How to prepare your site for AI Overviews and generative search?
Clients increasingly ask AI instead of Google. I show what to do technically and content-wise so generative search engines cite you.
In short
GEO (Generative Engine Optimization) means preparing your site so generative search engines — AI Overviews in Google, ChatGPT, Perplexity, Claude — can correctly read, cite, and recommend your business. The key elements: clear, direct answers at the top of each section, schema.org structured data, server-side rendering, an llms.txt file, and consistent, verifiable facts (prices, scope, contact).
More and more people don’t scroll through ten blue links — they ask an AI assistant and get one answer with a few sources. If your business isn’t among those sources, for some clients you simply don’t exist. GEO is the work you do to get there.
How GEO differs from classic SEO
SEO fights for a position in the results list. GEO fights to be cited in a generated answer. The foundation is shared — a fast, well-structured site with valuable content — but the emphasis shifts: the language model has to be able to pull one concrete, unambiguous sentence from your site and safely repeat it.
The gist
AI doesn’t cite “pretty” sites. It cites sites you can pull a clear fact from without risk of getting it wrong. Your job is to state that fact plainly.
Seven things that genuinely help
- 01Answer “directly” at the start of each section. The first sentence after a heading should carry the full answer — a definition, a number, a specific — with the context coming only after.
- 02Add schema.org structured data (Organization, Service, FAQPage, BlogPosting, BreadcrumbList). It’s a map the model moves along faster and more confidently.
- 03Render content on the server. If text only appears after JavaScript runs, some bots won’t see it. SSR solves this by default.
- 04Keep facts consistent across the whole site — prices, timelines, scope, and contact must be identical on the page, in the FAQ, and in the structured data.
- 05Add an llms.txt file with the most important facts about the company and offer in plain text — a shortcut for AI agents.
- 06Build content around real client questions (FAQs, guides), not around keyword phrases for the algorithm. Models answer questions, so they reward sites that answer them.
- 07Look after E-E-A-T: a named author, experience, verifiable data. Models prefer sources they “trust”.
What llms.txt is
llms.txt is a simple text file in the site’s root directory (e.g. web-craft.com.pl/llms.txt) where you summarize for language models what the business is, what it offers, and how to get in touch. It works like an extended, machine-readable summary — a cheap investment that tidies up what AI has to say about you.
How to check whether it’s working
- Ask ChatGPT, Perplexity, and Google (AI Overviews) about your business and services — check whether the answer is correct and whether it names your site as a source.
- Verify the structured data in Google’s Rich Results Test.
- Check that the content is visible in the HTML without JavaScript (view the page source).
- Make sure prices and facts are the same everywhere — a mismatch is the most common cause of wrong citations.
Honestly
GEO gives no guarantee that a specific model will cite you — nobody controls AI answers. What it does give is a much better chance and, importantly, it overlaps with good SEO. Do one and you gain the other.
Where to start
If you’re building a site from scratch, build GEO in from day one — it’s cheaper than bolting it on later. If you already have a site, start with a technical audit: we’ll check rendering, structured data, and fact consistency, then decide what to fix first.
Frequently asked questions
What is GEO?expand_more
Does GEO replace SEO?expand_more
Am I guaranteed that AI will cite my business?expand_more
Have a similar question about your own project?
The first consultation and quote are free. I usually reply within one business day — with specifics, not a generated offer.