Glossary

Working definitions as this book uses them. Terms link to the chapter that treats them in depth.

AEO (Answer Engine Optimization) — Optimizing to be the cited answer in AI-generated responses (AI Overviews, ChatGPT, Perplexity). Overlaps GEO; AEO emphasizes question-shaped queries and answer-first content. → AI Search

AI Overviews / AI Mode — Google's AI-generated answer layers. Overviews appear above classic results; AI Mode (udm=50) is the fully conversational surface. Both draw on Google's index. → How discovery works

Answer engine — A system that returns one synthesized answer with citations instead of a ranked list of links.

Answer-first content — Page structure where the direct answer appears in the first block, self-contained enough for an LLM to quote, with elaboration after. → Content that gets cited

Baseline — The metrics snapshot (clicks, impressions, position, AI-visibility spot checks) recorded before making changes, without which you can't attribute results. → Measurement

Bot challenge — A WAF/CDN interstitial (CAPTCHA, JS proof-of-work, 429) served to suspected bots. When it fires on Googlebot or GPTBot, your site silently vanishes from search and AI answers. → Rendering & WAFs

Canonical URL — The URL you declare as the authoritative version of a page (rel=canonical). Reverse proxies that rewrite the Host header are notorious for emitting canonicals pointing at internal hostnames. → Reverse proxies

Citation — An answer engine's attribution of a claim to a source, usually a link. The GEO equivalent of a ranking. → GEO fundamentals

Crawl budget — How much a crawler is willing to fetch from your site per unit time. Mostly a large-site concern; small sites' problem is usually crawl permission, not budget.

Crawlability — Whether a bot can fetch and read your pages at all: robots permissions, no challenges, server-rendered content. The gate every other tactic sits behind. → AI crawlers

DMARC / DKIM / SPF — The three DNS-based email authentication standards. Together they decide whether your mail lands in inboxes; they're also a trust signal for your domain generally. → Email trust

DNS-AID — An emerging pattern for resolving a domain to its agent/MCP endpoint via DNS records (SVCB + TXT + _index._agents + TLSA/DANE under DNSSEC), giving agents a DNS-native trust path. → Manifests and DNS-AID

Domain property (sc-domain) — A Search Console property type covering every protocol/subdomain of a domain, verified via DNS. Required for whole-domain visibility; supports service-account API access. → Search Console

E-E-A-T — Experience, Expertise, Authoritativeness, Trustworthiness: Google's framework for evaluating source quality. Not a metric you set — an impression your evidence earns. → Entities and trust

Entity — A distinct "thing" (organization, person, product, place) that engines model beyond keywords. Consistent structured data and sameAs links teach engines your entity. → Entities and trust

GEO (Generative Engine Optimization) — Optimizing so generative AI systems cite your property for category and intent queries, not just brand queries. → GEO fundamentals

GBP (Google Business Profile) — Google's local business listing system; feeds the map pack, local knowledge panels, and local AI answers. → Business Profile

GPTBot / OAI-SearchBot / ChatGPT-User — OpenAI's three crawlers: training-data collection, search indexing, and live user-initiated fetches respectively. Each honors robots.txt separately. → Crawler registry

Host-rewriting proxy — A reverse proxy that changes the Host header before your app sees it (common in multi-tenant platforms). The root cause of a whole family of robots/canonical/sitemap bugs. → Reverse proxies

IndexNow — A push protocol (adopted by Bing and others) that notifies engines of changed URLs instantly instead of waiting for a crawl. → IndexNow

Indexing API (Google) — Google's push API, officially scoped to JobPosting and BroadcastEvent pages only — not a general "index my site faster" tool. → Search Console

Intent query — A search expressing a need ("connect custom domains to my SaaS") rather than a name ("customdomain.ai"). The valuable, contested half of discovery. → Keyword strategy

JSON-LD — The recommended syntax for schema.org structured data: a <script type="application/ld+json"> block, independent of your visible HTML. → Structured data

@graph / @id — JSON-LD mechanics for shipping multiple cross-referencing nodes in one block; @id lets a page's Service point at the sitewide LocalBusiness node without repeating it. → LocalBusiness schema

llms.txt — A proposed markdown index file for LLM consumption. Reality check: most deployments get zero bot requests and Google says it doesn't use it; it earns its keep mainly on docs sites read by coding agents. → llms.txt

Map pack — The 3-result local block in Google results, driven by GBP data, proximity, and reviews rather than classic page ranking.

MCP (Model Context Protocol) — The open protocol AI agents use to discover and call tools. Your MCP server is a product surface for agents the way your website is for humans. → AI Agents

MCP Registry — registry.modelcontextprotocol.io, the canonical listing of MCP servers, mirrored by community directories (mcp.so, Smithery, Glama, PulseMCP). → MCP Registry

NAP consistency — Name, Address, Phone appearing identically everywhere (site, GBP, directories). Inconsistency fragments your local entity.

Noindex — A directive (meta tag or header) telling engines not to index a page. Several CMS settings set it implicitly — the trap catalog is longer than you'd think. → Sitemaps and robots

OfferCatalog — The schema.org structure for a tiered pricing catalog. Must hang off a Service node (hasOfferCatalog) — not SoftwareApplication. → Rich results

Answer-time retrievability — Whether an engine can fetch your page at the moment it composes an answer. Distinct from being indexed; blocked live-fetch bots (ChatGPT-User, Claude-SearchBot) kill citations even for indexed pages. → AI crawlers

Rich results — Enhanced result presentations (stars, FAQs, software cards, breadcrumbs) earned by specific schema types passing eligibility rules. → Rich results

robots.txt — The crawler permission file at your domain root. The first thing to check in any invisibility investigation — including whether the served file matches the one you think you deployed. → Sitemaps and robots

sameAs — The schema.org property linking your entity to its profiles elsewhere (GitHub, LinkedIn, social, registries). How engines consolidate your identity. → Entities and trust

Search grounding — An AI assistant running live web searches to inform its answer. The dominant route to being cited today. → How AI finds and cites

SERP — Search Engine Results Page.

Service-area page — A local page targeting "service + city". Legitimate when the business truly serves that city with evidence; a spam pattern when mass-produced. → Service-area pages

server.json — The manifest describing an MCP server for the registry: name (reverse-DNS namespace), endpoints, auth, description. → MCP Registry

Tokenization crisis — When a brand name made of generic words (e.g. a dotted domain like "customdomain.ai") gets split into its parts by the engine, making even branded search fail. → Keyword strategy

UnitPriceSpecification — The schema.org structure for recurring prices ("$249 per month" as price + referenceQuantity with P1M). → Rich results

URL Inspection API — Search Console's per-URL diagnostic API: index status, detected canonicals, structured-data errors, as Google sees them. → Search Console

Vacant niche — A query space with real intent and no committed owner — winnable in weeks while head terms take years. The core of realistic keyword strategy. → Keyword strategy

.well-known — The URL path convention (/.well-known/...) for machine-readable discovery documents: OAuth metadata, agent manifests, security.txt. → OAuth discovery