Glossary.
The terms that come up most often in modern search and AI visibility work, defined briefly. Useful before a strategy call, or as a reference when reading the blog.
Plain-English definitions.
- Answer Engine Optimization (AEO)
Optimising content so answer engines like ChatGPT, Perplexity, and Google AI Overviews surface it as a direct answer rather than one of ten blue links. In practice AEO and GEO are used interchangeably and both overlap heavily with modern SEO.
- AI Overviews
Google's generative answer block that appears above the traditional organic results for many queries. It cites source pages, but most users read the synthesised answer without clicking through, so being one of the cited sources is now a primary visibility goal.
- AI Search
Search behaviour where the user asks an AI system a question and receives a synthesised answer instead of a list of links. Most early-stage research now starts in AI search, often before the user visits any website.
- AI crawler
An automated bot that fetches web pages to feed an AI system, either for training or for live retrieval when answering a question. Common ones include GPTBot and OAI-SearchBot from OpenAI, ClaudeBot from Anthropic, Google-Extended, and PerplexityBot. If your robots.txt blocks them, you cannot be cited.
- Answer engine
A tool that answers a question directly in conversational language instead of returning a ranked list of links. ChatGPT, Perplexity, Gemini, and Google AI Overviews are answer engines, and the unit of visibility in them is a citation, not a position.
- Answer-first content
A writing structure that states the direct answer to a question in the first sentence or two, then expands with detail underneath. Answer engines extract the opening lines most often, so leading with the answer raises the odds of being quoted.
- Author authority
The signals that tell search and AI engines who wrote a page and why they are credible on the topic. It is built through consistent bylines, Person schema, an author archive page, and verified profiles the author is linked to across the web.
- Backlink
A link from another website to yours. Backlinks remain a core trust and authority signal for classic search, and the third-party pages that carry them are also frequent sources that AI engines read and cite.
- Canonical URL
The URL you declare as the primary version of a page when the same or similar content sits at more than one address. Set with a rel=canonical tag, it tells search engines which version to index and credit, which stops duplicate versions from competing with each other.
- Chunking
The step where an AI retrieval system splits a page into small passages before deciding what to quote. Engines cite the chunk that best answers a sub-query, so self-contained sections that make sense on their own are easier to lift than long unbroken prose.
- Citation
The source a generative engine attributes when it answers a query. Being cited, rather than just ranked, is the new measure of visibility, since the citation is what carries attribution and trust back to your domain.
- Citation rate
How often your domain is cited across a tracked set of prompts and engines, usually expressed as a percentage of answers. It turns AI visibility from a feeling into a number you can move and report on.
- Co-citation
When two brands or sources are named together in the same answer or on the same third-party page. Being co-cited alongside recognised names in your category is a strong signal to AI engines that you belong in that set.
- Content decay
The slow loss of rankings, traffic, and citations as a page ages and competitors or facts move on. The fix is a scheduled refresh that updates data, examples, and internal links rather than leaving the page to drift.
- Core Web Vitals (CWV)
Google's set of page-experience metrics for loading, interactivity, and visual stability, currently Largest Contentful Paint, Interaction to Next Paint, and Cumulative Layout Shift. They are a ranking factor and a proxy for whether a page is technically healthy enough to be crawled and read cleanly.
- Cosine similarity
A maths measure of how close two pieces of text are in meaning once each is turned into a vector. Retrieval systems use it to score candidate passages against a query, which is why content that matches the meaning of a question can be pulled even when it does not match the exact words.
- Crawl budget
The number of pages a search engine is willing to crawl on your site in a given window. It matters mainly for large sites, where crawling wasted on low-value or duplicate URLs can starve the pages you actually want indexed.
- Crawlability
Whether search and AI bots can reach and read your pages at all. Blocked resources, robots.txt rules, heavy JavaScript, and login walls all reduce it, and a page that cannot be crawled cannot be ranked or cited.
- Digital PR
Earning mentions and links on trusted third-party sites through stories, data, and expert commentary. Because AI engines lean heavily on independent sources, digital PR is one of the most direct ways to earn the co-citations that move AI visibility.
- E-E-A-T
Google's shorthand for Experience, Expertise, Authoritativeness, and Trustworthiness, the qualities its raters use to judge content, especially on topics that affect money or health. AI engines lean on the same kinds of signals when deciding which sources are safe to cite.
- Embedding
A list of numbers that represents the meaning of a piece of text so a machine can compare it to other text. Search and AI systems convert both your content and the user's query into embeddings, then match them by meaning rather than by exact keywords.
- Entity
A distinct thing that search and AI engines can identify and reason about, such as a person, company, product, or place. Modern search is organised around entities and their relationships, not just strings of keywords.
- Entity disambiguation
The work of making sure engines connect a name to the right entity when several share it. Clear schema, consistent details, and sameAs links to verified profiles help an engine tell your brand or author apart from others with the same name.
- Entity signals
The machine-readable data that helps engines understand who a person, company, or product is and how they relate to other entities. Examples are Person and Organization schema, a consistent name and address across the web, and sameAs links pointing at verified profiles.
- FAQ schema
JSON-LD that marks up question-and-answer pairs on a page so engines can read them as discrete answers. Google now shows FAQ rich results only for a narrow set of sites, but the markup still helps AI engines lift clean question-answer blocks.
- Featured snippet
A short answer Google pulls from a ranking page and shows at the top of the results. Featured snippets were the pre-AI version of answer-first visibility, and the content patterns that won them tend to help with AI citations too.
- Generative Engine Optimization (GEO)
The practice of shaping a site's content, structure, and entity signals so that generative AI engines cite it when answering relevant questions. GEO overlaps heavily with SEO but adds work on quotable answers, entity clarity, and citation-friendly formatting.
