When your clients put a question to ChatGPT, Gemini, Perplexity or Copilot, the answer reaches them without a visit to your site. The question is no longer whether you rank on the first page of Google, but whether you are the source the model cites. That is what we call your AI footprint.
The good news: this footprint is neither luck nor tooling. It is a matter of editorial architecture, and it can be diagnosed. This article gives you the reading grid, a ten-minute test and seven questions to place your brand. You leave with a method, not a promise.
What AI search visibility actually means
AI search visibility is the work that makes your brand visible, cited and faithfully rendered in the answers generative models produce. Where classic SEO aims at a position in a results page, this aims at a place inside the answer the model gives the user directly.
The shift is concrete. A growing share of searches ends without a click: the user reads the synthesised answer and moves on. Search engines started the movement with their summaries; generative models are accelerating it. Your content can feed the answer without your brand ever being named. Your knowledge circulates, your editorial capital dissolve, and nobody knows it came from you.
SEO, AEO, GEO: telling them apart
Three acronyms coexist, and it is worth separating them once and for all.
- SEO — search engine optimisation: being present in the results pages of engines such as Google or Bing.
- AEO — answer engine optimisation: being the direct answer, in a box or a synthesised reply, with no click required.
- GEO — generative engine optimisation: being cited, and correctly attributed, in the answers of generative models.
The difference between GEO and SEO fits in one sentence: SEO fights for a link in a list, GEO fights for a mention in an answer. This is not one more flavour of SEO. It moves the stake from ranking to citation.
Why AI does not cite your brand
Because on any question in your field, you sit in one of three states. Knowing which one is already the beginning of the audit.
- Cited with attribution. You are named and the model links to your site. That is the target: visibility and attribution together.
- Cited without attribution. Your content is reused, sometimes word for word, but you are no longer identified as the source. Your knowledge travels, your brand does not.
- Absent. You do not appear at all. The place is held by others: competitors, media, institutions, third-party sites.
In most cases, absence is not a knowledge gap: the model knows your subject but no longer recognises you as its source. The problem sits upstream, in the way your content is worded and structured. An interchangeable brand is cited by nobody.
The ten-minute test
Here is a simple, repeatable protocol you can run today to measure your citability.
- Choose three real questions your clients ask. Their words, their problem, not your brand name. “How do I cut my delivery times”, not “what does company X do”.
- Put them to four tools: ChatGPT, Gemini, Perplexity and Copilot. In a private window, logged out, so personalisation does not distort the reading.
- Record your state for each answer: cited with attribution, cited without attribution, or absent. Record who holds the place when you are absent.
In a few minutes you have a first map: which subjects the model cites you on, which ones it forgets you on, and to whose benefit.
Seven questions to place your AI footprint
Beyond the test, seven questions situate the maturity of your AI footprint. Answer yes or no.
- When the model is asked about your field without your name, do you appear?
- When you are cited, is the link to your site there?
- Is the content of the answer faithful to your discourse, and current?
- When you are absent, do you know who takes your place?
- Does your offer hold in one clear sentence a model could reuse without betraying it?
- Is your brand voice recognisable, or interchangeable with any player in your sector?
- Are your key contents structured to be quoted: explicit headings, sharp definitions, self-standing answers?
What your answers reveal
Two profiles keep coming back, and each points to a precise cause.
- Mostly “absent” or “cited without attribution”. You have a visibility gap, not a knowledge gap. The model knows your subject but no longer recognises you as its source. This is common, and it is reversible.
- “No” to questions five, six and seven. The cause sits upstream: a blurred offer, an unformalised voice, content that resists reuse. A model renders well only what is clear at source.
The diagnosis is almost always the same. Visibility inside generative models is not a technology problem. It is an editorial architecture problem.
Three levers of a citable brand
Three levers, in this order.
- Clarity at source. A sharp offer, put in one accurate sentence. What is clear to a reader is clear to a model.
- A voice that holds. Recognisable from one piece to the next. An interchangeable brand voice is cited by nobody, and voice is how memorability is won.
- EEAT criteria and editorial citability. Structuring the material so it can be reused and attributed: definitions, self-standing answers, explicit headings, clear signals.
Our principle takes few words: not more content, more coherence. Including in the way models tell your brand.
From self-assessment to a full AI footprint audit
The ten-minute test gives you a photograph. A full AI footprint audit gives you the map. Where the self-assessment probes three questions, the audit runs a structured corpus, by persona, by theme and by platform.
The method we run on engagements rests on a few simple, demanding principles.
- Prompts grounded in reality, modelled on what your audiences actually search, in their words, without jargon. Search behaviour transfers from one tool to the next: someone who phrases a question a certain way in a search engine phrases it the same way to a model.
- Several questioning stances, because a director, a subject expert and an end user do not ask the same question on the same topic, and do not get the same answer.
- A scoring grid that measures not only presence but fidelity: accuracy, alignment with your discourse, quality and nature of the sources cited in your place.
The substitution effect is measurable, and often striking. On a recent engagement for a reference body, a single very common query concentrated more than a million impressions for a few thousand visits: the answer was being delivered almost without a click. When a model answers in your place on that kind of query, the stake is no longer traffic. It is editorial sovereignty — your capacity to remain the source of reference on your own subject.
At the end of the audit, three deliverables: a precise map of your AI footprint, editorial recommendations that trigger citation, and technical recommendations so your material is faithfully rendered by the models.
Where to start
Run the ten-minute test on your three key questions. If most answers leave you absent or cited without attribution, this is not a technical fate: it is a sign that your editorial architecture deserves to be taken back to source. That is precisely the work of an AI footprint audit — moving from the photograph to the full map. Let’s talk about your AI footprint.
—