You may have heard about a brand new marketing channel over the past few months… GEO. So, what is it and does it actually exist?
GEO is Generative Engine Optimisation — essentially, the practice of optimising to appear in Generative AI, like Large Language Models (LLMs) and AI results.
And while there’s nothing wrong with the concept of optimising for AI, we don’t totally agree with coining it “GEO”.
Because really, it’s still SEO — the search engine is just an LLM.
The fact is: it’s all still SEO. The same old SEO fundamentals still apply but now, because of the shift toward generative systems some aspects matter a little more — but the foundations remain unchanged.
So, let’s get into it.
- “The next frontier of SEO”
- “Optimising for AI answers instead of search results”
- “Ranking inside LLM responses”
- “New techniques that replace old SEO”
- “A separate channel next to Google Search”
Most of the people pedalling GEO services are relabelling existing SEO best practices — like structured data, authority building, expert content, semantic depth — as though they are entirely new.
Let us be clear: it’s all still SEO.
How LLMs differ from search engines
Search engines crawl → index → rank.
LLMs generate → predict → learn using vetted data sources.
This is the basis for why certain SEO elements — like schema, authority, topical depth — matter a little bit more now. Don’t get us wrong, they’ve mattered for a while for traditional search. But we’ll expand on this a bit later.
SEO used to optimise for:
- Crawlers
- Indexes
- Ranking algorithms
Now it must also optimise for:
- Generative models
- Answer engines
- Retrieval-augmented generation
But the core pillars remain:
- Content
- Context
- Structure
- Authority
What does the AI playing field look like?
So, AI is the new shiny form of ranking. But where? And what does it look like?
💥 AI Overviews (Google)
Hybrid of traditional ranking + LLM summarisation.
🌀 ChatGPT (regular mode + browsing / AI mode)
Pulls from model training, trusted citations, and live web retrieval.
👁️🗨️Perplexity
Retrieval-first, citations-heavy, rewards authoritative sources.
✨ Gemini
Similar retrieval + generative synthesis model.
So, how are answers formulated?
Audience insights
You’ve probably seen roundup content before… Articles like ‘The Top 10 Restaurant in Brisbane’ or ‘Brisbane’s Best Digital Marketing Agencies’. In the early days of optimising for these AI results, showing up in these articles is proving to be effective.
If you think about how LLMs work, they take knowledge that exists online and use that to formulate an answer giving users a list of options. Particularly if you ask something like, “I’m in Brisbane, what’s the best restaurant?”
It just makes sense that ChatGPT would trust the consensus from these types of content.
Now, there’s nothing wrong with earning your spot in a roundup article like this. But, in the history of SEO, we’ve seen basic, logical ranking factors get black-hatted over and over again — like link building.
So, this has big potential to become pretty unethical… In fact, we’re already seeing it.
AI likes to quote quotable sources
Again, we’re talking about the consensus of it all. If you’re a quotable source. Ie. You’ve got a good domain authority because your website’s backlinks are full of reputable publications and websites quoting you, then AI is going to feel more comfortable with quoting you, too.
In fact, in a lot of cases, it’s citing you from another source that’s quoting you.
Fan-out queries
When you type a phrase into an LLM, it uses fan-out queries to retrieve all kinds of related data and information.
Keeping on the Brisbane restaurant theme… Imagine you search, “Best restaurants in Brisbane for my birthday.”
LLMs aren’t taking that query and just searching for that. It’s searching for all related facets:
Facet: Seating capacity & layout
- Which restaurants in Brisbane accommodate large groups?
- Brisbane restaurants with private dining rooms
- Venues with banquet-style seating
Facet: Menu structure
- Restaurants offering share plates suitable for groups
- Dietary-friendly group dining options (vegan/gluten-free)
Facet: Occasion
- Best Brisbane restaurants for birthday groups
- Casual group dining options in Brisbane
- Upscale restaurants suitable for corporate group dinners
Facet: Price
- Affordable group restaurants in Brisbane
- Group set-menu pricing comparisons
Facet: Neighbourhood
- Group-friendly restaurants in Fortitude Valley
- Group dining options in South Bank
- Best group-friendly restaurants in Brisbane CBD
An LLM won’t just answer the question. It’ll consider all questions at once to determine the very best answer to give you all the context you might need.
Semantic search using NLP and context
Semantic search uses Natural Language Processing (NLP) to not only understand your query, but the intent behind your query. It goes beyond keyword matching to understand the context and deliver you good and relevant results.
