Every five years or so, our industry decides it needs a rebrand. Some clever marketing type invents a shiny new acronym, and suddenly everyone’s scrambling to become an “expert” in something that didn’t exist last Tuesday. The latest offender? GEO – Generative Experience Optimisation.
GEO is marketing bollocks invented by people who want to sell you expensive courses about “optimising” for AI search and LLMs (Large Language Models) when all you’re actually doing is bog standard SEO for the search engines that feed those AI systems.
Let me be as clear as I can possibly be: GEO does not fucking exist.
We can make up the term. We can use the term. We can sell £997 courses about the term. But the term doesn’t actually mean anything beyond “I want to separate you from your money.”
Why is the term GEO bollocks?
Pretty much every single Large Language Model (LLM) uses one or more search engines to create summaries of documents pulled by that search engine. This is why they’re better named “Summarisation Engines” because they’re just summarising a set of documents the Search Engine retrieved.
This means you are NEVER optimising for the LLM itself. You’re always optimising for the Search Engine that the LLM uses to pull back documents.
Still confused?
It’s like claiming you’re “optimising for the printer” when you’re formatting a document. No, you’re optimising the document – the printer just outputs what you give it.
But everyone’s saying GEO is the future!
Of course they are. It’s much harder to sell “keep doing good SEO” than it is to sell “learn this brand spanking new technique or pay me £0000s to do it for you!”
Let’s look at some of the absolute drivel being peddled:
Claim 1: “GEO is different from SEO because it focuses on AI-driven search engines.”
Bollocks. You’re still optimising for search engines. The AI is just presenting the information differently. Bing literally admits to using schema from your website (an SEO staple for years) to help decide whether you appear in AI overviews.
Claim 2: “GEO requires a whole new strategy approach.”
Complete rubbish. Good SEO has nearly always been about creating relevant, helpful content that answers user questions. That hasn’t changed. In fact, I’d say it’s more important than ever.
Claim 3: “You need to optimise differently for generative AI.”
No, you need to optimise for the search engine that feeds the AI. And guess what? That’s SEO. Always has been.
Claim 4: “GEO is about predicting what users want before they ask.”
You mean like… understanding search intent? That thing SEOs have been doing for over a decade? Awesome, you just described SEO.
The proof is out there…
Test after test shows that AI overviews predominantly come from the top 20 search results in Google or Bing. Your AI visibility correlates directly with… wait for it… your SEO performance!
Yet everywhere I look, people are saying “SEO is dead” (again?), and “GEO is the future!!”. This is a solution looking for a problem.
Who’s using what?
Most “AI search” is powered by a regular search engine:
- Google’s AI uses Google Search (duh!)
- OpenAI uses Bing as part of its process
- Perplexity AI has its own engine but also uses others, that in turn use search engines such as Bing
- Anthropic uses Brave – you can find Brave Search here.
- And so on…
Note that if you’re optimising for Google, you’ll generally do well in other engines too. This isn’t exactly new information that should blow your mind.
AI search couldn’t exist without SEO
Saying you’re doing GEO shows a fundamental lack of understanding of how these systems actually work. The LLM summarisation engine is a predictive text engine. The documents it brings back are supposed to ‘ground’ the predictive response, so it hallucinates less and stays on topic.
The documents ground the summary. The search engine brings back the documents. Really good SEO is what matters here.
It’s like claiming you need special training to get your restaurant menu featured on Deliveroo. No, you need a good restaurant that’s already visible – the delivery app is just another way for customers to find you.
Where does this leave us?
The next time someone tries to sell you on GEO expertise, ask them how exactly an AI Summarisation Engine selects content. If they give you any answer other than “it uses a search engine to find relevant content first,” they’re either lying, or they don’t understand what they’re selling.
I’m not saying AI search isn’t impacting our industry – it absolutely is. But inventing a new terminology to describe “continuing to do good SEO” is just marketing bollocks designed to sell courses, services, and speaking gigs.
Keep creating excellent, relevant content. Keep optimising your technical SEO. Keep building your authority. That was true before AI search, and it’s still true now.
The only thing that’s changed is the number of consultants trying to rebrand existing knowledge as awesome brand new insights.
And that, my friends, is SEO.
Note: I’m not talking about the Large Language Model itself. If you want to try to data poison it or add your information to their training set, that’s a totally different thing that has nothing to do with AI search engines themselves, which are responding to a query or prompt.
