Say it precisely.

Answer engine optimization: making a page easy for an answer engine to extract a direct, self-contained answer from. It is about the page, not about predicting which answer an assistant gives.

Google’s conversational search experience. Like AI Overviews, it draws on pages indexed for Google Search.

Google’s AI-generated summaries shown within Google Search results. A page must be indexed and eligible for a snippet, and the site must be included in Search generative AI features in Search Console, which is the default.

The first passage under a heading, written so that it answers that heading on its own. Our working range is 40–60 words; no engine publishes a target.

A system that responds to a question with an answer rather than a list of links, often citing the pages it drew from. Assistants with web search and AI summaries in search results both behave this way.

Whether a page states a direct, extractable answer to the question it targets. One of the signals in GEOBee’s AEO lens.

Repeated page furniture such as navigation, footers, cookie banners and sidebars. Extraction tries to remove it; what it misses dilutes the content.

The URL you declare as the preferred version of a page with rel="canonical". Conflicting canonicals let a different URL take the credit.

Splitting a document into passages before indexing, so retrieval can return a section rather than a whole page. Passages that depend on earlier text lose meaning when split.

How specific, sourced and quotable a page’s claims are: named sources, dated facts, figures with context, sentences that make sense when quoted alone.

A distinct thing a system can recognize and connect facts to: an organization, a person, a product, a place.

Turning a fetched HTML page into the text a system actually reads, usually by discarding navigation, footers and other page furniture.

A system that writes a response with a language model, usually grounded in documents retrieved for that query.

Generative engine optimization: making a page discoverable, retrievable and worth referencing for systems that generate answers from retrieved sources. The term comes from a 2023 research paper of the same name.

A robots.txt control token, not a crawler, that governs whether content may be used for Gemini training and grounding. It does not affect Google Search.

Connecting a model’s answer to retrieved sources so it can be checked and cited, rather than relying only on what the model learned in training.

The JSON format for linked data that search engines recommend for structured data. It sits in a script tag and does not change what readers see.

A proposed markdown file at a site root that points language models to a site’s most useful pages. It is a proposal, not an access control.

GEOBee’s state for redirects, errors and non-HTML documents. They show a dash rather than a failing zero, so averages stay truthful.

Monitoring how assistants answer a fixed set of prompts over time. Useful as a trend signal; answers vary between runs, so single readings are weak evidence.

Google’s name for issuing several related searches behind one question in its AI search experiences, then combining the results.

RAG: a pattern where a model is given documents retrieved for the current query and writes its answer from them. Search-backed assistants work this way.

A plain-text file at a host’s root that tells crawlers which paths they may fetch, standardized as RFC 9309. Each host and subdomain has its own.

A schema.org property listing other URLs that identify the same entity, such as official profiles, helping systems connect your site to that entity.

A crawler that builds the index an assistant retrieves from when it searches. These decide whether you can be found.

Search engine optimization: the classic discipline of making pages crawlable, indexable and competitive in ranked search results. AEO and GEO build on it rather than replace it.

Rules such as nosnippet, max-snippet and data-nosnippet that limit what text a search engine may show from a page. On Google they also limit use in AI features.

A page that returns HTTP 200 but has no real content, such as an empty search page or a "not found" message. It looks valid to a crawler and is worthless to a reader.

Machine-readable facts about a page, usually JSON-LD using the schema.org vocabulary: what the page is, who made it, what it offers.

A crawler that collects content which may be used to train future models. Blocking one does not remove you from search products that use a separate crawler.

An agent that fetches one URL because a person asked an assistant about it. Some vendors document that robots.txt may not apply to these.