What is generative engine optimisation?

Generative engine optimisation, or GEO, is the practice of making a company findable, understandable and worth naming in the answers AI systems assemble when buyers research who to buy from.

It is not a separate discipline, and the people selling it as one are overstating their case. Google's own position is that optimising for generative AI search is optimising for the search experience, and is therefore still SEO. Its generative features are built on the same core ranking and quality systems and draw from the same index.

What is genuinely different is the outcome you are aiming at and the way you measure it. There is no position eleven in an assembled answer, and no second page. You are named or you are not — and nothing in a rankings report tells you which.

Source: Optimizing your website for generative AI features on Google Search, Google Search Central.

Where the term came from

From the market, not from the search engines. Nobody at Google or OpenAI announced a discipline called GEO. It emerged because AI answers started changing how buyers arrive, agencies needed a name for the work, and a new name sells more easily than an old one.

That does not make it meaningless. It describes something real: a shift in where the decision gets made. It does mean you should treat the acronym as shorthand rather than as evidence that a separate skill set exists.

So is it just SEO with a new name?

Mostly, and the honest answer has two halves.

The foundations are identical. A page that cannot be crawled, indexed or understood cannot be surfaced by anything. Covering a subject properly, answering real questions, being described consistently and loading correctly on a phone were good practice before any of this and remain the precondition for all of it.

The work overlaps almost entirely. The measurement does not overlap at all.

What changes is what you are optimising toward. Ranking asks where your page sits in a list. This asks whether your company is named in an answer a buyer reads instead of that list. The two can move in opposite directions, and only one of them is visible in your current reporting. The three aims are compared properly in GEO vs AEO vs SEO.

What the work actually involves

Five things, none of them exotic.

Questions rather than keywords. The unit of work is a question a buyer would actually type — "what does a palletiser cost to install on an existing line" — not the two-word term that describes your category. Answers get assembled from pages that answer questions.

The answer early on the page. A passage that makes sense lifted out of its page is a passage that can be quoted. One that builds to its point over eight paragraphs is not.

Coverage rather than one good page. Being relevant to a subject rather than to a single question is topical authority, defined on Build.

One consistent description of the company, stated the same way on your own site, in directories, in listings and in structured data. Contradiction is worse than silence, because it gives a system nothing settled to work from.

Corroboration from outside your website. Your own site establishes what you say about yourself. Independent sources are what turn that into something confirmed. This is the slowest part and the part that changes the most.

What it is not

It is not a markup trick. No schema makes a page eligible for an AI answer, and Google says so plainly in the same guidance quoted above.

It is not a fast channel. The work takes four to six months to show in the answers, and anyone offering a quicker route is describing something they do not control — the phases are set out in how long this takes to work.

And it is not a tool you buy. Tools that watch AI answers exist and are useful, but a dashboard does not write the pages, correct the listings or earn the citations.

Why we do not sell ourselves as a GEO agency

Because GEO is a method, and a method is a poor thing to buy.

What a manufacturer actually needs to know is narrower and more useful: when a buyer asks an AI who supplies equipment like yours, does your name come back, how often, alongside whom, and described correctly? That is measurable. The work that follows is GEO, along with several other things, and the label matters far less than the number.

It is also why the first thing we sell is a measurement rather than a retainer. You should not buy a method before you know whether you have the problem it treats.

How would you know whether it is working?

By asking the same questions again and counting.

Fix the wording of a set of buyer questions, put them to the systems on a schedule, and record which companies are named each time. Movement means the work is landing. Nothing for three months means it is early. Nothing at month six means something is wrong, and you will have the record to work out what.

Traffic is not the measure here, and neither is rank. Both will tell you a story about a different question.

What this article cannot tell you

It cannot tell you that GEO will work for your company. It describes a discipline, and disciplines have conditions — a market where buyers research before they contact anyone, a product with real technical answers behind it, and enough time for the work to compound.

It cannot give you a settled definition either. The term is only a few years old, used differently by different firms, and not defined by the companies whose systems it aims at. We have given ours plainly; treat anyone's definition, ours included, as a description rather than a standard.

And AI systems are probabilistic and change continuously. What works is judged by measurement over time rather than by argument, which is the whole reason we measure before we advise.

What to do next

Decide whether the problem exists before deciding what to call the solution. Thirty questions your buyers actually ask, put to the six systems in a logged-out session, with every named company written down, will tell you more than any definition of GEO can.

You can do that yourself in an afternoon, with no tool and no budget. If you would rather have it measured properly, against the competitors you name and across all six systems, that is the AI Visibility Audit.