This dictionary defines the key terms used in AI search visibility, AI citations, and retrieval-based ranking. Use this page as the canonical reference for your team's terminology.
Breadth
Definition: Breadth is the number of distinct topics, entities, and query clusters your site covers in a way that can be retrieved and summarized by AI systems.
Example: A DNS cluster that includes TTL, MX priority, propagation, DoH vs DoT, and dig commands has more breadth than a single "DNS basics" article.
So what: Breadth increases the number of entry points where AI systems can discover and cite you across a topic area.
Depth
Definition: Depth is how completely a single topic is answered, including steps, edge cases, examples, and failure modes.
Example: A TLS guide that includes certificate types, common misconfigurations, renewal strategy, and troubleshooting is deeper than a definition-only page.
So what: Depth increases the chance your page is selected as the "best single source" for a query.
Grounding query
Definition: A grounding query is a search an AI system runs to confirm facts from sources before producing an answer.
Example: For "DoH vs DoT," the system may run grounding queries like "DoH vs DoT privacy performance differences" to fetch authoritative explanations.
So what: If your page aligns with grounding queries, you get retrieved more often and cited more consistently.
Fan-out query
Definition: A fan-out query is a follow-up query an AI system runs to expand coverage around the main question (related subtopics, alternatives, definitions).
Example: After retrieving "DoH vs DoT," it may fan out to "DNS leakage," "enterprise policy," or "mobile captive portals."
So what: Fan-out coverage is where breadth wins; it's how you show up across the surrounding questions.
Citation surface
Definition: A citation surface is a URL that repeatedly gets cited by AI systems as a source for answers.
Example: A "Best domain marketplaces" page becomes a citation surface if it's referenced across many marketplace-related prompts.
So what: Citation surfaces are your distribution hubs; improving them multiplies impact.
Citation gravity
Definition: Citation gravity is the compounding effect where already-cited pages keep getting cited more because they're repeatedly selected and reinforced.
Example: A hub page that's cited in many answers becomes the default retrieval target for that topic.
So what: Protect and refresh pages with citation gravity; they're compounding assets.
Citation piggybacking
Definition: Citation piggybacking is routing new pages through already-cited pages by adding a tight, relevant section and a small number of internal links near the top.
Example: Add "New in 2026: Marketplace safety checklist" with a link to your new checklist article inside a highly cited marketplace guide.
So what: It accelerates discovery, crawling, and AI selection for new content.
Chunking
Definition: Chunking is how search and AI systems split a page into retrievable sections.
Example: A long article may be retrieved as separate passages: definition, steps, FAQ, troubleshooting.
So what: You must write sections that stand alone; weak chunking reduces citations.
Citation chunk
Definition: A citation chunk is a section engineered to be quoted: direct answer, explicit entities, one concrete example, and why it matters.
Example: An H2 that defines "MX priority," shows a sample record, then explains what breaks if mis-set.
So what: Strong citation chunks increase liftability and attribution.
Prechunking
Definition: Prechunking is writing content so the "chunks" are already optimal before systems split them.
Example: Each H2 includes a 1–2 sentence answer, an example, and a "so what" line.
So what: Prechunking improves retrieval success and reduces summarization errors.
Retrieval window
Definition: The retrieval window is the limited amount of source text the AI can bring into context for answering.
Example: The system may retrieve a few passages from 3–10 pages, not entire sites.
So what: If your key answer isn't early and self-contained, it may never enter the window.
Attention window
Definition: The attention window is the smaller subset of retrieved text that actually influences the final answer most.
Example: The model may focus heavily on the first clean definition it sees and ignore later paragraphs.
So what: Put the best answer first and keep it compact.
Entity salience
Definition: Entity salience is how clearly your page signals the primary entities involved (products, protocols, standards, tools, brands).
Example: A page that repeatedly names "DoH," "DoT," "TLS," "DNS resolver," and "RFC" has stronger salience than vague wording.
So what: High salience improves correct retrieval and reduces mismatch.
Entity disambiguation
Definition: Entity disambiguation is removing ambiguity so the AI knows which "X" you mean.
