The AEO Glossary: 60 Terms Every Marketing Team Should Know

Visibility and citation metrics

The terms in this section are the ones you will hear in board meetings and quarterly reviews. They describe what the AI is doing with your brand and how to count it.

Citation share. The percentage of AI responses, across a defined query set, that mention or link to your brand. This is the headline AEO metric most programs report against.

Citation rank. Where your brand appears within a single AI response when multiple brands are cited. Being cited first carries different weight than being cited fourth.

Share of voice. Your citation share divided by the combined citation share of you plus your top competitors. A more useful comparative metric than raw citation share for crowded markets.

Mention rate. The percentage of responses that mention your brand by name without necessarily linking. Mentions count toward brand awareness even when they do not drive clicks.

Link rate. The percentage of responses that include a clickable link back to your domain. Lower than mention rate, but closer to traditional referral measurement.

Sentiment score. Whether the AI describes your brand positively, neutrally, or negatively when it cites you. Negative sentiment in citations is one of the most actionable AEO findings.

Persona coverage. The number of distinct buyer personas in your target market for whom you appear in AI answers. Programs often win for one persona before they win for all.

Query coverage. The percentage of your tracked query set where you appear at all (in any rank, with any sentiment). The baseline measure of presence.

Recency-weighted citation share. Citation share weighted toward more recent measurements, smoothing out one-off model behavior shifts. Useful for reporting trends rather than snapshots.

Citation source diversity. The number of distinct domains and content types the AI pulls from when citing your brand. More diversity usually means more durable visibility.

Content and article architecture

The terms here describe how AEO-optimized content is built. If you have read a content brief from an AEO vendor recently, these terms appeared in it.

Answer capsule. A short, structured summary at the top of an article (usually four to six bullets) designed to be lifted directly by an AI as a complete answer.

Schema markup. Structured data added to a page (most often JSON-LD) that tells search engines and AI crawlers what kind of content they are looking at. FAQ, HowTo, Article, and Organization are the most common schemas for AEO.

FAQ block. A question-and-answer section at the end of an article, marked up with FAQPage schema, that gives AI models discrete question-answer pairs to retrieve.

H2 query mapping. Writing section headings (H2s) as plain-language questions that match how users phrase prompts to AI tools. Strongly correlated with citation lift.

Internal linking pattern. The structure of links between articles on your site. Hub-and-spoke patterns, where a pillar page links to detailed subpages, perform better for AEO than flat link patterns.

Content depth signal. The combination of word count, structured elements (tables, lists, FAQs), and topical completeness that signals to AI models that an article is a thorough answer rather than a quick post.

Semantic completeness. Whether an article covers all the related concepts a model would expect in a thorough answer. Missing concepts hurt citation rate even when the central topic is well covered.

Table density. The number of structured tables in an article. Articles with at least one data table get cited at roughly 2.5 times the rate of articles without tables.

Bold-key phrasing. Bolding the key term in a definition or answer (as this glossary does). AI tokenizers weight bolded text slightly higher in some retrieval setups.

Listicle stacking. Building articles as ordered or unordered lists of named items (tools, tactics, terms). Listicles get cited frequently for "best of" and "examples of" queries.

Entity building and knowledge graph

This is the technical side of AEO that most teams underinvest in. These terms describe how AI models recognize your brand as a distinct entity worth citing.

Entity disambiguation. The process of making your brand identifiable to AI models when other brands share your name. Schema, Wikipedia, and Wikidata are the primary tools.

Knowledge graph. The structured representation of entities (people, companies, products, places) and their relationships that underlies AI grounding. Google has the most public one, but every major AI provider runs an internal equivalent.

Wikidata entry. A structured record on Wikidata for your brand. The single highest-leverage entity-building action for most companies.

Wikipedia presence. A Wikipedia article about your brand. Hard to earn (notability requirements are strict), but enormously valuable for AI citation when you qualify.

sameAs links. Schema markup that lists your brand's profiles on other authoritative platforms (Wikipedia, Crunchbase, LinkedIn, GitHub). Helps AI models stitch identity across sources.

Founder entity. A schema record for your founder or key executives linked to your organization entity. Models cite people-led companies more readily when the people are well-defined entities.

Topic authority. A model's implicit ranking of which entities it considers expert on which topics. Topic authority compounds over time and is the single best predictor of citation share growth.

Brand-as-noun status. When AI models will use your brand name as a generic noun (using "a Stripe" to mean a payment processor, for example). The ultimate entity-building goal.

Entity co-occurrence. How often your brand appears alongside related concepts in the broader web corpus. Models use co-occurrence patterns to infer which topics you are credible on.

