AI visibility
We measure whether AI systems mention your brand when someone asks them about what you sell, and we work on the content so they start doing it.
A search engine returns ten results and the user chooses. An AI returns one answer and cites two or three sources. Showing up stops being a question of position and becomes binary, and that changes how content has to be written and structured.
We ask ChatGPT, Perplexity, Gemini and Claude the questions a customer in your industry would ask and check whether the brand appears. Measurement is recurring, because the same prompt doesn't return the same answer two days in a row. That leads to the actions: which brands hold that space, which sources back them and what your content is missing to get into the answer.
AI visibility
4 areas
Do AI assistants mention you when someone asks about what you sell?
We measure a set of queries from your industry across several models, on a recurring basis: whether you appear, where in the answer and how often.
Starting pointWhat do they say about you, and who do they think you serve?
How an LLM sums up your offering, what audience it attributes to you and what use cases it associates with you. When it doesn't match what you sell, the problem is in your content.
How they describe youWho shows up instead of you?
Which brands win the answers you don't and which sources back them. That's the map of what needs to be built.
Who holds your spotCan AI crawlers read you, and is what they find useful?
We review access for GPTBot, OAI-SearchBot and Googlebot, and rewrite the content with specific data, clear entities and direct answers, which is what an LLM cites.
ActionableThe first measurement ends in a report like this one: how often the model retrieves you when nobody names you, what it knows about you when asked by name, the tone it uses about you and who takes your place. The same questions are repeated every month, and what matters is how the results move.
Baseline report · Artificial intelligence · Measurement
ChatGPT
The baseline in ChatGPT
30 questions on visibility, brand knowledge and sentiment · first measurement
Recall
Absent from all 10 general questions
Brand knowledge
Model's confidence when answering about the brand
Sentiment
Tone of the answers about the brand
Sentiment by topic
Who shows up in general questions
Share among the five most cited
Your brand: 0 mentions. This is where the map of what to build comes from.
First measurement. The questions are repeated every month to track progress.
AI visibility
We start from the questions someone would ask on the way to you, not from a keyword list. What gets evaluated is the model's full answer, not a position on a list.
The same prompt gives different answers from one day to the next. A one-off audit tells you nothing: we measure on a recurring basis and work on the trend.
Checking that AI crawlers can read your website is quick to fix. What takes effort, and what decides, is having content worth citing.
Nobody controls what a model answers or can guarantee a mention. What we can do is measure where you stand, find out who holds your spot and work on the reasons they would cite you.
Trademarks of their respective owners. We name them because we measure visibility in them; this implies no relationship with any of them.
The work
Nobody controls what a model answers. What can be done is measure, find out who holds your spot and work on what is within your reach. These are the workstreams, in the order they open.
01
Measure
Where you stand, and who holds your spot
The questions someone would ask on the way to your business, put to the four models on a recurring basis: whether you appear, in what position, with what description and which sources back the answer.
Which competitors show up for each query and which sources back them. It's almost always a handful of pages: directories, comparisons or industry media.
What value proposition it attributes to you, for which audience and which use cases. When it doesn't match what you sell, the problem is in your content.
02
Technical
Fixed in days
GPTBot, OAI-SearchBot and Googlebot can read the website, the content is in the HTML and not behind JavaScript, and there are no leftover blocks. Fixed in days, and part of the SEO and GEO service. See the page →
Who you are, what you do and where, stated the same way wherever models look for it. That's what lets them recognize you as the same brand everywhere.
If you want to be cited, they have to be able to read you; if your business is selling the content itself, the decision may be the opposite. We explain each option and carry out the one you choose.
03
Content
This is the work
Direct answers to real questions, verifiable figures and dates, clear entities and text that stands on its own outside its page. This is the real work.
Being in the directories, comparisons and media that models already use as references in your category.
04
Measure again
Month by month
The same set of queries, month by month, to see whether it moves. The series over time is the data; a one-day audit is an anecdote.
A search engine returns ten results and the user chooses among them; an LLM, the language model behind ChatGPT or Perplexity, returns a single answer and cites two or three sources. Showing up stops being a question of position and becomes binary: either you're in the answer or you don't exist for that query.
We start from the questions someone would ask on the way to your business and put them to ChatGPT, Perplexity, Gemini and Claude. For each answer we record whether you appear, in what position, with what description and which sources back it.
Checking that GPTBot, OAI-SearchBot and Googlebot can read your website, that the content is in the HTML and not behind JavaScript, and that there are no leftover blocks in the robots file takes days. It's a necessary condition, and many people mistake it for the whole job.
Nobody controls what a model answers, and nobody can guarantee a mention. Anyone who promises it either doesn't know how this works or is counting on you not checking.
Before we start
What we get asked most before we start working together. If yours is not here, just ask us.
Another question? Write to usSEO and GEO prepares the website: so Google ranks you and AI can read and understand it. AI visibility measures whether you actually get named: we ask the models your customers' questions, record who appears and with which sources, and work to get into those answers, where there is often no click. What gets measured is no longer position but how often you appear. The two reinforce each other: content that answers a question well works for both.
By repeating. The same prompt gives different answers from one day to the next, because models sample and because the sources they consult change. That's why we measure a set of queries on a recurring basis and work on how often you appear over time, not on a screenshot from one Tuesday.
No, and be wary of anyone who guarantees it. There's no inventory to buy and no position to reserve. What we give you is a measurement of where you stand, a map of who holds your spot and with which sources, work on what is within your reach, and repeated measurement to check whether it moves.
Longer than in search engines, because besides publishing the content, two things have to happen: the crawlers have to pick it up, and the sources the models consult have to pick it up in turn. It's reasonable to plan in months and to measure from the first month, to see movement even before there are results.
It depends on what you're after, and it's a business decision before a technical one. If you want to be cited, they have to be able to read you: blocking GPTBot while hoping to appear in ChatGPT is contradictory. If your business model is selling the content itself, the decision may be the opposite. We explain what each option means and carry out the one you choose.
It works, and sometimes better. In local and niche queries there are far fewer sources competing, so the bar for getting into an answer is lower than in a crowded category. What it takes is company information that is consistent and verifiable wherever the models will look for it.
Let's talk
We'll get back to you within 24–48h for a first, no-commitment conversation. If we can add value, we'll tell you clearly.
Initial strategic audit, at no cost
Thank you!
We've received your request. We'll be in touch as soon as possible.
Only your name and email are required.