Your customers ask an AI
where to book.
Does it answer your name?
Etalo measures how often your brand appears in AI assistants' answers — and shows why it doesn't.
Assistants queried
ChatGPT
Claude
Gemini
Perplexity
Mistral Le Chat
Grok
DeepSeek
Google AI Overviews
Google AI Mode
Bing Copilot
The shift
68 %
of US Google searches ended with no click at all between January and April 2026.
SparkToro study on Similarweb clickstream data, reported by Search Engine Land, 2026.
The answer has replaced the list of links
Your future customers no longer compare ten tabs. They ask a question, read one answer, and remember the two or three brands it names.
Inside that answer there is no second page. You are cited, or you do not exist — and unlike search, nothing in your current tooling tells you which.
Your Google position, your traffic, your impressions: none of those numbers measure what an assistant tells someone asking it for advice.
What Etalo measures
A brand position, quantified and tracked over time
A corpus of questions representative of your market is put to every assistant, several times, every month. What comes out is not an impression: it is a share of voice.
Share of voice
Your presence in answers, relative to every brand cited.
Rank
Where you sit among the brands assistants name in your market.
Presence rate
The share of answers naming your brand, with its confidence interval.
Domain citation
When the AI talks about you, does it read your site or only what third parties say?
Accuracy
The share of claims about your brand that survive verification, severity included.
Tone and positioning
How assistants place you when they cite you: the reference, an alternative, or nothing specific.
The real competition
Who gets cited instead of you
Etalo extracts every brand named in every answer, not just yours. You see your share of voice against your competitors, question by question and assistant by assistant.
The gaps are rarely the expected ones: a competitor can dominate ChatGPT and be absent from Perplexity, or own the comparison stage while you hold discovery.
A brand missing from an answer doesn't lose a position. It loses the entire conversation.
What nobody else shows
Why you aren't cited — and what to write to be
An assistant never searches the question it was asked. It breaks it into several queries, reads the pages those return, then writes. Those queries are what decide which brands are citable.
This is a content roadmap, not a diagnosis. A frequent query that never returns your domain is a page to write — and it is written for that phrasing, not for the original question.
Why these numbers hold
A measurement you can challenge line by line
A number nobody can verify ends up rejected wholesale. Etalo is built for audit, not for the demo.
The corpus is frozen
Questions don't change from one month to the next. That is what makes two campaigns comparable — a corpus regenerated every time produces nothing but noise.
Every question is asked several times
AI never answers the same way twice. Every rate is published with its confidence interval.
Everything traces back to the raw answer
Any percentage drills down to the exact sentence that produced it, with the assistant, the model and the timestamp.
The calculation is replayable
Counting rules can evolve without discarding history: the whole past is recomputed with the new version.
What nobody else measures
What AI says about you, and whether it is true
Counting citations does not tell you whether what is said is accurate. A price that no longer exists, a discontinued offer still attributed to you, a forum opinion repeated as fact: those are the errors that cost deals, and no visibility curve flags them.
Etalo checks the claims in the answers against a truth sheet you approve, and produces the list of inaccuracies ranked by severity — with the question, the assistant and the date of each finding.
An inaccuracy verdict not reviewed by a human blocks the report export. Accusing an answer of being wrong is too heavy a claim to publish on the word of one model judging another.
A heavily cited brand with a wrong price does not have a visibility problem.
The other end of the measurement
What AI crawlers actually do on your site
Everything above is measured inside the answers. Your access logs measure the other end: can the crawlers that feed those answers even read you?
Who came, and what they got
You upload a log — Apache, Nginx, a Cloudflare export — and Etalo pulls out every AI crawler that visited: how many times, on which pages, and what your server answered them.
A verdict per crawler, not a counter
Served. Never came. Refused by the server while robots.txt allows it. Came anyway while robots.txt forbids it. When your two files contradict each other the firewall wins — and nothing else tells you.
The human traffic sent by assistants
A GA4 acquisition export uploaded here surfaces the visits arriving from ChatGPT, Perplexity or Copilot — the ones your analytics files under "direct" because it cannot tell where they came from.
And when the file cannot settle the question, the screen writes a dash. An analytics export contains no crawler visit at all: showing zero would read as an absence, when it is the measurement that is missing. The reason is written on screen, in place of the figure.