- Grounding
Tying an AI system's answer to real retrieved sources instead of letting it rely on memory alone. Grounded answers cite pages, which is exactly the moment your content can be pulled in and credited.
- Hallucination
When an AI system states something that sounds confident but is wrong or invented. Clear, well-structured, well-sourced content lowers the odds that an engine describes your brand incorrectly.
- HowTo schema
JSON-LD that marks up the ordered steps of a genuine procedure. Google has wound down HowTo rich results, but the markup can still help engines parse step-by-step instructions, and it should only sit on content that really is a set of steps.
- Indexing
The stage where a search engine stores a crawled page so it can appear in results. A page has to be indexed before it can rank, and pages that are crawled but left out of the index are effectively invisible.
- Internal linking
The links between pages on your own site. Good internal linking spreads authority, shows engines how your topics relate, and helps both crawlers and AI systems find and understand supporting pages.
- Knowledge Graph
Google's database of entities and the relationships between them. Being represented cleanly in the Knowledge Graph helps engines describe your brand accurately and pull the right facts into answers and panels.
- Knowledge Panel
The information box Google shows to the side of the results for a recognised person, company, or thing. It is drawn from the Knowledge Graph and trusted sources, and a clean entity setup is what earns and keeps it accurate.
- Large Language Model (LLM)
An AI model trained on large amounts of text to predict and generate language, which lets it answer questions in a conversational way. LLMs power ChatGPT, Claude, Gemini, and the answer layer of modern search.
- llms.txt
A proposed plain-text file at the root of a site that gives AI tools a curated map of its most important pages. Adoption is early and no major engine treats it as required, but it is a low-cost way to signal your key content to agents that look for it.
- Long-tail query
A longer, more specific search phrase with lower individual volume but clearer intent. Conversational AI has pushed more search into long-tail, question-shaped phrasing, which rewards content that answers precise questions directly.
- Organization schema
JSON-LD that describes a business entity to engines. Useful fields include name, url, logo, a founder linked to a Person, areaServed, and sameAs. With Person schema, it forms the backbone of how AI engines understand a company.
- Person schema
JSON-LD that describes a person to search engines and AI crawlers. Key fields are name, jobTitle, description, knowsAbout, and sameAs links to verified profiles. It is a primary input to how AI engines describe a consultant or author when asked.
- Pillar page
A comprehensive page that covers a broad topic and links out to narrower supporting pages. The pillar anchors a topic cluster and is often the page that earns rankings and citations for the head term.
- Prompt
The question or instruction a user gives an AI system. In AI visibility work, a fixed set of buyer-intent prompts is the test bed you measure citations and share of voice against.
- Query fan-out
When an AI engine silently expands one prompt into several related searches to gather sources before answering. It means a single answer can pull from pages that match any of the sub-queries, not only the exact phrase the user typed.
- Query reformulation
When an engine rewrites a user's query before searching, sometimes into a semantically related phrase, sometimes into several sub-queries. It means exact-keyword targeting is no longer enough, since the page has to match the topic the engine reformulates toward.
- Retrieval-Augmented Generation (RAG)
A method where an AI system retrieves relevant documents at answer time and uses them to write a grounded, cited response, instead of relying only on what it memorised in training. Most cited answers in AI search are produced this way, which is what makes real-time content optimisation possible.
- Rich results
Search results that show extra detail such as ratings, prices, or FAQs, driven by structured data on the page. They can lift visibility and click-through, though Google has narrowed which types it shows over time.
- robots.txt
A file at the root of a site that tells crawlers which paths they may or may not fetch. It now governs AI crawlers too, so an over-broad rule can quietly block the bots that would otherwise cite you.
- sameAs
A schema property that links an entity to its other verified profiles across the web, such as LinkedIn, Crunchbase, or an official site. Consistent sameAs links are one of the strongest ways to confirm identity and merge scattered mentions into one entity.
- Schema.org
The shared vocabulary of types and properties that search engines agree to read, usually written on a page as JSON-LD. It is how you tell engines that a block of text is a person, an article, a product, or a set of steps.
- Search intent
The goal behind a search, usually grouped as informational, commercial, or transactional. Matching intent is the difference between a page that gets read and one that gets skipped, in both classic and AI search.
- Semantic HTML
Using HTML tags that describe what content is, such as headings, lists, and tables, rather than generic containers. Clean semantic markup makes a page easier for crawlers and AI systems to parse and quote accurately.
- Semantic search
Search that matches on meaning rather than exact keywords, powered by embeddings and vector comparison. It is why a page can be surfaced for a question it never uses the exact words of, as long as it covers the same idea.
- SERP
The search engine results page, meaning everything shown for a query, from organic links to ads, snippets, and AI Overviews. As AI answers take more of the page, the classic ten blue links are a shrinking share of it.
- Structured data
Machine-readable markup, usually JSON-LD following the Schema.org vocabulary, that labels the meaning of content on a page. It powers rich results and gives AI engines a clean, unambiguous description of your entities and content.
- Topic cluster
A content structure where one pillar page covers a broad topic and several supporting pages cover sub-topics, all linked back to the pillar. Engines reward the topical authority a well-built cluster signals more than they reward isolated high-ranking pages.
- Topical authority
The depth and consistency of coverage that makes a site a trusted source on a subject. It is built by covering a topic thoroughly across linked pages, and it is one of the clearest patterns AI engines reward when choosing who to cite.
- Vector database
A store built to hold embeddings and find the closest matches to a query fast. It is the retrieval layer behind semantic search and most RAG systems, the component that decides which passages an answer engine even considers.
- Zero-click search
A search that ends on the results page because the answer is shown there, in a snippet or an AI Overview, so the user never clicks a link. It makes being the cited source inside the answer matter more than winning the click.