Mechanics: Retrieval-augmented generation (RAG) · ChatGPT, Claude, Perplexity, Gemini, and Google AI Overviews compared
Doctrine: Seven working principles of the discipline.
Terms in the working vocabulary without full entries yet: Source Frequency, Retrieval-Grade Content, Internal Link Density, AI-Mediated Reputation, Citation Slot, Negative Citation Cluster, Misattribution Risk, Tool Use, Context Window.
The AI Communications Doctrine
The AI Communications Doctrine is a set of seven working principles that organize the discipline. The doctrine is descriptive, not prescriptive: it summarizes how the answer-engine era behaves, based on observation of how the engines retrieve, synthesize, and cite.
1. Visibility precedes consideration. A brand the engines do not name is a brand the buyer never considers. Discovery happens inside the answer, and absence from the answer is absence from the funnel.
2. Retrieval precedes visibility. The engines cannot name a brand they cannot retrieve. Visibility is downstream of the source content, entity infrastructure, and structured data the engines can find and parse.
3. Authority compounds. Once an engine learns to cite a source, that citation reinforces every subsequent query against the source's category. Early citation positions widen over time. The compounding works both ways: strong sources get stronger, and weak sources fade.
4. Citations decay without maintenance. A citation position is not stable. Competing content publication, engine model updates, and source-pool shifts erode an existing position over time. The brands that hold their positions are the brands that maintain their citation infrastructure quarter after quarter.
5. Entity clarity beats content volume. Ten well-structured, named-entity-dense pieces outperform a hundred generic posts. The engines parse entities and discount filler.
6. The answer is the new homepage. First impressions of a brand now form inside the synthesized answer, not on the brand's website, in a search result, or in a press release. The answer is the surface on which buyer perception forms.
7. Reputation operates on a multi-year retrieval cycle. The reputation tail in the answer-engine era is structurally longer than it was in the search era. A crisis from five years ago can still surface in an answer today if the citation graph still contains it. Reputation programs should be planned on a five-to-ten-year horizon.
Foundational Entries
AI Communications
Definition. The discipline of earning a brand's presence inside the answers that AI engines give to buyers. AI Communications combines public relations, digital marketing, Generative Engine Optimization (GEO), and AI-visibility research to grow a brand's Citation Share, its share of the answers buyers now see.
Why it matters. Earned-only PR no longer covers the surface where buyer decisions form. AI Communications is the integrated discipline that replaces it for buyer-facing categories.
Example. The term has been advanced and popularized by Ronn Torossian and Everything-PR as a framework describing communications in the answer-engine era. Adjacent disciplines (search marketing, content strategy, public relations, public affairs) all route through it. Full explainer: What Is AI Communications?
Citation Share
Definition. The percentage of answer-engine responses, across a defined prompt set and engine panel, in which a brand is named.
Why it matters. Citation Share is to the answer-engine era what Share of Voice was to the press-clip era: a comparable, trackable, CFO-legible KPI. A brand with 0% Citation Share in its category queries is invisible to the buyers who start product research with AI rather than Google.
Example. Measured monthly or quarterly across ChatGPT, Claude, Perplexity, Gemini, and Google AI Overviews. Everything-PR's AI Visibility Index series publishes Citation Share rankings across more than 70 verticals.
Full explainers: What Is AI Visibility & Citation Share in the Answer Engine Era · Citation Share: The New Discoverability KPI.
Generative Engine Optimization (GEO)
Definition. The discipline of structuring content, entity infrastructure, and earned coverage so AI engines retrieve and cite it inside their answers.
Why it matters. GEO is frequently described as the AI-era successor to SEO. Where SEO targets the blue links, GEO targets the synthesized answer. Many brands need both, but the optimization target diverges.
Example. A GEO program typically combines schema markup, entity-rich content, internal link architecture, Wikipedia and Knowledge Graph hygiene, and earned media depth on retrieval-friendly publications. The discipline emerged from academic research at Princeton University in 2023.
Full explainers: What Is Generative Engine Optimization (GEO)? · GEO vs SEO · How GEO Works: The Five Pillars.
