What are AEO and GEO, and is either one real?
Answer engine optimisation, generative engine optimisation, AI SEO. Three names for largely the same work, one of which Google says is just SEO. What the terms mean, where the discipline genuinely differs, and where it does not.
The vocabulary arrived faster than the evidence. AEO, GEO and AI SEO are used interchangeably by most people selling them, including by people who will tell you they are quite different. Here is what each is supposed to mean, and what survives scrutiny.
The definitions
| Term | Stands for | Usually means |
|---|---|---|
| AEO | Answer engine optimisation | Work aimed at being named or cited inside a single AI answer. Framed around the individual response. |
| GEO | Generative engine optimisation | The broader version: how a brand appears across AI-generated answers generally, including whether it is described accurately and whether competitors are recommended instead. |
| AI SEO | — | Usually a synonym for either, sometimes used to mean ordinary SEO with AI features mentioned. |
In practice the distinction between AEO and GEO is not commercially meaningful. Almost nobody sells them as separate products, and the deliverables are the same. If a provider draws a sharp line between them, ask what changes in the work.
What Google says about the whole idea
“Optimizing for generative AI search is optimizing for the search experience, and thus still SEO.”
Google discourages the AEO and GEO framing for its own platform. A Forrester analyst has put it more bluntly, describing AEO as significantly but not fundamentally different from SEO, and saying proponents tend to exaggerate the differences to carve out a new budget line.
That should be the starting position, and any honest provider in this category should be able to say it out loud before explaining where they think it is incomplete.
Strip the jargon and what is left?
Take any AEO or GEO deliverable list and remove the vocabulary. What remains is: be crawlable, render server-side, publish clear factual content with headings, keep it fresh, earn third-party mentions, keep your listings accurate.
That is technical SEO plus digital PR. It is not nothing, and it is not new. And the items that genuinely are new to this discipline — llms.txt, vector embedding optimisation, speakable schema — are precisely the ones with no supporting evidence.
Where it genuinely does differ
Three things are real, and they are the honest reason the work is worth anything.
The gates are different
A business can be perfectly healthy in Google and structurally invisible to ChatGPT, because ChatGPT discovers URLs through Bing rather than Google. Blocking OAI-SearchBot in robots.txt removes you from citation eligibility while leaving your Google rankings untouched. Those failure modes do not exist in classic SEO.
The signal mix is different
Unlinked mentions of a brand correlate with AI visibility about three times more strongly than backlinks. Reddit and YouTube carry weight well beyond their classic search role. And the share of Google AI Overview citations coming from the organic top ten fell from about 76% to about 38% between mid-2025 and 2026 — ranking and being cited have measurably come apart.
The engines disagree with each other
In our own measurement across 29 answers, the five major assistants cited 204 distinct domains and not one was cited by all five. ChatGPT and Perplexity shared no sources at all. There is no single 'AI search' to optimise for, which is a genuine structural difference from optimising for one dominant search engine.
So is it real?
The gates and the measurement are real. The growth claims are weaker than they are sold as, and the flagship technical tactic failed its only rigorous test. The honest description is a narrow discipline with a few verifiable levers, sold with the confidence of a mature one.
What follows from that is mostly about pricing and promises rather than about whether to bother. Published UK prices sit at parity with ordinary SEO, and at the low end below it, which is roughly the right position for work that overlaps substantially with SEO and adds a handful of genuinely new checks.
Words to be suspicious of
- 'Ranking' in an AI assistant. There is no stable ranking. Asking an identical question repeatedly returns the same list of brands under 1% of the time.
- 'AI visibility score'. Composite figures with no stated construction, over a sample far too small to support the decimal places shown.
- 'Getting into the training data'. Unverifiable by construction — training composition is opaque, so the claim cannot be checked either way.
- 'Vector embedding optimisation'. The actionable advice underneath is ordinary topical authority.
- 'AI-first content'. Usually means the same content with more headings.
Common questions
- What is the difference between AEO and GEO?
- AEO frames the work around being named or cited in an individual AI answer; GEO frames it around how a brand appears across AI answers generally. In practice the deliverables are the same and almost nobody sells them separately, so the distinction is rarely commercially meaningful.
- Is AEO just SEO with a new name?
- Substantially, and Google says so directly — its documentation describes optimising for generative AI search as still SEO. What genuinely differs is that the discovery gates are different, unlinked mentions outweigh backlinks by roughly three to one, and the assistants run on different indexes so there is no single thing to optimise for.
- Is AEO worth paying for?
- The verifiable parts are worth having: the crawler and indexation checks, an accurate Google Business Profile, and honest measurement across several assistants. The growth claims are weaker than they are marketed as, and published UK prices sit at parity with ordinary SEO, so treat a large premium as a claim that needs justifying.
Sources
Every claim above should be checkable. Where a study has limits, they are stated rather than left out.
- Google Search Central — generative AI search guidanceGrade A. Provider documentation.
- Forrester — analyst commentary on AEO versus SEOGrade B. Named analyst position.
- Ahrefs — 75,000-brand correlation study; AI Overview citation source share over timeGrade B. Correlational and observational.
- SparkToro — 2,961 prompt runs, reproducibility of brand listsGrade A.
- Sentinay — original measurement, 29 answers across five assistants, 17 August 2026Grade B. Our own single run.
More on choosing an aeo agency
- How to choose an AEO agency in the UK
What the published UK prices actually are, the strongest case against this whole category, the three places that case fails, and the questions that separate real work from vocabulary.
Want to know where you stand?
We ask five AI assistants for a company like yours and send you a free report showing how often you came up and who came up instead.
Get my free report