AI visibility
See whether the answer names your brand, and who it names instead
A buyer asks ChatGPT which company to hire. The answer names three of them. Nothing in a rank tracker tells you whether one of the three was you.
Crownler sends the questions your buyer would actually type to ChatGPT, Claude and Gemini, keeps the whole answer, and reads two things out of it: who was named, and which pages the model cited to say it.
What AI visibility actually measures
AI visibility is how often a brand is named in the answers generative engines write, in what company it is named, and on the strength of which sources. It behaves differently from a ranking: there is no page one, the answer is a paragraph, and the model writes it again every time it is asked.
So the unit changes. Instead of a position you get a mention, and instead of a competitor sitting one row below you, you get the list of sources the model leaned on. That list is where the next month of work comes from, because a source you can reach is a mention you can earn.
It also means the measurement has to keep its evidence. A score with no answer behind it cannot be checked by the client paying for it, so every answer is stored word for word, with the date and the model that wrote it.
What Crownler measures today
Written the way the tool behaves, not the way a landing page would prefer it to behave. What is still missing is on this list too.
- Measured today
ChatGPT, Claude and Gemini
Each prompt goes to the three models through their APIs, and the answer comes back whole: stored in full, with the date, the model and the cost of the call.
- Measured today
Mentions and share of voice
Your brand and the rivals you register, counted once per position in the text, so a brand with more registered nicknames cannot inflate its own number.
- Measured today
The sources the answer cited
Every URL that shows up in the answer, grouped by domain and kept in the order it was cited, so you can see which pages are feeding the model.
- Measured today
The tone around the mention
A window of words around each mention, read as positive, neutral or negative. It is a heuristic, and it is labelled as one on screen: useful to separate praise from complaint, not to judge a brand.
- Measured today
AI Overview on Google
Read from the search result itself: whether the block appeared for that query, and which domains and URLs it cites.
- Not yet
Perplexity
It exists in the data model and it is not being queried: the key is not configured yet. A surface we cannot answer for does not ship as if it worked.
This is the screen
The product running on a demonstration site, captured as it is, not redrawn for the page.
Screens from a demonstration project. The brand, the domain and the numbers are invented: no client data is shown.
How it runs
01
Register questions, not keywords
A prompt is a whole question, in your language and your market, the way a buyer would type it. You say which intention it belongs to, and it joins the territory that already carries your Google positions.
02
It runs weekly, and on your click
Each pair of prompt and model runs once a week. Refresh runs it again on the spot, and the same pair is reused for 40 hours, so two clicks in one afternoon cost one call.
03
You read the answer, not only the number
The report shows mentions and share of voice, and under them the answer itself with its cited sources. It has a public link you can send to the client, carrying your agency brand when white label is on.
Who this is for
- Agencies that have to answer, in the monthly report, whether the client shows up when the buyer asks a model.
- In-house teams in a category where the first question of the month goes to ChatGPT and not to a salesperson.
- Consultants who need the evidence in writing, because the claim is new and the client will ask how it was measured.
Who builds this
The company behind Crownler is backed by Google for Startups.
The tool is built by three people who do SEO for a living, and who use it on client work before it ships.
Digo Garcia
AI, technical SEO and engineering
Two decades of the Brazilian internet, with work at Letras.mus.br, Cifra Club and Meliuz. Doing SEO since 2003.
Andre Mousinho
SEO and GEO strategy
Led SEO strategy at Rock Content to more than 7 million visits a month, and led SEO for four brands in the LWSA group.
Gabriel Raposo
Link building and digital PR
Built one of the most respected authority and press operations in Brazil, with projects for Agibank and BTG.
Questions about AI visibility
Does this scrape ChatGPT?
No. Each model is queried through its official API and the answer is stored whole. Nothing is read off a logged in session, and nothing depends on a screen scraper that breaks the week the interface changes.
How often is it measured?
Once a week for each pair of prompt and model, plus any time you click Refresh. The result of the same pair is reused for 40 hours, so a repeated click does not become a repeated bill.
Is Perplexity included?
Not yet. The surface is in the data model and the key is not configured, so it is not being queried. The day it runs, it appears with the same evidence as the other three.
Can I see the answer, or only a score?
The full answer is stored and shown on the report, with the sources it cited. A number with nothing behind it cannot be checked by the person paying for it.
Does this replace rank tracking?
It sits next to it. The same intention can be held on Google and lost on ChatGPT, and Crownler follows both on the same territory. The method page explains how the two are joined.
What does it cost to run?
Every model call draws on the monthly allowance of your plan, and the allowance is debited before the call is made, so a plan never spends past its limit by accident. The numbers are on the plans page.
Related pages
Same tool, a different question each time.
- SEO toolWhat the whole toolkit does, what it costs, and what it replaces.
- AI SEO toolGoogle and AI answers measured on the same ground.
- Generative engine optimizationWhat GEO is, and what a tool can measure about it.
- Answer engine optimizationBeing the source an answer is built from.
- AI Overview trackingThe block above your result, query by query.
Plans from $79 to $379 a month. What changes between them is how many sites you run at once and how many of the 15 modules come in the licence.
See it on your own brand
Open the workspace, enter the brand and the questions your buyers ask, and read the answers the models give today.