Canonical entity URL. The single URL that should be treated as the authoritative source for information about your brand. Usually your homepage, ideally the same URL on Wikidata.

Model and platform concepts

Marketing teams running AEO programs do not need to know how a transformer works, but a few model-level concepts come up often enough that they should be in your vocabulary.

Retrieval-augmented generation (RAG). The pattern where an AI model fetches relevant documents from a corpus at query time and grounds its answer in those documents. Most consumer AI tools use RAG for current-information queries.

Grounding. The act of an AI model anchoring its answer to specific retrieved sources. Grounded answers carry citations; ungrounded answers come from model weights alone.

Grounding window. The portion of a model's response that is actually grounded in cited sources versus generated from training data. Some models ground only the first paragraph, then drift.

Training cutoff. The date after which a model has no training data. Anything published after the cutoff only enters the model via retrieval, not via training.

Crawl frequency. How often an AI provider's crawler visits your site. Higher crawl frequency means faster citation pickup for new content.

Citation surface. The visible part of an AI response where citations appear. Surfaces vary from inline footnotes to source cards to sidebar links.

Reranker. The component in a RAG system that reorders retrieved documents by relevance before passing them to the model. Reranker scoring is increasingly important for which sources actually get cited.

Browse mode. A feature in some AI tools (most prominently ChatGPT) where the model fetches live web pages during a query. Browse-mode queries are where most "AI search" citation share gets won.

Provider-specific signals. The ranking signals each AI provider uses that may differ from each other. Perplexity weights recency heavily; Gemini weights Knowledge Graph alignment heavily; Claude weights apparent authority heavily.

Measurement and reporting

The vocabulary here keeps AEO programs honest. These are the terms you should hear in vendor reports and ask about if you do not.

Query set. The defined list of prompts that a measurement platform tests against AI tools to measure your citation share. Should reflect actual buyer language, not just keyword research.

Sample size. The number of distinct queries and the frequency of testing. A 50-query weekly sample is the floor for statistically meaningful measurement.

Test plan. The documented methodology covering which models, which queries, which personas, and which cadence a program measures against. Required for repeatable reporting.

Baseline. The starting citation share measured before any AEO work begins. The reference point for measuring lift.

Lift. The change in citation share over time, usually reported month over month and quarter over quarter.

Attribution window. How long after a content publish you continue to attribute citation lift to that piece of content. Usually 90 to 180 days.

Citation half-life. The decay rate of an article's contribution to your citation share as fresher content takes its place. Varies widely by topic and publishing competitive intensity.

Cohort reporting. Reporting citation lift by cohort (e.g., "articles published in March") rather than by total share. Helps isolate which content investments are working.

Cost per citation point. The dollar cost of moving citation share by one percentage point. The cleanest unit economics measure for AEO programs.

Holdout testing. Reserving a portion of your query set or content output as a control to validate that observed lift is causal, not coincidental.

Program and org concepts

The final group of terms is about how AEO programs get run inside companies. These come up in vendor scoping, org design, and budget conversations.

AEO maturity stage. Where a company sits in the maturity curve from no measurement to industry-default citation share. Most B2B companies are in stage one or two.

Content operations layer. The set of processes, templates, and tools that produce AEO-optimized content at scale. The difference between a one-off content effort and a program.

Shared service model. Running AEO as a centralized function (often used in private equity portfolios or enterprise multi-brand setups) rather than as a per-team capability.

AEO retainer. The monthly fee structure most AEO agencies use, typically covering research, content production, schema, and reporting. OnlyAEO uses retainer structures sized to portfolio scope.

Citation tracking platform. Third-party software that automates query testing and citation measurement across AI tools. Pricing typically runs $1,500 to $5,000 monthly per organization.

AEO scorecard. A defined rubric (usually 100 to 110 points) for evaluating whether a single article meets AEO standards. OnlyAEO's scorecard covers structure, schema, entity signals, and content depth.

Approval gate. A review step in the content pipeline where a marketing leader approves articles before publish. Some programs gate every article; some sample-audit; some skip entirely.

Brand voice profile. A documented description of how a company writes, used to keep AEO-produced content sounding human and on-brand rather than AI-generic.

Topic queue. The prioritized list of upcoming article topics, scored by query volume, competitive gap, and brand fit. The backbone of a working AEO content calendar.

AEO RACI. The responsibility matrix for who owns research, writing, schema, publishing, and measurement. Programs without a clear RACI tend to stall around month four. At OnlyAEO we publish a default RACI at engagement kickoff and adjust it to each client's org chart.