Definition
An answer engine is a search system that answers a question directly with a generated, coherent answer instead of a list of links. ChatGPT with search, Perplexity, Google AI Mode and AI Overviews, Microsoft Copilot and Claude with web access are answer engines: they retrieve sources, have a language model write the answer and show the sources as citations. The user reads the answer and at most clicks through for depth.
The difference from a classic search engine lies in the output and in user behaviour. A search engine gives ten options and lets the user choose; an answer engine chooses itself, combines several sources and presents one outcome. Questions become longer and more conversational ('which SEO tool suits a Shopify store with 400 products and no marketing team'), users ask follow-up questions in the same session, and the number of sources that come into view is small: usually three to eight per answer.
Technically, answer engines work as RAG systems. A question is split into sub-questions (query fan-out), pages or passages are retrieved for each sub-question from an own index or a partner index (Bing for ChatGPT and Copilot, Google's index for Gemini and AI Mode, an own crawler for Perplexity), and the model writes an answer from them. Which pages get retrieved depends on crawlability, speed, relevance and trustworthiness; which ones end up cited depends on how concretely and independently the passage answers the question.
Answer engine optimization (AEO) and generative engine optimization (GEO) are the disciplines that respond to this. The core: pages that answer questions directly in the first paragraph, independently quotable sections, entities named explicitly, evidence with sources, a recognisable author, structured data, server-rendered HTML and no blocks on AI crawlers. Beyond that, the brand picture the model has from external sources counts: mentions, reviews and comparisons determine whether you are named for a recommendation question.
Measurement works differently than in Google: there is no position, but a percentage of answers in which you are mentioned or cited, per engine and per topic, obtained through prompt sampling. Traze tracks that per engine, links every citation to the page that earned it and writes and rewrites pages specifically to be chosen as a source more often.