AI visibility glossary
AI visibility has its own language. This glossary explains every term in plain English — what it means, why it matters, and what you can do about it.
Use it to understand why AI skips some sites and cites others — and to get your team speaking the same language when it comes to getting found.
01
Fundamental terms describing the transition from traditional search to AI synthesis and agent-driven discovery.
Optimizing pages and site signals so AI systems can reliably retrieve, understand, and cite your content in generated answers.
Strategies to earn visibility inside generative engines (chat assistants and answer engines) where output is synthesized, not ranked links.
Optimization focused on inclusion, attribution, and accuracy inside systems that produce direct answers from multiple sources.
A system that reads and fuses information from several sources to produce a single response, often with citations.
Search experiences where the primary interface is a generated summary/answer rather than a list of blue links.
Traffic or leads originating from users who found you via an AI assistant recommendation.
The share of relevant assistant answers where your domain is cited as a source.
How often your brand/product is named in answers, regardless of whether a clickable citation is provided.
A journey where the user gets an answer without visiting a site; success is measured by being cited and trusted, not clicks.
A response that attempts to fully satisfy the query in one output, reducing the need for follow-up browsing.
The process of combining information from multiple documents into a single coherent response.
Explicit linking or referencing of the sources used to support a generated answer.
An answer constrained by retrieved sources so claims can be traced back to evidence.
A cue that increases the likelihood a model will use/cite a page (clarity, provenance, structured data, consistent identity).
How up-to-date the content is relative to the query; critical for time-sensitive or rapidly changing topics.
The underlying goal behind a query (learn, compare, buy, troubleshoot), used to decide which content format best satisfies it.
A natural-language question with context and follow-ups, typical of chat and assistant interfaces.
Workflows where an AI agent plans multiple steps (search, read, compare, act) to solve a task.
A question that requires combining multiple facts or steps of reasoning across sources.
Biases in how a system selects sources (e.g., clarity, authority, format, accessibility, and structured signals).
Any interface where answers are consumed (chat, overviews, voice, browser sidebars), each with different citation and formatting constraints.
02
Classic search terminology that still matters—especially for crawlability, indexing, and trustworthy site architecture.
Practices that improve visibility in traditional search engines by aligning content, structure, and authority signals with ranking systems.
The results page shown by a search engine, including organic results, ads, and rich features.
A word or phrase representing a search demand; often used to plan content and measure search performance.
Why a user searches (informational, navigational, commercial, transactional) and what content format best satisfies it.
Optimizations on the page itself: titles, headings, content clarity, internal links, and structured markup.
The HTML title used for page identification in browsers and search previews; also a strong summarization cue.
A short summary used in previews; not a direct ranking factor in many systems but influences click behavior and context cues.
Use of H1–H6 to structure content; helps both humans and machines map sections and extract answers.
Links between your pages that help crawlers discover content and help models navigate topic relationships.
Clickable text of a link; provides topical context about the target page.
The practical limit of how many pages a crawler will fetch from a site within a time window.
A declared primary URL for a piece of content to reduce duplication and consolidate signals.
Substantially similar content across multiple URLs; can dilute signals and confuse retrieval.
Enhanced search presentation driven by structured data (e.g., product, FAQ, review).
A prominent excerpt intended to answer a query directly; conceptually similar to answer-engine extraction.
Clicks divided by impressions; a key performance metric for classic search surfaces.
How often a page appears in results for a query, regardless of clicks.
Search systems rewriting or broadening a query to retrieve more relevant documents.
Perceived expertise in a topic area, built through comprehensive, internally consistent coverage.
03
Terms used to plan content and site signals for assistant-led discovery, citations, and generated answers.
A broad umbrella term for optimizing for AI-influenced discovery across assistants, overviews, and chat-based search.
Using AI tools to accelerate SEO work (research, drafts, clustering) while still optimizing for human and crawler requirements.
Optimizations tailored to how LLMs retrieve, summarize, and cite sources, including format, clarity, and entity definition.
Content designed so key facts are easy to extract and cite (definitions, tables, concise claims with evidence).
Ensuring assistants describe your product the way you intend by providing canonical language and unambiguous definitions.
Pairing each important claim with supporting proof (data, references, policies) that a model can cite.
A page that contrasts alternatives using consistent criteria, useful for best X and X vs Y queries.