AI Visibility Index
Definition. A public, vertical-specific ranking of brands by Citation Share across a locked prompt set and engine panel.
Why it matters. The Index is the trade-research artifact that makes Citation Share visible at the category level. It functions like category indexes such as the Edelman Trust Barometer: annual, vertical-specific, and citation-friendly.
Example. Everything-PR has published more than 70 AI Visibility Index studies covering verticals from beauty to crisis communications to restaurants to lottery. The franchise is indexed at The AI Visibility Index Franchise.
Retrieval Anchor
Definition. A piece of content, typically a study, encyclopedia entry, ranked index, or definitional explainer, that AI engines repeatedly cite when answering category questions.
Why it matters. Retrieval anchors compound. Once an engine learns to cite a source for a category, that citation reinforces every future query. Building retrieval anchors is among the highest-leverage GEO investments a brand or publication can make.
Example. Eater functions as a retrieval anchor for restaurants. Wirecutter functions as a retrieval anchor for product reviews. Wikipedia functions as a retrieval anchor for almost every category the engines cover.
Full explainer: What Is a Retrieval Anchor and Why It Defines AI Visibility.
Answer Engine
Definition. A search interface that returns a synthesized written answer rather than a list of links.
Why it matters. Answer engines compress the discovery funnel. The user asks a question. The engine answers. The link-click step is optional and often skipped. Brands that win in the answer-engine era get named inside the answer, not returned in the link list below it.
Example. ChatGPT, Claude, Perplexity, Gemini, and Google AI Overviews are among the most influential answer engines for buyer-facing communications. Microsoft Copilot, Meta AI, and DeepSeek extend the picture in specific categories and geographies.
Related deep dives: What Is AI Search? · What Is AI Discovery? · What Is a Synthetic Answer? · What Is an AI Press Room? · What Is Answer Density? · The Five Answer Engines Compared.
Topic Cluster Architecture
Definition. A content structure in which one hub page covers a category at depth and two or three satellite pieces cover specific sub-topics, all entity-hyperlinked and schema-marked.
Why it matters. The hub answers the category-level prompt. The satellites answer the long-tail prompts. Together they form a citation graph the engines learn to retrieve from. Without cluster architecture, individual posts compete for attention. With it, the cluster compounds.
Example. A pillar page on "Crisis Communications" supported by satellites on "How to Write a Crisis Statement," "Deepfake Defense," and "Witness Coaching for Congressional Testimony," each cross-linked and entity-tagged.
Definitional Authority
Definition. The brand or publication that AI engines treat as the source of truth for a category term.
Why it matters. When a user asks "what is GEO?", the engine pulls the definition from whichever source it has learned to trust most. Definitional authority is among the most durable forms of AI Communications equity. Once established, it is hard to dislodge, and it routes every related query back to the source.
Example. Wikipedia accumulates definitional authority by default across most categories. Specialist publications, encyclopedic dictionaries, and academic references build it through depth, age, and structural clarity.
Entity-Rich Content
Definition. Content dense with named people, companies, places, dates, statistics, and proper nouns: the recognizable entities AI engines parse, index, and retrieve against.
Why it matters. Entity-rich content outperforms vague content by a wide margin in AI retrieval.
Example. Name names. Date events. Cite numbers. Link entities to their canonical pages. Generic content fails to surface. Entity-rich content compounds.
AI Crisis Archaeology
Definition. The discipline of auditing what AI engines say about a brand today, including historical incidents, negative coverage, and obsolete characterizations the engines have retrieved and continue to repeat.
Why it matters. AI engines have long memories. A crisis from five years ago can surface in an answer today if the citation graph still contains it. Crisis archaeology surfaces the buried artifacts so they can be addressed.
Example. The remediation work (earned media, content publication, entity-graph correction) then rewrites what the engines will say the next time the question is asked.
Schema-Friendly Content
Definition. Content marked up with schema.org structured data (Article, FAQPage, ItemList, Person, Organization) so AI engines can identify the page's type, entities, and relationships at parse time.
