GEO, or generative engine optimisation, is the practice of structuring your content so that AI tools reference it when they answer a question. Where traditional SEO tries to win a position on a results page, GEO tries to make your site one of the sources an AI draws on when it writes its answer. The tools in question are the ones people increasingly ask instead of searching on traditional search engines – ones like ChatGPT, Google’s AI Overviews, Perplexity, Gemini and Claude.
That’s the idea in brief. The rest of this page explains how it works, how much of it is truly new, and what really makes a difference. We’ll be straight with you throughout, including – perhaps slightly contentiously – discussing where we think elements of GEO are overstated.
Why people are talking about GEO
People are starting to ask AI assistants the questions they used to type into Google. When they do, they often get an answer without clicking anything (which is why this is called “zero-click search”). The AI reads a handful of sources, writes a reply in its own words, and sometimes names where the information came from.
If your business is one of those named sources, you have reached someone who was quite likely researching to help them make a decision, and you’ve done it with the AI’s implied endorsement. If you’re not, you’re invisible on that surface, no matter how well you rank on Google.
That’s the opportunity, and the risk, in two sentences. GEO is the work of being one of the sources rather than none of them.
Know this: the term “GEO” is a mess
You’ll see this idea called several things. GEO is the most common. You will also see AEO (answer engine optimisation), AIO, LLMO, and a few others. They mostly describe the same work.
The term generative engine optimisation comes from a 2023 academic paper, not from an AI company’s marketing department, which is unusual and, it has to be said, a little bit reassuring. Since then the label has been pulled in different directions by everyone selling a tool for it.
Our advice is not to get attached to the acronym. The names will keep changing. The underlying question doesn’t: when an AI answers something in your field, does it mention you? Optimise for that, and you’re covered, regardless of what you call the technology/process/science/techniques deployed.
How an AI assistant decides what to cite
To do GEO well it helps to know, in plain terms, what happens when someone asks an AI assistant (or chatbot, ot whatever you want to call it) a question. It’s not as mysterious as it sounds.
Most AI search tools don’t simply invent an answer from memory. They search the live web first, read what they find, and write a reply based on it. The technical name for this is retrieval-augmented generation, but the idea is simple: retrieve some pages, then generate an answer from them.
Two things about that process matter for you:
Firstly, the AI assistant usually breaks the question into several smaller searches rather than one. Ask it for the best CRM for a small recruitment agency and it may silently search for small business CRMs, CRMs for recruiters, and CRM pricing, then combine what it finds. So the page that gets cited is often the one that cleanly answers a narrow sub-question, not the one targeting the big obvious phrase.
Secondly, the AI does not copy your page. It reads it, extracts the part that answers the question, and rewrites it. That rewrite is where citations are won or lost. A page that states its answer plainly and early is easy to extract. A page that buries the answer under a thousand words of preamble is not.
Is GEO just SEO with a new name?
This is the honest question, and it deserves an honest answer: mostly, but not entirely.
In May 2026, Google published its first official guidance on optimising for its AI features. Its position was blunt. Optimising for AI search, it said, is still SEO. The pages that AI tools cite tend to be the pages that were already crawlable, fast, trustworthy and useful. If your SEO foundations are weak, no amount of GEO tactics will rescue you, because the AI is drawing from the same web that everyone else is.
So the foundation is unchanged. A site that can’t be crawled, that loads slowly, or that no one trusts will struggle in AI answers for the same reasons it struggles in search.
What GEO adds to this is a change in emphasis. Answering questions directly matters more than it used to, because, as we’ve seen above, clarity increases your page’s chances of being extracted and cited by AI. Getting mentioned favourably on other sites matters more too, because AI systems build their answers from third-party sources – reviews, trade publications, forums – rather than trusting your own marketing copy about yourself. And the way you measure success changes, because a citation can influence someone who never visits your site at all.
That is the real state of things. GEO is not a separate discipline you bolt on. It is SEO done well, with a few new habits, aimed at where people now find answers. Anyone selling it as a revolution that makes your existing SEO obsolete is overselling.
What makes a page citeable
There is no trick, and the specifics deserve their own guide. But the principles are consistent and none of them are exotic.
Answer the question early. Say the useful thing in the first paragraph, then expand. This helps the AI extract you, and it happens to help human readers too.
Be specific and verifiable. AI tools favour content with concrete facts, figures and clear claims over vague, hedging prose. If you can cite a real number or a dated source, do it.
Earn mentions elsewhere. A large share of what AI tools cite is earned coverage, not a brand’s own website. Being referenced by a publication, a directory or an industry site raises your odds more than yet another page on your own blog does.
Demonstrate genuine expertise. These systems are built to favour content that shows real experience and authority. For a business, that means content written by people who actually do the work, not generic copy assembled to fill a page.
Keep it readable. It’s possible to over-structure content for machines, with endless bullet points and robotic phrasing, to the point where a human bounces off it. That’s self-defeating, because a page no one wants to read earns no links or trust, and links and trust are what get you cited. Write for the person first.
How you measure whether it’s working
This is the part most people skip and it’s the part that decides whether the effort was worth it.
Traditional rankings only tell you part of the story now, because a citation can do its job without a click. So you measure a few things alongside your usual reporting.
Watch for referral traffic from AI tools in your analytics. It is usually small at first, often zero, which is exactly why recording the starting point matters. Track whether your brand appears when you ask AI tools the questions your customers would ask. And keep an eye on branded searches, since people often see you in an AI answer and then look you up directly.
None of this is instant. Building the kind of authority AI tools reward takes months, in the same way ranking does. Anyone promising faster is selling something.
Where to start
If you’re doing SEO properly already, you’re most of the way there. The additional work is a change in habit: answer questions directly, prove your expertise, earn mentions on sites that matter and measure the new places where answers appear (what UX designers call “the surface”) as well as the old ones.
If your SEO foundations are shaky, start there, because GEO built on weak foundations doesn’t work well.
We work on SEO and GEO together, because for most businesses they are the same project. If you want to talk through where your site stands, our SEO and GEO service page explains how we approach it, or you can get in touch directly.
