Brand visibility in AI answers: what AI says about a company.
GEO, generative engine optimisation, is the work that gets a brand to appear, with accurate information, in the answers of ChatGPT, Gemini, Perplexity and Google AI Overviews.
The report is a document: a person at AIR reads it and goes through it with the company. Fixes to the website, listings and content are made together.
- ChatGPT, Gemini and Perplexity: dozens of variants of each question
- Google AI Overviews: checked on the main questions
- Report read and annotated by a person
AI visibility: the answer comes before the click
Customers ask ChatGPT, Gemini and Perplexity which solution to choose and who to turn to. When the AI answers, the customer often opens no website at all: they read a few names and choose among them.
A brand that is missing, or shown with wrong information, loses enquiries without ever seeing them. The conversation happens elsewhere, outside the website's analytics tools.
How to appear on ChatGPT: five questions to measure
Is the brand named?
How many answers it appears in, for each question and each engine. Everything else starts from this figure.
Next to whom?
Which competitors appear alongside it, who comes first, and when the AI recommends others.
Is it true?
Prices, locations, opening hours, offer: every wrong or outdated detail is flagged, with the answer that contains it.
In what words?
What the AI knows the brand for, which needs it links it to, and whether the tone is the intended one.
Where does it come from?
The pages, portals and reviews the AI draws on. That is where you act to change the answer.
What to do, in order
A list of changes to the website and content, from most to least useful. Then we measure again.
How we work on GEO
A single AI answer is too small a sample: the words and names change with every request. Each question is asked in dozens of variants on ChatGPT, Gemini and Perplexity, and we look at the proportions, not the lucky hit. AI Overviews are checked separately, because they cannot be queried systematically. The same method as our processes.
What the client keeps
- A report you can read in ten minutes, with the AI's real answers as evidence.
- The brand's presence by question and by engine, with competitors alongside.
- Every wrong or outdated detail, with the source to correct.
- Changes to the website, listings and content, in order of usefulness.
- A new measurement after the fixes, to see what has changed.
GEO vs SEO: what changes with AI Overviews and AEO
SEO measures position among the links. GEO measures what the AI's written answer says. AEO, answer engine optimisation, is the term used for engines that answer questions.
Frequently asked questions on visibility in AI answers
What is GEO?
GEO, generative engine optimisation, is the work that gets a brand to appear, with accurate information, in the answers of ChatGPT, Gemini, Perplexity and AI Overviews.
What is the difference between GEO and SEO?
SEO looks at position among Google's links. GEO looks at what the AI's written answer says about a brand, which customers often read without opening any link. The pages you work on are largely the same; the measurement is different.
Are GEO and AEO the same thing?
Almost. AEO, answer engine optimisation, means optimising for engines that answer questions. Here we use GEO for all the work on AI-generated answers.
How do you appear on ChatGPT?
A brand appears in answers when the sources the AI draws on describe it clearly and consistently: the website, portal listings, reviews, and content that answers customers' questions. You start by measuring what the AI says today, then you correct those sources.
How do you measure visibility in AI answers?
Each typical customer question is put to ChatGPT, Gemini and Perplexity in dozens of variants. We count the answers in which the brand appears, next to whom, and with what information. Google AI Overviews cannot be queried systematically: they are checked separately, on the main questions.
Do you need access to the company's systems?
No. The work concerns what the AI sees, which is what is public. We need customers' typical questions and the competitors to compare.
Can you change what the AI says?
Correction goes through the sources the AI draws on. The website, portal listings and content are fixed together with the company, then we measure again to see the effect. The outcome depends on sources and engines: we measure it, we don't promise it.
How often should the measurement be repeated?
After each round of fixes and when new models come out, as they often change the answers. The frequency is agreed according to the sector.
Who reads the report?
A person at AIR, who goes through it with the company and prepares the fixes to the website, listings and content with them. The report is discussed together.