GEO audit — AI search optimization
Your customers ask ChatGPT. Does it mention you?
We put the questions your buyers actually ask to three AI assistants, and measure whether you are cited, who gets recommended in your place, and what to fix. Once, or every month.
Report within 48h. Secure payment by Stripe. No commitment.
- 40
- real questions
- 3
- AI assistants
- 3
- times each question is asked
- 360
- measurements in total
Search no longer stops at Google
When a buyer asks an AI "which supplier would you recommend?", they do not get ten blue links. They get three names. You are one of them, or you do not exist.
One answer, not a results page
An AI assistant cites three or four sources and recommends two or three brands. There is no second page to fall back on.
Ranking well is no longer enough
AIs read forums, comparison sites, videos and niche articles. A site ranking first on Google can be completely absent from their answers.
Your analytics cannot see it
A recommendation that leaves you out generates no visit at all: it shows up nowhere in your stats. You lose without ever knowing.
Cited, or recommended?
Two different things — and the gap between them is often the real finding.
- Cited
The AI reads your site
Your domain is among the sources it consults to form an opinion. That does not mean it names you.
- Recommended
The AI advises you
Your name is written into the answer your buyer reads. That is the one that turns into a visit, a call, a booking.
- Recommended first
The AI names you at the top
You are among the two or three names that open the answer. Past that, people rarely read on.
For your revenue, being recommended matters more than being cited. In one market we measured, a site was read as a source in 47 answers out of 120 and named in only 29: twenty-seven times, the AI opened that site and then recommended someone else.
How we measure
No question contains your name. That is the condition for the result to mean anything: we look at whether the AI mentions you spontaneously, not whether it can describe you when prompted to.
- 01
40 questions written for your market
Discovery, comparison, best choice, alternative to a competitor, local search, purchase decision: the 40 questions cover the intents a buyer actually moves through. They are written from your business, your site and your area — never generic, never recycled from one client to the next.
- 02
3 AIs asked, 3 times each
Perplexity, ChatGPT and Claude, with web search enabled. Each question is asked three times, because an AI never answers the same way twice: a single measurement proves nothing.
- 03
360 answers analysed
We count your citations, your recommendations, where they sit in the answer, and every domain the AIs went to read to form an opinion.
The assistants we query
Three engines, all with web search enabled — under the real conditions in which your buyers use them.
Perplexity
Sonar
Built-in web search. The most source-driven engine of the three.
ChatGPT
GPT, consumer model
The model actually served inside ChatGPT, not an API variant.
Claude
Sonnet 4.5
Native web search, long and highly structured answers.
Two markets measured
Same method, two markets, opposite outcomes
We measured two industries in exactly the same way. The results have nothing in common — which is precisely why a market average tells you nothing.
In one market measured, nobody exceeds 8% of mentions. In another, a single player takes 31%.
Read the full studyHotels24%One property takes one answer in four. Seven names, no more.
Read the full studyCampsites18%The most-cited site is not a campsite: it is the regional tourist board.
Read the full studyIn one market nobody holds the ground. In the other, a single player takes a quarter of all answers. Your industry looks like one or the other — the audit tells you which, and where you stand in it.
What you get
A report you read in fifteen minutes and turn into decisions.
Your citation rate
Out of 360 answers, how often your site is a source the AI actually read. Broken down per assistant.
Your recommendation rate
How often your brand is named in the answer, and in which position — being cited first is not worth the same as being cited last.
The competitors taking your place
A ranking of the brands AIs recommend instead of you. Not the ones you think you have: the ones the data points to.
The sources AIs read
The domains cited most often in your market — forums, comparison sites, trade media. That is where your presence is decided.
The questions where you are absent
The question-by-question detail, so you can see exactly which topics you do not exist on.
Where to start
The three or four levers your data points to: which domains to get cited on, which topics you are absent from, what is holding your site back. Derived from the measurement, not from general principles.
What the report looks like
Here is a real extract from a study we published: accommodation around Mont-Saint-Michel bay.
90 answers · 10 traveller questions · French, English, German
| Property | Named | Share of answers |
|---|---|---|
| Auberge Saint-Pierre | 22 times | 24% |
| Château de Boucéel | 7 times | 8% |
| Manoir de la Roche Torin | 5 times | 6% |
| Hôtel Ermitage | 2 times | 2% |
Extract from a public, reproducible market study. Your report follows the same format, over 40 questions and 360 measurements.
Why them
Booking is the source AIs read most — 47 times across 90 answers. And yet, of the dozens of places in the bay, only seven are ever named, and the top one is cited twice as often as Booking. Being listed is not being recommended.
What the report concludes
For a property absent from those seven names, the lever is not its own website: it is existing in the content the AI reads — tourist boards, travel blogs, guides — where those seven are described in one sentence.
See the full studyIs this for you?
Let us say it up front: the audit is not worth it for everyone.
Yes, if
- Your customers compare before buying, and search online to do it
- Your competitors are identifiable — they can be named
- A recommendation can turn into a sale, a call or a booking
- Your market is big enough that buyers ask questions about it
No, if
- Your market is so narrow that nobody asks an AI about it
- Your customers come through tenders, network or referral — never through search
- You expect a guaranteed ranking: we measure and advise, we do not promise a position
Two plans
The snapshot, or the film. No commitment either way.
The Audit
€149excl. VAT
one-off
Where you stand today.
- 40 questions written for your market
- 3 AI assistants, each question asked 3 times
- Your citation and recommendation rate
- The competitors taking your place
- The sources the AIs read in your market
- A technical check: what, on your site, stops an AI from reading it properly
- Where to start: 3-4 levers drawn from your data
- PDF report within 48h, invoice with VAT
The Monitoring
Recommended€59excl. VAT/mo
cancel any time
First month €149 excl. VAT — your full audit — then €59 excl. VAT per month.
Your visibility, watched month after month.
- Everything the Audit contains
- Mentions gained and lost since last month
- The competitors that overtook you, and on which questions
- New sources that started citing you — or ignoring you
- This month's fixes, in order of impact
- Cancel any time, on request
Set up by hand for now: write to us and we reply within 24h.
Report delivered within 48 business hours — fully refunded if we run significantly late.
What next?
Would you rather the fixes were made for you? Reply to your report and we will talk it through. Quoted case by case, from what your own figures point to.
Start your audit
Five fields. They are what we use to write your 40 questions — the more precise you are, the more accurate the audit.