What GEO is, and whether you need to care
GEO stands for generative engine optimisation. It describes the work of being named when an AI assistant answers a customer’s question. This page explains what that actually involves, and how to tell the difference between a firm measuring it and a firm selling the acronym.
The short version
GEO is the work of making a business the one an AI assistant names when a customer asks it a question, rather than the one it leaves out.
That’s a different job from ranking on Google, because the assistant does the choosing rather than handing the customer a list. It names two or three businesses. Everyone else is absent — and absent in a way the customer never notices, because there’s no page two to click through to.
Why a new term appeared at all
Because the old one stopped covering the job.
Search engine optimisation grew up around a list of links. The work was to be high on the list. When a system answers directly instead, the question changes from where am I on the list to am I in the answer at all — and the inputs aren’t the same.
Whether that deserved a new acronym is arguable. What isn’t arguable is that the two can disagree: we’ve measured sites scoring well on conventional SEO that went unnamed in most of the assistant answers we tested.
What the work actually consists of
Three things, in roughly this order of neglect.
- Being readable. An assistant has to get through the page. Content that only appears after scripts run, or that is locked inside images, is content it may never see.
- Being checkable. Claims that exist only on your own website are weaker than claims corroborated somewhere else — a professional register, a college, a journal, a conference program.
- Being attributable. A named person with qualifications attached carries more weight than anonymous copy. The system is trying to work out who’s speaking.
For a regulated health service that list is convenient, because none of it is restricted. The material you aren’t allowed to use — testimonials, before-and-after photos, outcome claims — isn’t on it.
How to tell whether a firm is measuring GEO or selling it
One question does most of the work: can they show you the questions they will ask, and how they will score the answers?
- If the answer is a fixed list of questions, run on a stated schedule, scored on something specific — they’re measuring.
- If the answer is “AI-ready content” or “GEO included” with no question set, no schedule and no scoring — the acronym is doing the work a measurement should be doing.
A second question: do they treat “named in the answer” and “cited as a source” as the same thing? They aren’t. A site can be a listed source in half the answers and named in the text in a fraction of that.
Whether you need to care yet
Honestly: it depends on whether your customers ask assistants, and that varies by treatment and by age group more than by anything else.
What we wouldn’t do is treat this as urgent just because it’s new. The reason to look at it is that the work overlaps heavily with things that were worth doing anyway — a readable site, a named practitioner, corroboration outside your own website. If assistants turned out to matter less than expected, none of that would be wasted.
Common questions
- Is GEO different from AEO?
- They overlap and different firms draw the line differently. We use AEO for being the source behind a direct answer, and GEO for being named inside an answer written by a generative assistant. The distinction matters less than whether either is being measured.
- Do we need to do anything special for ChatGPT versus Gemini?
- They answer differently and we measure both, but the underlying work — readable, checkable, attributable — is the same.
- Can we pay to appear in AI answers?
- Not in the way you can pay for a search ad, and we don’t run paid advertising. Be careful with anyone who implies otherwise.
- How would we know if it’s working?
- By measuring the same questions the same way over time. A single check tells you nothing, because the answer to the same question changes from one day to the next.