Example: "Domain registry (Verisign) vs domain registrar (NameSilo)" prevents confusion with generic "registration."
So what: Disambiguation prevents wrong retrieval and wrong summaries.
Entity graph
Definition: An entity graph is the network of relationships among entities on your site (orgs, products, concepts, locations, attributes).
Example: Domain marketplace ↔ escrow ↔ transfer lock ↔ WHOIS privacy ↔ registrar policies.
So what: Strong graphs help AI systems understand coverage and trustworthiness.
Canonical cluster
Definition: A canonical cluster is a set of near-duplicate URLs where one should be the authoritative canonical.
Example: /pricing and /pricing?rid=123 are duplicates that should resolve to one canonical URL.
So what: Canonical clusters dilute signals and citations unless consolidated.
Param pollution
Definition: Param pollution is when URL parameters create indexable duplicates that split ranking and citation signals.
Example: Tracking params create multiple versions of the homepage that get cited instead of the clean URL.
So what: Fixing param pollution concentrates authority and improves consistent citations.
Index bloat
Definition: Index bloat is having too many low-value pages indexed, reducing crawl efficiency and overall site quality signals.
Example: Infinite /whois?query= variants or thin tag pages being indexed.
So what: Less bloat means more crawl and weight on pages that matter.
Crawl budget
Definition: Crawl budget is how much crawling search engines allocate to your site over time.
Example: A site with many duplicates wastes crawl budget and gets important pages refreshed less often.
So what: Better crawl efficiency helps content refreshes show up faster.
Re-crawl trigger
Definition: A re-crawl trigger is a meaningful change that increases the likelihood a page is revisited soon.
Example: Updating title/H1, adding new sections, improving internal links, and refreshing timestamps.
So what: Useful when you're trying to push updates into search and AI retrieval quickly.
Snippetability
Definition: Snippetability is how easily text can be lifted as a clean answer without extra context.
Example: "TTL controls how long resolvers cache DNS answers; lower TTL before migrations to reduce downtime."
So what: High snippetability increases selection in AI answers and featured snippets.
Answer box
Definition: An answer box is a 40–60 word summary placed near the top designed for copy/paste retrieval.
Example: A single paragraph that defines a concept, states the decision rule, and includes a constraint.
So what: This is the most "quotable" unit on a page.
Source grounding
Definition: Source grounding is when an AI ties its claims to retrieved sources rather than model memory.
Example: It cites a page for "DoH vs DoT" instead of guessing.
So what: Grounding is where citations come from; your goal is to be the grounded source.
Hallucination pressure
Definition: Hallucination pressure is when the system is forced to guess because sources are missing, vague, or contradictory.
Example: A page that never gives concrete steps causes the model to improvise.
So what: Reduce hallucination pressure with explicit steps, examples, and definitions.
Query framing
Definition: Query framing is structuring queries with entities, constraints, and context to force better retrieval.
Example: "DoH vs DoT for enterprise networks: performance, policy control, and security tradeoffs."
So what: Framing determines what sources get pulled in.
Query expansion
Definition: Query expansion is adding related terms and synonyms to broaden retrieval coverage.
Example: "WHOIS privacy" + "domain privacy" + "redacted WHOIS" + "ICANN policy."
So what: Expansion helps you cover variations and long-tail prompts.
Freshness bias
Definition: Freshness bias is a preference for newer sources when topics change quickly.
Example: A "2026 marketplace" update can outrank and out-cite a "2024" guide.
So what: Refresh top citation surfaces on a predictable cadence.
Content decay
Definition: Content decay is performance loss over time as information becomes outdated or competitors publish stronger answers.
Example: A "2025 trends" page stops being cited when 2026 sources exist.
So what: Refresh prevents decay and preserves citation gravity.
Provenance
Definition: Provenance is the traceable origin of a claim: who said it, where, and when.
Example: Citing official protocol docs, policy pages, or primary sources for rules and standards.
So what: Strong provenance increases trust and citation likelihood.
Attribution likelihood
Definition: Attribution likelihood is how often an AI will name your brand when it uses your content.
