Every AEO & GEO termthat matters in 2026.

Optimizing your content to be cited inside AI answer engines like ChatGPT, Perplexity, and Google AI Overviews.

The broader discipline of being retrievable, quotable, and trusted by generative AI engines.

The architecture every AI answer engine uses — retrieve documents, ground the generated answer in them.

Pricing model where users plug in their own provider API keys, the tool pays no per-token cost.

How many tokens of retrieved text an engine can feed its LLM at once — caps the number of pages it can cite per query.

Numerical vector representations of text that let engines compute semantic similarity — the math behind RAG.

The infrastructure that stores billions of document embeddings so engines can run nearest-neighbor search in milliseconds.

When a model is further trained on a specific corpus — distinct from RAG (which uses retrieval at inference time, not training).

The corpus a model was trained on — distinct from the live web it retrieves citations from at query time.

What the user actually wants from a query — informational, navigational, commercial, or transactional.

The composite answer an AI engine writes by weaving together retrieved sources — the unit of AEO output.

When an AI engine generates incorrect facts not supported by retrieved sources — risk grows when retrieval coverage is poor.

The writing pattern where you state the answer in the first 100 words — strongly correlated with AI citation.

A single URL that covers what/who/how/pricing for a topic — the format top-cited URLs share.

When an answer engine decomposes one user prompt into 3-5 sub-queries, then aggregates citations across all.

Querying the REAL answer engines (vs simulating retrieval logic) using BYOK API keys.

Ensemble retrieval that combines dense embeddings + sparse BM25 + authority signals — what every modern engine actually uses.

A second-stage scoring model that reorders the top 50 retrieval candidates before sending the top 5 to the synthesis model.

When an AI engine decides to invoke a web-search / web-fetch tool mid-conversation to ground its answer.

How engines split long pages into retrievable pieces — H2 boundaries are the dominant split signal.

The technique of embedding structural fields (headings, lists, schema) separately from body text — +17.3% citation lift per the arXiv paper.

Publishing comprehensive coverage of a topic so an engine cites YOU instead of the original primary source.

Using HTML elements for their intended meaning (<article>, <nav>, <aside>) — gives AI engines structural context.

Sending an empty HTML shell + JS bundle that renders content in the browser — invisible to most AI crawlers.

Server returns fully-rendered HTML per request — visible to all AI crawlers without JS execution.

Pre-rendering pages at build time to static HTML — fastest possible TTFB, perfect AI-crawlability.

Static pages that re-render on a schedule — best-of-both: fast like SSG, fresh like SSR.

Serving pre-rendered HTML to bots while serving SPA to humans — a transitional fix, not a long-term strategy.

Optimizing for non-English search/AI queries — massive underserved TAM where most AEO tools don't work.

Content with long-term relevance — outperforms time-sensitive content in cumulative AI citations.

The network of links between your own pages — engines use it to discover content + transfer authority within your site.

The visible clickable text of a link — provides semantic context engines use for retrieval + ranking.

Ranking individual page sections (passages) rather than whole pages — lets long pages rank for specific sub-questions.

A leading answer engine that generates responses with inline citations.

Google's AI-generated answer summary that appears above traditional search results on ~30% of queries.

The OpenAI crawler that powers ChatGPT's web search retrieval — different from GPTBot (training).

Anthropic's retrieval crawler for Claude's web tool — separate from ClaudeBot (training).

Perplexity's retrieval crawler — the one that fetches pages for citation in its answer panel.

The component of an AI answer engine that decides which web pages to read before generating an answer.

Crawlers used by AI coding agents (Cursor, Claude Code, Copilot) to fetch documentation on demand.

OpenAI's autonomous browser-agent that performs research tasks on the user's behalf.

OpenAI's TRAINING-data crawler — distinct from OAI-SearchBot (retrieval).

Google's TRAINING-data crawler for Gemini and Vertex AI — distinct from Googlebot (search).

Any system that synthesizes a written answer from retrieved sources instead of returning a list of links.

Google's full-page conversational answer surface — distinct from AI Overviews (which sit above the SERP).

When an agent decomposes a task into multiple search steps and acts on the results — the next generation past one-shot answer engines.

The component that decides which LLM (and which retrieval pipeline) handles each query — invisible to the user but consequential for AEO.

The number of pages a crawler will fetch from your site in a given time window — a hard ceiling on indexation depth.

The boxed direct answer at the top of Google search — the predecessor to AI Overviews.

Google's expandable related-questions section — a goldmine for sub-query targeting.

Google's branded sidebar with structured facts about a person/place/brand — the visible output of the Knowledge Graph.

Searches that resolve entirely on the SERP/answer page — share rising past 60% as AI Overviews expand.

The schema.org type that signals your page contains questions + answers — the most-cited schema by AI Overviews.