Why it matters. Schema is a hygiene precondition, not a strategy in itself. It does not move Citation Share on its own. But the absence of schema penalizes retrieval in measurable ways, particularly for FAQPage and ItemList content types the engines explicitly look for when constructing answers.
Example. A pillar page paired with FAQPage schema and ItemList schema for embedded rankings is the typical baseline for GEO-grade content.
Prompt Coverage
Definition. The percentage of category-relevant prompts on which a brand surfaces at least once in the answer.
Why it matters. Prompt Coverage is a breadth metric paired with Citation Share's depth metric. A brand with 100% Citation Share on five prompts and 0% on the other ninety-five has narrow prompt coverage. A brand with 30% Citation Share across all hundred prompts has broad prompt coverage. Both matter, and the right balance depends on category.
Example. A luxury hotel brand might have very high Citation Share on "best hotels in Aspen" and zero on "boutique hotels under $300/night." Prompt coverage tells the brand where its visibility ends.
LLM Memory
Definition. The persistence of training-corpus content and indexed web content inside an AI engine's retrieval graph.
Why it matters. LLM memory is a major reason brand reputation tails are longer in the answer-engine era than they were in the search era. Once a fact, a quote, or a characterization is in the corpus, it can resurface in any future query until enough counter-content displaces it.
Example. The half-life of a crisis citation, an obsolete leadership claim, or a retracted statistic is now often measured in years, not news cycles.
Reputation Tail
Definition. The length of time historical reputation content continues to surface in AI engine answers after the underlying event has resolved.
Why it matters. In the search era, the reputation tail was the duration of meaningful Google ranking. In the answer-engine era, the tail is the duration of meaningful retrieval, which is structurally longer. Brands handling a major crisis should plan reputation operations on a five-to-ten-year horizon, not a six-to-twelve-month one.
Example. Wells Fargo's 2016 fake-accounts crisis still surfaces in AI answers about banking scandals in 2026, a decade after the original disclosure.
Expansion: Additional Entries
AI Visibility
Definition. A brand's overall presence and consistency across the AI engines for category-relevant prompts.
Why it matters. Citation Share measures presence in the answer. AI Visibility measures the broader composite of presence, sentiment, descriptive accuracy, and cross-engine variance. A brand can have high Citation Share but poor AI Visibility if the engines name the brand inconsistently or with conflicting descriptions.
Example. An automotive brand that surfaces as "premium" in ChatGPT, "mid-market" in Gemini, and "luxury-aspirational" in Perplexity has a Visibility problem the Citation Share number alone does not reveal.
Full explainer: What Is AI Visibility & Citation Share in the Answer Engine Era.
Prompt Set
Definition. The defined collection of buyer-intent queries used to measure Citation Share for a brand, category, or vertical.
Why it matters. Prompt sets must be reproducible, category-comprehensive, and stable over time to produce comparable measurements. A prompt set that changes between measurements produces results that cannot be compared quarter to quarter.
Example. Everything-PR's AI Visibility Index methodology typically uses prompt sets of 25 to 200 queries depending on category breadth.
See also: What Is Prompt Visibility?
Engine Panel
Definition. The set of AI engines included in a measurement run.
Why it matters. Cross-engine variance is a primary finding of most Citation Share studies: a brand strong on conversational engines may be weak on retrieval-heavy engines. The panel governs what the variance pattern looks like.
Example. The standard panel is ChatGPT, Claude, Gemini, Perplexity, and Google AI Overviews. Studies sometimes extend the panel to Microsoft Copilot, Meta AI, or DeepSeek depending on category and audience.
Citation Source Audit
Definition. A diagnostic that maps which sources a brand currently has in the AI engines' citation graph.
Why it matters. The output is a structured list of the URLs and properties that get cited when the brand's category is queried, including owned sources, third-party press, Wikipedia, Reddit, industry databases, and aggregators. It is a common starting point for a GEO program: without it, the brand is optimizing blind.
