AEO and demand generation glossary

Every term below is defined in one self contained sentence, the way an answer engine needs to read it. Where a definition rests on published research, the study is linked.

AI search

Answer engine optimization (AEO): Answer engine optimization is the practice of structuring content so AI answer engines quote it and name the brand as a source inside a generated answer.The unit of competition is the passage that gets cited, not the position of a link.
Generative engine optimization (GEO): Generative engine optimization is the academic term for the same discipline: optimizing pages for citation and absorption inside generative search systems.arXiv, April 2026
Answer engine: An answer engine is any system that responds to a question with a written answer rather than a list of links, such as ChatGPT, Perplexity, Gemini, Claude, Copilot, or Google AI Overviews.
Citation selection: Citation selection is the first stage of AI visibility: the platform triggers a search and decides which retrieved sources to attach to its answer.arXiv, April 2026
Citation absorption: Citation absorption is the second stage: how much of a cited page's substance actually shows up in the generated answer text.A page can be cited and contribute almost nothing, which is visibility without influence.arXiv, April 2026
Citation re-ranking pass: A citation re-ranking pass is the evaluation an AI system applies between search retrieval and answer citation, using criteria that differ from search ranking factors.Documented criteria include source type credibility, consensus across sources, evaluative depth, a self ranking discount, and claim specificity.Scientific Institute for Generative Intelligence (SIGI-2026-056), March 2026
Retrieval augmented generation (RAG): Retrieval augmented generation is the architecture behind most answer engines: retrieve documents, place them in the model's context, then generate an answer grounded in them.
Passage: A passage is the chunk of a page an answer engine retrieves and reads, typically a section rather than the whole document.
Chunking: Chunking is how a retrieval system splits a page into retrievable units, which is why self contained sections outperform ideas spread across a long narrative.
Direct answer block: A direct answer block is a short, self contained paragraph, usually 40 to 60 words, placed immediately under a question shaped heading so a model can lift it without rewriting.
Claim specificity: Claim specificity is how precise and verifiable a statement is, and it is one of the strongest observed predictors of whether a model cites a page.A dated number with a named source outperforms the same idea stated as an opinion.Scientific Institute for Generative Intelligence (SIGI-2026-056), March 2026
Consensus detection: Consensus detection is a model's tendency to prefer claims corroborated by multiple independent sources over claims that appear on only one site.Scientific Institute for Generative Intelligence (SIGI-2026-056), March 2026
Citation fingerprint: A citation fingerprint is the mix of page types AI engines rely on when answering questions in a given industry, and it differs sharply between categories.In B2B technology services, listicles captured 61 percent of tracked citations in 2026.DeltaV Digital, July 2026
AI Overviews: AI Overviews is Google's generated summary that appears above traditional results and cites a small set of supporting pages.Canadian Conference on Artificial Intelligence (PMLR 318), June 2026
AI Mode: AI Mode is Google's fully conversational search surface, where the entire result is a generated answer with citations rather than a ranked list.
Zero click search: A zero click search is a query resolved on the results surface itself, so the user gets the answer without visiting any of the cited sources.
llms.txt: llms.txt is a plain text file at a site's root that lists its most important pages in a form language models can read cheaply.It is a proposed convention, not an official standard, and no major engine has confirmed using it as a ranking or citation input.
AI crawler: An AI crawler is a bot that fetches pages for a language model product, such as GPTBot, OAI-SearchBot, PerplexityBot, ClaudeBot, or Google-Extended.Most of them do not execute JavaScript, so client rendered content can be invisible to them.

Search

Schema markup: Schema markup is structured data in JSON-LD that states explicitly what a page is, such as Article, FAQPage, Product, or DefinedTerm.
Entity: An entity is a thing a search or answer system recognises as distinct, such as a company, person, or product, independent of any one page about it.
E-E-A-T: E-E-A-T stands for experience, expertise, authoritativeness, and trustworthiness, the qualities Google's rater guidelines use to judge content quality.
Server side rendering (SSR): Server side rendering means the server returns complete HTML for a page, so crawlers that do not execute JavaScript still see the content.
Prerendering: Prerendering generates static HTML for each route at build time, giving non JavaScript crawlers full markup without moving to a server rendered framework.
Canonical URL: A canonical URL tells crawlers which address is the authoritative version of a page, preventing duplicate versions from splitting signals.
Topical authority: Topical authority is the depth and interlinking of a site's coverage of one subject, which raises the odds any single page on that subject is trusted.

Demand generation

Demand generation: Demand generation is the practice of creating and capturing buyer interest in a category and a product, measured in pipeline rather than form fills.
Lead generation: Lead generation is the subset of demand generation focused on collecting contact details from people who are already in market.
Dark social: Dark social is buyer research and discussion that happens in places analytics cannot attribute, such as private communities, group chats, and AI assistants.
Ideal customer profile (ICP): An ideal customer profile is the firmographic and behavioural definition of the accounts a company can win and retain profitably.

Measurement

Prompt set: A prompt set is a fixed list of buyer questions run repeatedly across AI engines to measure whether and how often a brand is cited.
Citation share: Citation share is the percentage of tracked prompts in which a brand appears as a cited source, and it is the closest AI search equivalent of share of voice.
Share of model: Share of model is how often a brand is named in generated answers for a category, whether or not a clickable citation is attached.
Pipeline coverage: Pipeline coverage is open pipeline divided by the bookings target for the same period, with three to four times treated as healthy in 2026 B2B SaaS.Causo, June 2026
CAC payback: CAC payback is the number of months of gross profit required to recover the cost of acquiring a customer, and it is the controlling efficiency metric for most 2026 boards.Causo, June 2026
Magic number: Magic number is net new ARR divided by prior period sales and marketing spend, used as a quick read on go to market efficiency.Causo, June 2026
Win rate: Win rate is closed won opportunities divided by all closed opportunities, and it compressed to roughly 19 percent on average across a 655,000 opportunity dataset.Ebsta and Pavilion, via PipelineGrader, July 2026
Pipeline velocity: Pipeline velocity is the rate at which pipeline converts to revenue, combining opportunity count, average deal size, win rate, and cycle length.
Marketing sourced pipeline: Marketing sourced pipeline is the value of opportunities whose first recorded touch belongs to marketing, and it systematically undercounts channels that create demand without a click.
Incrementality: Incrementality is the pipeline a channel creates that would not have happened otherwise, measured with holdouts or geo tests rather than attribution reports.
Blended CAC: Blended CAC is total sales and marketing spend divided by all new customers, including those attributed to organic and word of mouth.Benchmarkit, June 2026

Want the longer version?

The guide library covers each of these in depth, with sourced data and worked examples.

Browse all guides