Guides to answer engine optimisation

Answer engine optimisation, or AEO, is the work of becoming a useful source for the answers people receive from ChatGPT, Google AI Overviews, Gemini, Claude, Perplexity, and other search-connected assistants. It builds on ordinary search fundamentals but measures a different outcome. A page can rank, be technically readable, and still be absent when a buyer asks for a recommendation. These guides separate the work from the evidence so technical readiness is never presented as proof of a citation.

Use three evidence states when reading any recommendation here. Eligible means a named crawler can access the intended page and the delivered HTML contains the information it needs. Competitive means the site has a credible page or external source for the buyer question, with clear claims and support that can stand beside the pages already being cited. Observed means a live answer was captured with its question, provider, surface, cited URLs, and timestamp. These states are related, but they are not a blended score and one never substitutes for another.

The library is organised around four reader jobs. Learn explains AEO, GEO, AI visibility, how answer retrieval works, and how to evaluate tools or costs.Measure covers crawler access, readable content, provider surfaces, citations, competitors, and the evidence needed for a baseline. Act turns a diagnosed constraint into a prioritised technical, content, or trust change.Prove covers comparable checks after publication, monitoring over time, and what a result can honestly support without claiming causation.

Start with the stage matching the constraint you can see now, not the tactic you expect to use. If the site has never been measured, establish access and one live answer first. If competitors are already being recommended, compare the exact cited pages before creating more content. If a change has shipped, preserve the baseline and repeat the same question on the same surface. Each guide states its limits, sources, and next step so you can move through the library without turning a sample into a trend or a correlation into proof.