01Notes · 21 August 2026
Nine Names for Four Ideas: the AI search vocabulary problem
Quick answer
AEO, GEO, LLMO, LLM SEO, AI visibility, AI answer optimisation, AI citation optimisation, AI discovery optimisation, generative search optimisation. Nine names in current circulation.
Underneath them are four genuinely distinct ideas. The gap between nine and four is where a lot of budget goes.
Key takeaways
- Four real ideas: being retrieved, being used in an answer, being described accurately, and being measured across platforms.
- AEO and GEO overlap by roughly 80%. Sold separately, they bill twice for one body of work.
- LLMO and LLM SEO are the same thing again. No separate scope should exist for them.
- A new name is the cheapest way to look specialist. Naming is faster than being good.
- Ask what a term contains that another does not. If the answer is vague, it is a label, not a discipline.
- Build the programme around outcomes, because half these words will not survive two years.
The four ideas that are actually distinct
Strip the vocabulary away and four separate things are happening. They need different work, and confusing them is what makes scoping conversations circular.
One: being retrieved. Can a system find and read your page at all? Crawler access, rendering, indexation. Unglamorous, and the most common blocker we find.
Two: being used. Once retrieved, does the system select your page as a source? Answer clarity, query breadth, factual specificity.
Three: being described accurately. When you are named, is what is said about you correct? This is entity work, and it is the one most often skipped.
Four: being measured. Knowing any of the above is true, per platform, over time. Genuinely new, and the only one where the old toolkit gives you nothing.
Every term in circulation is describing some combination of those four.
| Term | Usually means | A distinct discipline? |
|---|---|---|
| SEO | Retrieved and used, on search engines | Yes, and it still does most of the work |
| AEO | Used inside an assistant | Partly. ~80% shared with GEO |
| GEO | Used in AI summaries inside search | Partly. Same 80% |
| LLMO / LLM SEO | Whatever AEO means | No |
| AI answer optimisation | Whatever AEO means | No |
| AI citation optimisation | Being used, again | No |
| AI discovery optimisation | Usually all of it, vaguely | No |
| Entity optimisation | Being described accurately | Yes. Different work entirely |
| AI visibility measurement | Knowing, per platform, over time | Yes, and it is the genuinely new one |
Four rows say yes. Five say no. A proposal with five line items drawn from the "no" rows is one body of work billed five times.
Why the names keep multiplying
Not conspiracy. Incentive.
A new category name is the cheapest way to look like a specialist. Coining "AI discovery optimisation" takes an afternoon. Becoming genuinely good at measuring citation behaviour across five platforms takes months and produces a service that is harder to explain.
And buyers cannot easily tell the difference, because the vocabulary is new enough that nobody is confident challenging it. That is a temporary condition, and it is worth using while it lasts — in your favour, by asking harder questions.
One question to ask a proposal
One question, and it works on any of the nine terms.
"What does this line item contain that the one above it does not?"
A real distinction produces a specific answer: different sampling, different deliverable, different platform. A label produces a paragraph about how AI is changing everything.
We hold this position at cost to ourselves. It removes several billable lines from our own proposals, and the overlap table on our services page tells you exactly which of our disciplines are 80% the same thing.
The second useful question is about durability. If GEO stops being the fashionable word in eighteen months, does the programme still make sense? A programme built around being cited, accurately, on the platforms your buyers use survives any renaming. One built around an acronym does not.
What this means for you
- Count the distinct disciplines in your proposal. If more than four, ask what separates them.
- Treat AEO and GEO as one scope unless someone can show you what differs beyond measurement.
- Write your scope in outcomes, not acronyms. Vocabulary changes; being cited does not.
02Common questions
Questions this raises
Is GEO different from AEO?
Barely. AEO usually means being used inside an assistant; GEO usually means appearing in AI summaries inside a search engine. The optimisation overlaps by roughly 80%. Only the measurement genuinely differs.
What about LLMO or LLM SEO?
Another name for the same body of work. If a proposal treats it as a separate discipline with its own budget line, ask what it contains that AEO does not.
So the terms are meaningless?
Not meaningless — imprecise. There are real distinctions underneath, just fewer of them than there are names. Four ideas, by our count.
Why do so many terms exist?
Because a new category name is the cheapest way to look like a specialist. Naming something is faster than being good at it, and buyers cannot easily tell the difference.
Which term should we use internally?
Whichever your team already understands. What matters is that your scope document says what work happens, not which acronym it sits under.
Will these terms still exist in two years?
Some will not, and that is a reason to build a programme around measurable outcomes rather than around a vocabulary that may not survive.
Sources
Every figure on this page comes from one of the following. Where a claim is ours rather than a source’s, it carries an evidence label instead.
- Rankability — State of AI Search, 1,645 citation observations across 31 topics, 2026. Source for the finding that no schema type showed a consistent independent citation advantage, and for the first-100-words and query-breadth findings. Independent study
- Google Search Central — Optimizing your website for generative AI features (15 May 2026, updated 10 July 2026), structured data documentation, and the 2026 deprecation of FAQ rich results. Primary
- Searlixa sampling across client engagements, labelled Observed with sample and date. Our own
Ask an assistant
Hand this page to an assistant and ask it to summarise or challenge the argument. If it gets something wrong about us, that is itself a useful finding.
→Next step