Answer Engines: What AEO and GEO Actually Change

The reader is no longer only a person

Search engines list pages. Answer engines read pages and write an answer, often without the visitor ever opening your site. Whether your company is described correctly in that answer depends on how readable and consistent your site is to a machine. That is the whole subject of answer-engine optimisation (AEO) and generative-engine optimisation (GEO).

Start with pages a crawler can read

If the content only exists after JavaScript runs, many crawlers see an empty page. Render the page's real content into the HTML that is served, so the first response already contains the headings, the text, and the links.

Say who you are in one consistent place

Answer engines combine many sources. If your name, location, services and contact details differ between your site, your profiles and your listings, the summary will be a blend of them. Pick one wording and use it everywhere, starting with a clear description of what the company does on the home page.

Structured data must match the visible page

JSON-LD helps a machine read a page, but only when it describes what a visitor can also see. Questions and answers in structured data should be the same ones printed on the page. Data that claims more than the page does is a reason to be ignored.

Write answers, not slogans

A page that opens with the direct answer to a question, then explains it, is easy to quote. Use headings that are real questions or clear statements, keep each section to one idea, and avoid claims you cannot support, because an engine will repeat them.

Give crawlers and models a summary

A plain-text file such as llms.txt at the site root can summarise the company and link to the key pages for language models. It is a convenience, not a ranking switch, and it is only useful if it is accurate. Keep robots rules explicit about which crawlers are welcome.

On our own site

This site renders every public page to static HTML at build time in both languages, generates its structured data from the same content shown on the page, lists the search and answer-engine crawlers it welcomes in robots.txt, and publishes an llms.txt summary. None of this promises a result: it removes reasons for a machine to misread us.