A page that frames competitor options and differentiators, often used in commercial investigation queries.
A page centered on a specific job-to-be-done so assistants can match you to intent-driven prompts.
Stating the what it is and who it's for early to prevent wrong summaries from limited context.
Structuring navigation and pages around core entities (product, features, industries) to improve retrieval and disambiguation.
The real user questions asked in assistants, which may differ from traditional keyword phrasing.
Choosing the format that best matches common assistant outputs (lists, steps, tables, TL;DR, FAQs).
A short, self-contained statement that remains correct when quoted out of context.
Controlling what shows when the brand is searched so assistants pull consistent facts (profiles, docs, reviews).
A cluster of interconnected pages that cover a topic comprehensively, signaling expertise and improving navigation for agents.
New, unique value your page adds beyond common summaries; increases selection when many sources look similar.
Evidence originating from you (data, docs, changelogs) that supports claims and improves trust.
A schedule for updating key pages so assistants retrieve current facts (pricing, features, policies).
When assistants cite third-party pages about you instead of your canonical pages due to better structure or trust signals.
The practice of maintaining canonical pages that assistants consistently use as the source of truth for core facts.
04
How AI systems fetch, rank, and compress information—useful for designing pages that survive retrieval and summarization.
A pattern where a model retrieves documents at query-time and generates an answer grounded in those sources.
A store optimized for similarity search over embeddings.
The length of each passage; too small loses context, too large wastes limited context space.
Ordering retrieved chunks by relevance before sending them to a model.
A second-stage model that refines which passages are most useful after initial retrieval.
Combining lexical retrieval (keyword) with semantic retrieval (embeddings) to improve recall and precision.
Embedding-based retrieval that matches meaning rather than exact keywords.
Generated content that is not supported by evidence; often triggered by ambiguity or missing sources.
The limited amount of text the model can process at once; influences how much of your page is actually read.
Selecting and formatting the best passages to fit within the model's context window.
The maximum amount of text a model can attend to in a single response.
When a model calls external tools (search, browse, APIs) to gather information or take actions.
How a system handles disagreeing sources; clear canonical pages reduce conflicts and improve accuracy.
05
Writing and information architecture patterns that improve extractability, reduce ambiguity, and increase citation likelihood.
How easily text can be parsed for meaning; shorter sentences and clear structure reduce misinterpretation.
The ratio of meaningful information to boilerplate; denser pages are easier to summarize within limited context.
Whether the first screen communicates what you do, who it's for, and why it's credible.
A precise statement of the primary benefit and differentiation.
The target customer definition; helps assistants match you to intent-driven prompts.
The outcome a user hires a product/service to achieve; a strong framing for AI queries.
Language that prevents confusion with similarly named entities or overlapping categories.
A compact section containing key facts (pricing, features, geography, policies) in a machine-extractable form.
A table of measurable attributes that models can cite and compare.
Question-and-answer formatting that maps directly to assistant behavior and reduces paraphrase errors.
Use of headings, bullets, and short paragraphs so both people and models can locate key statements quickly.
A standard pattern where a term is defined in one or two sentences before deeper explanation.
Using the same names for the same things across pages (features, plans) to avoid conflicting summaries.
Pricing and packaging stated unambiguously (numbers, units, limits) to prevent incorrect assistant answers.
Explicitly stating constraints and exclusions to prevent overclaiming in summaries.
A clear, extractable way to reach you (email, form, phone) that models can surface confidently.
Evidence of results (metrics, case studies) tied to specific claims.
Designing content in reusable blocks so retrieval can pick the right piece without needing the whole page.
Links that provide definitions and supporting pages to keep retrieval grounded.
A page or section that records updates; helps assistants confirm what changed and when.
06
Standards and modeling terms that reduce ambiguity and help machines understand who/what your site is about.
Machine-readable markup that describes entities and relationships on a page.
A shared vocabulary for structured data used by many search and parsing systems.
A uniquely identifiable thing (brand, person, product, place) that can be referenced consistently.
Making it clear which entity a term refers to when names overlap.
A network of entities and relationships used to support retrieval and reasoning.
A schema property linking an entity to authoritative profiles (e.g., social, knowledge bases) to confirm identity.
Structured data describing a company, including name, URL, logo, and identifiers.