The JSON-based serialization for embedding schema.org structured data in your HTML.

The shared vocabulary maintained by Google, Microsoft, Yahoo, and Yandex for structured data on the web.

Schema.org markup (typically JSON-LD) that gives engines machine-readable context about page content.

Google's name for the enhanced SERP cards (FAQ accordions, recipe stars, product prices) that schema.org markup unlocks.

The single preferred URL for a page, declared via <link rel=canonical> — prevents AI engines from splitting citations across duplicates.

Link attribute telling crawlers not to pass authority through this link — used for sponsored/user-generated links.

Meta tag or HTTP header telling crawlers not to add this page to their index.

HTTP header equivalent of <meta name=robots> — controls crawler behavior on non-HTML files (PDFs, images).

Link tag declaring language/region variants of a page — tells engines which version to serve for which locale.

SERP enhancements (sitelinks, prices, ratings) generated from structured data — increase visual surface area on SERPs.

How a URL renders when pasted in Discord — derived from OG + Twitter card tags, viewed by 200M+ monthly users.

The visual preview generated when a URL is shared on social platforms — driven by og:image + og:title.

Meta tags that constrain how engines may snippet your content: max-snippet, max-image-preview, max-video-preview.

The meta tag standard for declaring how URLs render when shared — og:title, og:description, og:image.

Twitter/X's meta tag standard for share-card rendering — distinct from Open Graph, takes precedence on X.

Google's Experience-Expertise-Authoritativeness-Trustworthiness framework — directly inputs to AEO citation likelihood.

The knowledge-graph identifier that signals your brand is a recognized entity — highest-weight EEAT signal.

A standardized text file at /llms.txt telling AI agents which pages summarize your site — adopted by ~10% of domains in 2026.

The structured entity-relationship graph (Wikidata, Google KG) AI engines use to verify brands and concepts.

The /robots.txt file telling crawlers what they can fetch — the 2026 retrieval-vs-training distinction matters most.

Google's category label for pages affecting health/finance/safety — held to 2-3x stricter E-E-A-T scrutiny.

An author whose credentials + visible track record qualify the page for E-E-A-T-sensitive citations.

The 'By [Author Name]' attribution at the top of a page — the cheapest E-E-A-T signal to add.

A dedicated page at /authors/[name] with credentials, prior work, and external profile links.

The browser's parsed semantic structure of a page — what screen readers AND many AI crawlers consume.

Roles that define page regions (banner, main, navigation, contentinfo) — give crawlers + screen readers structural waypoints.

Google + AI engines crawl the mobile version of your site first — desktop-only or mobile-broken content gets ignored.

Serving identical content to bots and humans — engines penalize sites that show bots one version and users another.

Adjusting the AEO scoring weights for industry-specific signal importance — healthcare prioritizes E-E-A-T 3x harder than e-commerce.

The percentage of AI-generated answers in your category that cite YOU vs. competitors.

Three-level page architecture (macro/meso/micro) validated to lift citation rate +17.3%.

Percentage of cited URLs in your category that are yours, vs competitors.

Whether your page has 4+ H2 sections covering the buyer query's subtopics — the dominant chunking signal.

How close two pieces of text are in embedding space — the cosine distance between their vectors.

Google's three loading/interactivity/visual-stability metrics: LCP, INP, CLS — table-stakes for AI Overview eligibility.

Time until the largest above-the-fold element renders — under 2.5s is good, over 4s fails Core Web Vitals.

Sum of unexpected layout shifts during page lifetime — under 0.1 is good.

How responsive a page feels after the user clicks/taps — replaces FID as the Core Web Vitals interactivity metric.

Time from request start to the first byte of response — the floor for every other Core Web Vital.

How comprehensively a site covers a topic cluster — a stronger ranking signal than backlinks for AI citation.

The rate at which a site publishes new content — moderate consistency beats sporadic high-volume bursts.

The gradual loss of traffic + citations as content ages without updates.

A page with zero internal links pointing to it — discoverable only via sitemap, often ignored by AI crawlers.

Number of clicks from the homepage to a given page — pages deeper than 3 clicks see degraded crawl frequency.

Percentage of queries where an AI engine generates an answer instead of returning a SERP — AI Overviews now hit ~50%+ for common informational queries.

Percentage of generated answers (per query) that include a citation to your site — the AEO equivalent of SERP CTR.

Composite metric combining cite rate, position, and query coverage — the AEO industry-standard KPI.

Your rank among the set of pages an AI engine considered for citation — usually the top 50 retrieved, not the full web.

Citation share broken down by AI engine — ChatGPT, Claude, Perplexity, Gemini, AI Overviews don't behave identically.

The time period after which content is treated as stale — varies by topic: 30 days for news, 12 months for evergreen guides.

How tightly a generated answer's claims map to retrieved source content — high groundedness = low hallucination risk.