AEO vs GEO vs SEO

SEO, search engine optimisation, is the old contract: Google shows ten links, you fight for the top of the list, the user chooses. Rank is the metric, the click is the prize.

AEO, answer engine optimisation, targets a different surface: ChatGPT, Perplexity, Gemini and Google's AI results do not show a list, they write one answer. There is no position two. You are named in the answer, or you are absent, and one of your competitors is the recommendation.

GEO, generative engine optimisation, is the same goal under the name a 2023 research paper gave it. If a tool sells you GEO and another sells you AEO, they are selling the same work.

The unit of competition changes. A results page has room for everyone on page one; an answer has room for whoever the model names, which is usually a handful of brands and a handful of cited sources. Being tenth stops being a consolation prize and becomes invisibility.

The evidence changes. A ranking is stable enough to check on Monday and trust on Friday. A generated answer is sampled from a model, so the same question returns different answers on different days. Any serious measurement has to ask repeatedly and report how many answers a number is based on, which is why every metric in Aureo carries its n.

The lever changes. In SEO you optimise your own page. In AEO the answer often leans on third party sources: review sites, comparison articles, forums. If the pages an assistant cites do not mention you, your own site can be perfect and the answer will still send buyers elsewhere. The work is knowing which sources the engines lean on and getting onto them.

The plumbing is shared. Assistants retrieve pages before they write, so everything technical SEO always asked for still applies: a crawlable site, real text that survives without JavaScript, structured data, a canonical URL, a sitemap. Two things are new on top: AI crawlers have their own user agents that your robots.txt and CDN must actually admit, and a page only gets quoted if it contains a sentence worth quoting. You can test the readable half of this in a minute with our free check, fifteen checks, no account.

First, make yourself readable: pass the technical checks above, on your own domain. Second, find out what the answers say today: ask the questions your buyers ask, across engines, more than once, and write down who is named and which sources are cited. Third, close the gaps that measurement names: publish the page that answers the question directly, and get onto the third party pages the answers already lean on. Then measure again, because an answer you fixed in March can quietly change in April.

LLM SEO: how assistants pick their sources covers the retrieval side in more depth, and the sample report shows what a weekly measurement of all of this looks like on a real brand.

• Are AEO and GEO the same thing?
  Near enough that arguing the difference is a waste of a meeting. Both mean optimising to be named and cited inside a generated answer. GEO comes from a 2023 research paper on generative engines, AEO grew out of the older answer box vocabulary. Vendors pick whichever sells; the work underneath is the same.
• Does good SEO automatically give me good AEO?
  No, but bad SEO almost guarantees bad AEO. Assistants retrieve pages before they write, so being crawlable, fast and well structured still matters. What SEO does not give you is being quotable: an answer engine needs a sentence it can lift, a claim it can attribute, and a page that answers the question directly.
• Should I stop doing SEO?
  No. People still search, and the engines that write answers read the same web that Google ranks. The honest framing is a shift of budget, not a replacement: keep the technical base, and spend the content effort on questions people ask assistants, not just keywords they type.
• How do I measure AEO?
  Ask the assistants the questions your buyers ask, repeatedly, and count. One answer proves nothing: the same question to the same engine returns different answers on different days. A useful measurement samples each question several times per engine, reports how often you are named, and says how many answers that count is based on.