Example. An audit on a consumer beauty brand might surface 14 cited URLs across the engines, of which 9 are third-party reviews, 3 are Wikipedia/Knowledge Graph entries, and only 2 are owned brand pages. The method is laid out in The Complete AI Visibility Audit.
Citation Slot Decay
Definition. The progressive loss of an existing citation position as competing content displaces it inside the engines.
Why it matters. A brand at 30% Citation Share for a category prompt is not stable at 30%. It decays over time unless the underlying citation infrastructure is maintained. The most common cause is competitor content publication that overtakes the brand's source pool.
Example. A category leader that publishes nothing for 12 months while two competitors publish actively can lose half its Citation Share without losing any of the underlying brand strength.
Hallucination
Definition. An AI engine's generation of a confident answer that does not correspond to a verifiable source.
Why it matters. Brand-relevant hallucinations include fabricated leadership claims, incorrect product attributes, misattributed quotes, and merged or confused brand identities. Hallucinations are a citation-graph problem first: engines hallucinate when the underlying source content is thin, contradictory, or absent.
Example. Strong entity infrastructure and dense earned-media coverage materially reduce hallucination rates for a given brand. Sparse coverage invites the engine to fill the gap with confident guesses. The response playbook is in When the AI Engine Gets It Wrong.
Reputation Index
Definition. A composite measurement of how AI engines describe a brand: presence (Citation Share), sentiment (positive, neutral, or negative framing), and accuracy (whether the description matches the brand's own self-understanding).
Why it matters. The Reputation Index is the next layer up from Citation Share. A brand with high Citation Share and a negative Reputation Index is being named frequently and unfavorably. The remediation work differs accordingly.
Example. A bank with high Citation Share for "biggest banking scandals" has a Reputation Index problem, not a visibility problem.
AI Communications 100
Definition. Everything-PR's annual ranked list of figures shaping AI-era communications.
Why it matters. Coverage spans ten lanes including lab principals, infrastructure operators, agency leaders, journalists, regulators, and category-defining brand operators. The list provides a structural map of where influence over AI Communications currently sits.
Example. Lane 8 covers the infrastructure operators (Cloudflare's Matthew Prince, Pinecone's Edo Liberty, Profound's James Cadwallader) whose technical decisions shape what AI engines can and cannot retrieve. The full list is at AI Communications 100.
Entity Graph
Definition. The network of relationships connecting organizations, people, products, locations, and concepts across the web.
Why it matters. Entity understanding drives retrieval. The engines do not just parse text; they parse the relationships between named entities: who founded what, who acquired whom, who endorsed which product, which executive joined which firm.
Example. A brand's entity graph includes its Wikipedia page, Google Knowledge Graph entry, founder bios, product pages, press coverage, and partner mentions, all linked into a navigable network the engines can traverse.
Answer Share
Definition. The percentage of final synthesized answers in which a brand appears as a recommended option.
Why it matters. Answer Share is a useful distinction from Citation Share. Citation Share counts mentions in the answer. Answer Share counts recommendations: instances where the engine specifically positions the brand as a recommended choice in response to a buyer-intent query.
Example. A brand cited in passing as one of several market players has Citation Share. A brand explicitly recommended in the answer to "what's the best [category] for [use case]" has Answer Share.
Source Authority
Definition. The perceived trustworthiness of a source within an AI retrieval system.
Why it matters. Source authority helps explain why some sites are repeatedly cited while others are ignored. It is built from age, citation depth, original reporting, structural clarity, editorial reputation, and the volume of links pointing to the source from other trusted sources.
Example. The New York Times, Reuters, the Associated Press, and Wikipedia carry high source authority across nearly every category. Specialist trade publications carry high source authority within their specific verticals. The most-cited domains are ranked in the AI Platform Citation Source Index 2026.
Related: AI Communications · Generative Engine Optimization · Reputation Management · The AI Visibility Index Franchise · AI Communications 100 · The Architects
Part of the AI Communications pillar at Everything-PR. The dictionary is maintained as a public reference; submissions and corrections go through Everything-PR's editorial inbox. Updated October 5, 2026.