Where your data lives
Designed, built and hosted in France
Not a badge: the publisher, the server and the database each carry an address, and you can check it before taking our word for it.
| What lives where | The address, and where to check it |
|---|---|
| The publisher | Galaxy SEO — Aix-en-Provence, France. Name, address and registration number are on the legal notice page: you know who you are talking to before you sign. |
| The server, the database, your files | o2switch — Clermont-Ferrand, France. The site, the database, your question set, your collected answers, your truth sheet and the logs you upload all live there. |
| The questions put to the assistants | At the assistants' publishers, so outside France: the questions go to them, that is the measurement itself and it cannot happen anywhere else. Each one is named in the privacy policy, along with what it receives. |
Of the seven tools our own landscape review compares, Etalo is the only one whose publisher is established in France: the other six are registered in Austria, the Netherlands, Germany, Singapore and the United States.
Public legal notices of each publisher, checked on 22 September 2026. This check covers the seven tools in the landscape, not the whole market.
What it asks of your time
Nothing, once the corpus is written
The measurement repeats itself, and the figures come to your tools rather than the other way round.
Campaigns restart on their own
A start date and a frequency — daily, weekly, fortnightly or monthly. The next campaign is created, armed and collected with nobody there. Manual launch stays the default.
You choose the assistants
Ten surfaces available — ChatGPT, Claude, Gemini, Perplexity, Mistral Le Chat, Grok, DeepSeek, Google AI Overviews, Google AI Mode and Bing Copilot — and you decide which are queried, campaign by campaign. The screen shows the number of API calls before you confirm.
Thirteen signed CSV feeds
One URL per feed, no login, ready to paste into Looker Studio, a spreadsheet or a warehouse. From the campaign summary down to which page was cited, on which question, by which assistant.
A written read-out, not just curves
The campaign analysis is written in your language from that campaign's figures — what moved, why, and what to fix first.
What you receive
Every month
| Deliverable | What it contains |
|---|---|
| Share of voice | Your presence in answers, your rank among cited brands, and the gap with the previous campaign. |
| Competitive map | Who owns which stage of the buying journey, on which assistant, on which questions. |
| Lost questions | Questions where competitors are cited and you are not — ranked by frequency, with the incumbent to displace. |
| Factual accuracy | False claims about your brand — prices, guarantees, terms, scope — ranked by severity, with the answer they came from. |
| Named or linked | The share of mentions carrying a link, those that do not, and the answers citing your site without naming you. |
| AI Score and Benchmark | A technical audit of your site across 18 checks, and the same audit on the competitors AI actually cites. |
| Data feeds | Thirteen signed CSV feeds, ready to plug into your dashboard with nothing exported by hand. |
| Derived queries | What assistants actually search for, and the pages they read instead of yours. |
| Correction plan | Actions to take, prioritised by impact and effort, with an owner and a deadline. |
| PDF report | All of the above in a single laid-out, paginated document: cover, summary, campaign figures, technical diagnosis and plan. The one you circulate in a board meeting. |
| Email alerts | An AI crawler that can no longer read your site, an axis collapsing, a competitor overtaking you: you hear about it the same day, not in the next report. And the other way round too: what improves is reported, with the figure before and after. |
Everything exports to CSV, and the full report to PDF. Read access can be opened to your teams or your agency, by brand and by country.
How it works
First numbers in two weeks
A one-hour interview
Your ranges, your markets, your real competitors. It is the only moment that takes time on your side — and it is what makes the measurement worth anything.
Corpus built and validated
200 questions representative of your market, which you review and approve before it is frozen.
First campaign
Every assistant queried, several passes, with and without web search. You receive the report and the correction plan.
Monthly tracking
The same corpus, every month. From the second campaign on, you are no longer reading a snapshot but a trajectory.
The blog
What measuring teaches you
Tool comparisons, method and sourced figures. When a figure is quoted, so is its source, with the date it was read.
-
Why my brand doesn't appear in ChatGPT
You asked the question, and a competitor came out instead of you. Before rewriting anything, work out which of the four causes applies: they share neither the test, nor…
-
How much does a GEO audit cost
Two of the three most-cited tools no longer show any price at all. Here are the ones still public, what the API calls behind them really cost, and why the gap between…
-
AI prompt volume does not exist
Every AI visibility tool ends up printing a number next to each question: "2,400 searches a month". Nobody measures that. Here is where the figure actually comes from…
Let's start by measuring
where you stand.
An initial audit on your market, your competitors and your real questions. You see your numbers before deciding anything.
Or book the one-hour interview directly — Pick a slot →
Email directly — contact@etalo.ai