don't, and what access we need. Fixed price, no open-ended hours, and the plan's dates in writing. The price of this report is not deducted from the quote: they are two different pieces of work, and you have already received this one in full.

3. Your technical team does it, with our specs

If you have a developer, or another company runs the site, every point in the plan comes with a technical spec for them to carry out. If you like, we review the result when they finish.

And whatever you decide: if in six months you want to know whether the needle has moved, the same 30 questions get asked again and compared with today's. It is the only way to know.

EXITUS AGENCY · Barcelona
[email protected]
Sample report · fictitious business
22 September 2026