Classic search returns a ranked list; you read the pages and decide. AI search does the reading. The system runs one or more web searches for your question, fetches some of the pages, and writes an answer that names a few of them as sources. The products that do this are the answer engines: Perplexity, ChatGPT with search on, Google's AI Overviews and AI Mode, Gemini, and Claude with web search.
Why it changes the job
Ranking still matters, because the search behind the answer is a search. But ranking is no longer the result. A page can rank third and not be cited, and a page can be cited from page two if it answers the question more directly than the pages above it. The number of sources is small: in our check on 2026-10-21, ChatGPT cited 1.9 sources per answer and Perplexity 10.0. Everyone else in the results list is invisible.
Google's AI features documentation describes the eligibility side: a page can appear in AI Overviews and AI Mode once it is indexed and can be shown with a snippet, with no extra markup. OpenAI and Perplexity each run a crawler for their search index, OAI-SearchBot and PerplexityBot, and both say to allow it if you want to appear.
The three steps that decide it
- Being read: the crawlers can reach the page.
- Being retrieved: the page is in the index and matches the question.
- Being chosen: the page answers the question directly, with facts the model can quote and a source it can name.
The names people use for working on this, answer engine optimization, generative engine optimization, LLM optimization, all describe the third step. Our guide to generative engine optimization covers it in full.
