GEO (generative engine optimisation)
Why it matters
A growing share of questions is asked to an AI assistant rather than a search engine. The assistant answers with the websites it deems clear and reliable, and it cites them. An association's website that answers "who can benefit", "how to register", "where and when" without detour has every chance; a website that talks about itself in vague terms, none. The difference with classic SEO: the engine shows a list of links, the assistant writes an answer and picks two or three sources.
What it changes in writing
- A short answer at the top of each page, before the development: that is what the assistant picks up
- Dated facts, with their source, rather than adjectives: "open on Tuesdays from 9 to 12 since 2024" rather than "widely accessible"
- Real questions and answers, those your publics ask, with self-contained answers
- A glossary of your sector's words, one definition per page
- Structured data (schema.org) that tells the machine what the page is
- An llms.txt file summarising the website for assistants, like a robots.txt for engines
How to check
Ask two or three assistants the questions your publics ask, and look at their sources. If your website is not there, look at which website is and what it does better: usually, it answers more directly. It is a test to repeat every quarter, like looking at your queries in Search Console. GEO does not replace SEO: the same clear, structured, sourced pages serve both.
In Kern4Good
Every website of the programme delivers structured data and the llms.txt file, from the launch website. This website applies the recipe to itself: a short answer at the top of every definition, sourced facts in the guides, an FAQ per page. Answer-oriented writing is part of the content work, a few days a month, on your side after training or as support on quote, on the Pricing page.
Further reading
Do we need a special tool for GEO?
No. You need pages that answer clearly, sources, a glossary and structured data, which the programme's website provides. The rest is writing.
How do we know whether an AI cites our website?
Ask it the questions your publics ask, and look at its sources. It is a test to repeat every quarter, like looking at your queries in Search Console.
What is the llms.txt file?
A text file at the root of the website summarising what the organisation is, what it offers and where to find what, written to be read by an AI assistant. It guarantees nothing, it helps the machine understand the website without guessing.
Do AI assistants send visitors?
Fewer than search engines, but visitors who already got an answer and come to check or act. For an association, being cited correctly often counts more than the number of clicks.