In this article
I got my first client enquiry via ChatGPT in February 2025, not Google, not an ad. An AI model pulled up SmåSeo as an answer to a specific question about SEO for small businesses. These are the times we live in now.
What is GEO?
GEO — Generative Engine Optimisation — is the practice of optimising digital content to be cited, referenced, and recommended by AI-based search and answer systems. Put simply: GEO is SEO for generative search.
Where SEO is about ranking in a list of links, GEO is about being used as a source in a direct answer. AI models like ChatGPT, Perplexity, and Google's AI Overviews give users answers — not lists. And they pull these answers from a selection of websites they consider trustworthy and relevant.
GEO visibility happens in two ways: either your content is used as a source in an AI-generated answer in tools like ChatGPT or Perplexity, or it's included in Google's AI Overviews. Both build on the same foundation — high-quality content structured to be read and understood by machines.
What is generative search, and what is LLMO?
Generative search is search where you get a written answer instead of a list of links. You ask, the model answers, and a few sources get credited in a small row of links beside it. ChatGPT, Perplexity, Gemini, Copilot and Google's AI Overviews are all generative search.
LLMO — Large Language Model Optimization — is a newer name for the work of getting chosen as a source in those answers. You will also meet AEO (Answer Engine Optimization). The terms compete for space, but the actions behind them are the same.
| Term | Full form | What people mean by it |
|---|---|---|
| Generative search | — | The arena itself: search that answers with text instead of links |
| GEO | Generative Engine Optimisation | Getting cited and recommended in generative search |
| LLMO | Large Language Model Optimization | The same practice, seen from the language model's side |
| AEO | Answer Engine Optimization | The same practice, with the emphasis on direct answers |
My advice: don't spend energy picking the right acronym. Spend it on what actually moves you in generative search — clear definitions, paragraphs that stand alone, and one topic you cover more thoroughly than anyone else.
What's the difference between SEO and GEO?
SEO is about being found in a list of search results. GEO is about being used as a source in a direct answer.
| SEO | GEO | |
|---|---|---|
| Full form | Search Engine Optimisation | Generative Engine Optimisation |
| Visibility via | Organic search results | AI-generated answers and AI Overviews |
| Click required? | Yes | Not always |
| Primary signals | Content quality, links, technical health | Topical authority, structured content, brand |
| Metrics | Rankings, CTR, organic traffic | Mentions in AI tools, AI Overview citations, referral |
SEO and GEO are closely linked, but not as closely as many people assume. In July 2025, 76 per cent of citations in Google's AI Overviews came from pages in Google's top ten. By early 2026, that same number had fallen to somewhere between 17 and 38 per cent.
In other words: a high Google position helps, but guarantees nothing. Models increasingly cite pages sitting far down in, or entirely outside, the top ten. I go through why, and what we actually know about how models pick sources, in part 1 of the series on AI search.
Good SEO is therefore still the foundation, and if you're not visible in Google at all, that's where you start. But it's no longer enough on its own.
What I often see go wrong with GEO
As an SEO and GEO consultant, I unfortunately see that the enthusiasm for GEO leads some people to make poor decisions.
- Jumping on GEO at the expense of SEO. Some people read about GEO and start restructuring their entire content strategy — while their fundamental SEO problems remain unsolved. 9 out of 10 searches still happen in Google.
- Producing tons of AI-generated content to increase GEO visibility. AI models are trained on high-quality content — and they're good at identifying generic content without editorial oversight.
- Expecting quick, measurable results. As of today, there's no reliable rank tracker for AI search. GEO is as much about long-term brand building as it is about technical optimisation.
GEO: How to get it right
GEO isn't a revolution — it's a natural extension of good content practices.
- Write clear definitions early on. AI models tend to pull the first clear, self-contained definition of a term. "GEO is..." early in the text is more citable than a definition buried in paragraph three.
- Write for the paragraph, not just the page. Models pull precise answers from individual paragraphs, not from a page's overall authority. Every paragraph should stand alone and answer one specific question. This is one of the few GEO moves with real empirical backing.
- Answer specific questions directly. FAQ sections, "What is..." sections, and question-based H2s directly match how AI models "think".
- Use structured data. FAQ schema and Article schema help AI crawlers understand the structure. This increases the chance that the right section will be cited in the right context.
- Build topical depth. A website with 20 in-depth articles on one subject is cited more frequently than a website with 100 superficial articles on many topics.
- Prioritise the models your audience actually uses. The models have almost separate source ecosystems, so one generic GEO strategy fits none of them well. In Norway, around 94 per cent of mobile AI chatbot traffic goes to ChatGPT, against 5 per cent Gemini and under 1 per cent Copilot. If your customers are B2B or technical, Claude matters more than those numbers suggest.
- Ensure Bing indexing. ChatGPT and Microsoft Copilot use Bing as their technological foundation. Check that your site is indexed and that you haven't blocked Bingbot in robots.txt.
Your GEO action plan: Step by step
| Step | What you do |
|---|---|
| 1 | Check that your SEO foundation is solid — without it, GEO efforts are ineffective |
| 2 | Find out which AI models your audience actually uses, and prioritise accordingly |
| 3 | Review existing content and ensure clear definitions are placed early in each text |
| 4 | Break the text up so each paragraph answers one question and can be cited alone |
| 5 | Build out FAQ sections with direct, concise answers — add FAQ schema |
| 6 | Verify Bing indexing in Bing Webmaster Tools, check robots.txt |
| 7 | Consider llms.txt — easy to implement, but not a priority |
| 8 | Measure: GA4 referral, GSC AI Overview filter, manual testing in ChatGPT/Perplexity |
In summary: My take on GEO
If I'm being completely honest: GEO is real, but overhyped. We know too little to invest significant resources specifically in GEO optimisation. What we do know is that good SEO and good content are the best foundation for GEO visibility.
My advice: keep an eye on developments, implement the simple measures (Bing indexing, FAQ schema, clear definitions), and hold off on major strategy changes until we have better data. Patience is underrated.
If you want to go deeper into how AI search actually works under the hood, I cover it thoroughly in the series What we know about AI search. This article is about the actions; the series is about the mechanics.