Example: A named framework and a clearly branded definition increases attribution vs generic phrasing.
So what: Attribution is the difference between "being used" and "being credited."
Link adjacency
Definition: Link adjacency is how close an internal link is to the extractable answer text.
Example: A link placed immediately after an answer box is more likely to be followed than one in a footer.
So what: Adjacency is how you route new pages through citation surfaces.
Distribution hub
Definition: A distribution hub is a page designed to send authority, crawl, and users into related pages.
Example: A DNS hub that links to TTL, MX priority, propagation, and troubleshooting guides.
So what: Hubs turn one strong surface into many strong pages.
Topical moat
Definition: A topical moat is owning a dense, interlinked cluster so retrieval and citations default to your site.
Example: Multiple DNS pages with consistent internal linking and distinct coverage.
So what: Moats reduce competitor displacement.
Zero-click capture
Definition: Zero-click capture is winning exposure through AI answers/snippets even when users don't click.
Example: Being cited or named inside the AI response.
So what: Visibility becomes brand demand and downstream conversion.
AI Labs
Definition: NRLC AI Labs is the category layer for retrieval, agent, and schema infrastructure — the research and implementation practice that lets machines understand, retrieve, cite, compare, localize, and act on enterprise data.
Example: A team maps diagnostics findings to AI Labs pillars before scoping schema or agent work.
So what: AI Labs organizes infrastructure work into intelligence, agents, and schema categories.
Pillar: AI Labs guide →
LLM Compute Leak
Definition: LLM Compute Leak is the observable exposure of market demand, intent, brand preference, and citation pathways through AI retrieval behavior across answer surfaces, prompt fan-out, and agentic browsing. NRLC strategic framing unless separately documented as a proprietary product.
Example: Prompt fan-out reveals related sub-queries where a brand never enters the retrieval window.
So what: Useful for prioritizing where retrieval infrastructure investment matters most.
Pillar: LLM Compute Leak guide →
AI Agent Optimization
Definition: AI Agent Optimization is preparing entities, products, locations, services, pricing, availability, policies, and transaction paths so autonomous agents can discover, verify, compare, and complete actions without ambiguous inference.
Example: A product page with stable SKU IDs, currency-aware offers, and explicit checkout routes.
So what: Reduces inference cost and failed agent actions on commerce and service paths.
Pillar: AI Agent Optimization guide →
Enterprise Schema Implementation
Definition: Large-scale schema implementation is enterprise entity architecture: JSON-LD and source-of-truth records for catalogs, locations, services, people, offers, reviews, currencies, languages, and regions.
Example: Shade-level Product nodes with per-locale WebPage canonicals and store-level Offer availability.
So what: Schema at catalog scale prevents entity conflation in retrieval and agentic commerce.
Agent action route
Definition: An agent action route is the machine-readable path from discovery to completion — compare, verify stock, book, checkout, or procure — without human interpretation of ambiguous HTML.
Example: Service lookup → policy confirmation → booking URL with structured availability.
So what: Broken routes cause agent failures even when content ranks well.
Pillar: Agent action route guide →
Entity and source architecture
Definition: Entity and source architecture is governed entity IDs, JSON-LD graphs, and authoritative source pages that define how AI systems interpret products, services, people, and places as verifiable facts.
Example: A catalog where every variant has a stable @id aligned to its canonical source page.
So what: Foundation for retrieval, citation, and agent verification.
Currency-aware product schema
Definition: Currency-aware product schema is Product and Offer markup with explicit currency, regional price, and locale alignment so retrieval systems cite the correct regional price without conflating entities.
Example: Offer.priceCurrency and Offer.price per market on variant-level Product nodes.
So what: Prevents wrong-price citations and agent checkout errors in multi-market catalogs.
Language-aware entity schema
Definition: Language-aware entity schema uses localized names, descriptions, and per-locale canonical URLs so retrieval systems compare the correct regional entity.
Example: Product.name and WebPage inLanguage per locale with hreflang-aligned canonicals.
So what: Reduces cross-locale entity conflation in AI answers and agent comparisons.