It maps the query into a network of entities. So, using the restaurant example again, imagine you’re looking for a restaurant for date night — that query is made up of a list of entities: restaurants, suburbs, cuisines, occasions. Then there are the relationship entities: romantic → ambience, views → riverside dining, special occasion → degustation menus.
This is why a page can rank even if it never uses the exact phrase the user typed — because both search engines and LLMs understand the concept — not just the words.
What really matters more in the era of AI (the fundamentals that mean more now)
So, what are we focusing on more now that we’re doing SEO for both search engines and LLMs? Well, much of the same.
Schema: Making your content machine-readable
Schema Markup, or structured data, is code that lives on your website and helps bots understand exactly what is on the page and what they need to pay attention to.
A bot lands on your page and schema tells them what the page is: a service page or an article. Then it marks up the content with structured data to make it really clear what text relates to what throughout the page.
Schema helps search engines and AI systems understand the context of your content. It essentially marks up literal answers that AI can just have.
The other big thing with schema is that with both search engines and LLMs, they want to know you’re legit. That you’re a real business. By making your business details and NAP (name, address, phone number) easy to quickly identify, you make it easy to quickly verify your business — making your website more citable.
Authority: Being quotable (with digital PR)
As we touched on earlier, LLMs lean heavily on trusted sources. Digital PR was already essential with traditional SEO but now, it’s even more essential. Brands with external citations and news coverage are more likely to be quoted in AI answers.
Round-up content
Now, the potential for exploitation of this aside, obviously AI assumes “Top 10 X” lists are authoritative. If you appear in them, you’re likely to get mentioned as an answer to those commercial queries.
Topical authority: Proving contextual relevance
Both search engines and LLMs want you to be an authority on what they’re quoting you on. They don’t want to quote a plumber on a house painting query.
The deeper and more niche your topic clusters can go, the better. By covering your chosen field in depth, from all angles, you can build on that semantic depth and show up for both informational and commercial AI queries.
Fan-out queries and semantic search
Like we explained earlier, LLMs interpret intent, context, facets, and related topics. So, we need to be considering fan-out queries and semantic search in our content. For semantic search, content clusters, internal linking, and covering your topic in full is essential.
The semantic search aspects are just as important for those fan-out queries — it basically comes back to making sure you’re answering broader intent… Not just keywords.
Better content structure
Create a content structure that is both easy for humans to quickly skim and understand. But also, creates a really clear path for AI and bots to interpret and logically group and summarise.
Use plenty of descriptive headings with the right header tags, answer questions in a question and answer format, mark up your content with schema, and break it all up into bite-size pieces.
And of course, there are the experimental bits
Just like with traditional search, there are specialists who are theorising and experimenting with different ideas. There’s llms.txt — similar to robots.txt — which is meant to direct AI crawlers and tell them:
- which parts of your site they’re allowed to use for training
- which parts they shouldn’t use
- what attribution you expect
- how you’d prefer your content to be represented
Now, Google did add this to their Developers domain, but they have since removed it. There is no evidence that this actually works.
In the same way, people are marking up their metadata with AI directives, re-optimising their sitemaps, and optimising pages specifically for model training. While experimenting is a great way to find out what works — at this stage, all of these experimental bits are anecdotal and have no firm evidence that they actually work.
It’s also worth noting that, just like with Google and Bing, LLM algorithms will continue to evolve.
So… is GEO real? Yes, but really, no.
Yes, there are aspects of SEO that are now more heavily weighted and more important than before that we’ll want to focus on to get you ranking in AI and LLMs.
But also? No. Nothing here is fundamentally new. Realistically, the term GEO is just repackaging SEO.
Some agencies and specialists will disagree with this. And yes, there are extra things we can be doing, but good agencies will just continue to evolve in the SEO space.
After all, LLMs are just about searching for answers. Search engines are just about searching for answers. So, it’s all SEO.
SEO evolves but it’s nothing new 🤖
SEO is constantly evolving. We’re not selling GEO as a standalone product, because it doesn’t make sense to do so. Sure, there are things that can be done to get your website up to speed and get ahead in this new era of SEO. But ongoing, it’ll still come back to SEO.
But we are adapting our SEO strategies to ensure our existing clients are set up to rank in LLMs and these AI results.
We’re always keen to future-proof our clients’ SEO strategies.
Want a hand ranking on traditional search engines like Google or for LLMs like ChatGPT? We’d love to help. Get in touch with us. 👋