Markup describing an individual (author, executive) with role and identity links.
Markup describing a product with attributes like brand, offers, and identifiers.
Markup describing a service offering, useful when the product is delivered as a service.
Structured data summarizing review ratings for a product or organization.
Markup describing individual reviews and their authorship.
Markup for FAQ content that clarifies Q/A pairs for machine extraction.
Markup for step-by-step procedures and required materials.
Structured navigation that clarifies page hierarchy and improves discovery.
Site-level markup that can define search actions and canonical identity.
Page-level markup describing the type and key properties of a page.
Markup for editorial content, including author, publish date, and headline.
Markup describing datasets; useful when your site publishes data intended for reuse and citation.
07
How bots fetch pages and how your site allows or blocks them—critical for both SEO and AI visibility.
An identifier sent by a bot or browser; used in server rules and robots directives.
A file that provides crawling directives; can allow or disallow specific user agents and paths.
A file listing important URLs and metadata to help crawlers discover and prioritize pages.
A proposed convention for providing an LLM-friendly site summary and key links in a compact format.
Whether bots can reach your pages through links and allowed paths.
Whether bots can successfully download the page (status codes, auth, paywalls, blocks).
Whether the important content is present after rendering (especially for JS-heavy sites).
Rendering HTML on the server so bots and users receive content immediately.
Rendering in the browser; can hide content from bots that don't execute scripts fully.
A response code (200, 301, 404, etc.) that signals success, redirects, or errors.
A response indicating the URL doesn't exist; should be used for removed pages.
Ensuring one primary URL represents a piece of content to avoid duplication and confusion.
Splitting lists across pages; requires clear linking for discovery and retrieval.
Filter-based navigation that can generate many URL combinations; needs careful control.
Rules for URLs with query parameters to prevent duplicate or low-value pages.
Restricting request frequency; misconfigured limits can block legitimate bots.
Firewall or rules that deny bots; can inadvertently block AI crawlers and prevent visibility.
09
How to quantify visibility in both classic search and answer engines, and how to run improvement cycles.
The group of alternatives you compare against for visibility and citations.
Converting different signals into a comparable scale so they can be combined into one score.
Ranking fixes by expected impact and effort so work focuses on what moves outcomes.
A reasoned projection of how much a change could affect visibility or citations.
A structured change designed to test a hypothesis under controlled conditions.
Collecting user interactions (clicks, form submits) to evaluate outcomes.
URL parameters used to label campaigns and traffic sources for analytics.
Conversions divided by visits; a business outcome metric.
Tracking groups over time to understand retention and behavior changes.
Ongoing measurement to detect drift, regressions, or improvements.
A drop in performance after a change; requires diagnosis and rollback plans.
Recording what changed (content, markup, routing) so score changes can be explained.
A public or internal log of updates; helps users and agents understand what's new.
10
Technical quality signals that affect fetchability, comprehension, and compatibility across devices and AI modalities.
A set of user experience metrics used to quantify loading speed, responsiveness, and visual stability.
Measures perceived load speed by timing when the largest content element appears.
Measures responsiveness by timing how quickly the page responds to user interactions.
Measures visual stability by tracking unexpected layout movement.
How quickly a page becomes usable; impacts user satisfaction and some ranking systems.
A distributed network that serves assets closer to users and bots, improving speed and resilience.
Compressing and properly sizing images to reduce load while preserving quality.
Deferring offscreen resources until needed; must be implemented carefully for bots and accessibility.
Designing content usable by assistive technologies; often correlates with machine readability.
Text alternatives for images; improves accessibility and helps multimodal systems interpret visuals.
Attributes that label regions (nav, main, footer) to help assistive tech and parsers understand layout.
Using meaningful tags (article, nav, main) rather than generic divs; improves parsing and extraction.
Ensuring the site is usable without a mouse; a practical accessibility requirement.
Sufficient contrast between text and background; improves readability and compliance.
Full text of audio/video; makes content searchable and retrievable for RAG.
Layouts that adapt to different screen sizes; supports mobile and assistant surfaces.
Indexing that prioritizes the mobile version of content; requires parity between mobile and desktop.
Can't find what you're looking for? Get in touch.
Now apply it
Run a free audit on your site. Get a score on all 15 dimensions and a clear list of what to fix — in plain English.