Generative Engine Optimisation (GEO)
GEO is the practice of influencing what generative AI systems say about your business when somebody asks them for a recommendation.
We test your visibility across the platforms your customers already ask
This page explains what GEO is, where the term came from, how it differs from search optimisation, what appears to move the needle and what does not. Written to be useful whether you hire us or do it yourself. No jargon for its own sake, and no promises we cannot evidence.
Related pages
A short definition
Generative Engine Optimisation is the work of making a business legible, credible and quotable to systems that generate answers rather than list pages. The output you are optimising for is a sentence, not a position. If an assistant cannot confidently say who you are and why you are a sensible recommendation, it will name somebody else β not out of preference, but out of caution.
Why the term appeared
For twenty years the destination of a search was a page of links, and optimisation meant competing for those links. Assistants changed the destination. The user asks a question and receives a written answer, often with two or three businesses named inside it and no list to scroll. Nobody needed a word for optimising that until assistants started sending real enquiries. GEO is that word. It is new, some of what is sold under it is repackaging, and the underlying problem is nonetheless real.
How a generative engine reaches an answer
Simplifying, three sources feed any answer. First, what the model absorbed during training β broad, dated, and impossible to edit directly. Second, live retrieval: the system searches the web at question time and reads what it finds, which is why some assistants cite sources you can click. Third, connected structured data, such as a mapping or business-listing service the platform has licensed. Different platforms lean on different mixes, and that is the single most useful thing to understand, because it tells you which lever moves which assistant.
What generative engines appear to reward
From testing businesses across the six main platforms, the same themes recur. Consistency: your name, address, phone number and description agreeing everywhere they appear. Specificity: stating plainly what you do, for whom, and where, rather than marketing abstraction. Structure: machine-readable data that removes the need to guess. Corroboration: other credible sources saying the same thing about you. Recency: signals suggesting the business is active now rather than three years ago. None of these is a trick. All of them are things a careful business would want to be true anyway.
Why assistants hedge
You will often see an answer that describes a category rather than naming a company β "look for a local firm with good reviews" instead of a name. That hedge is a signal. It usually means the system found candidates but could not corroborate any of them well enough to commit, or found conflicting information and chose safety. When we see hedging across a sector, the opportunity is generally larger than in a sector where competitors are already being named confidently, because nobody has done the work yet.
The role of Google Business Profile
For any business with a location or a service area, the Google Business Profile is disproportionately important. It is structured, it is verified, it is current, and several systems either read it directly or read sources that mirror it. A profile with the wrong category, an old address, no services listed and no recent reviews is a weak foundation for everything else. Getting it accurate and active is usually the highest-ratio hour of work available, and it costs nothing but attention.
What GEO is not
GEO is not prompt trickery, and it is not a way to make a model say something untrue on your behalf. There is no tag you can add that forces an assistant to recommend you, and anyone selling that is selling nothing. There is no submission form to the assistants. There is no paid inclusion. Hidden text aimed at models rather than readers is the modern equivalent of white-on-white keywords and carries the same eventual outcome. The work is closer to careful information hygiene and reputation building than to a growth hack, which is why it takes patience and why it holds once it is done.
Common myths, briefly
That an llms.txt file will get you recommended β it is a helpful courtesy, not a ranking signal, and no major platform has committed to obeying one. That publishing a hundred AI-written articles will do it β volume without corroboration mostly adds noise, and the models are increasingly good at recognising it. That you must be mentioned on Reddit β genuine community mentions can help, manufactured ones are risky and obvious. That GEO replaces SEO β it does not, and if your enquiries currently come from Google, treating it as a replacement would be an expensive mistake.
Services businesses versus product businesses
The work differs. For a service business the question is usually "who should I use", and the answer turns on identity, location, credibility and reviews. For a product business the question is more often "what should I buy", and the answer turns on the product data itself β accurate specifications, availability, price, structured product markup, and third-party corroboration such as reviews and retailer listings. Both are GEO. The signals you prioritise are not the same, and an agency treating them identically is not paying attention.
How to measure it
The only honest measurement is to ask. Write down the questions your customers would genuinely type, run them across the platforms, and record whether you are named, what is said about you, and who is named instead. Repeat on a schedule, and keep the wording of the prompts identical each time or you are measuring your own phrasing rather than your visibility. Everything else β traffic, rankings, impressions β is a proxy that can move for unrelated reasons.
How often to re-test
Monthly is enough for most businesses. Weekly tells you more about the platformsβ natural variability than about your own progress, and it creates a temptation to react to noise. Keep a dated record of every run. Because these systems change without notice, an undated screenshot is close to worthless six months later, and a dated one is evidence.
Where to start if you are doing it yourself
Four steps, in order. Check that AI crawlers are not blocked in your robots.txt β this alone occasionally explains everything. Make your homepage state plainly who you are, what you do and where you operate, in the first paragraph, in words a stranger would use. Get your name, address and phone number identical across your site, your Google Business Profile and every directory you appear in. Add Organization or LocalBusiness structured data that matches your visible text exactly. Those four remove most of the common reasons a business is skipped, and none of them requires an agency.
The things people ask us first
Is GEO a real discipline or a marketing label?
Does GEO replace SEO?
How long does it take?
Can you guarantee an AI will recommend me?
Does an llms.txt file help?
Should I block AI crawlers to protect my content?
Is this worth doing for a very local business?
Want to see where you stand?
A free AI visibility audit, run by hand across all six platforms, with the answers quoted back to you.
Start With a Free Audit β