SEO & GEO Glossary

How should you use this glossary?

Start with a category, then open an entry to check its definition, practical use and common misconceptions. Related-term links provide a navigable concept path.

GEO

• GEO (Generative Engine Optimization): GEO (Generative Engine Optimization) is the practice of optimizing content for generative AI search engines like ChatGPT, Perplexity, and Google AI Overviews so a brand gets cited inside AI answers. Core tactics include Answer Capsules, entity consistency, structured data, and AI-readable llms.txt files.
• Answer Capsule: An Answer Capsule is a 40–60 word direct-answer block placed within the first 200 words of an article, structured as 'definition + key action'. It is the highest-ROI GEO writing pattern because ChatGPT, Perplexity, and Google AI Overviews preferentially extract self-contained, copy-pasteable sentences as citations.
• Entity Consistency: Entity Consistency means a brand, founder, and product use the same name, description, and links across Wikidata, Wikipedia, G2, Product Hunt, LinkedIn, etc. AI models run entity coreference resolution to decide whether sources refer to the same thing — inconsistency dilutes brand authority.
• Google AI Overviews: AI Overviews (formerly SGE) is Google's AI-generated answer at the top of SERPs, citing 3–10 sources and rolled out in 100+ countries. Cited links see ~40% higher CTR than classic blue links; uncited pages have lost ~30% of traffic on average.
• Google AI Mode: AI Mode is Google's full-AI search experience launched in 2026: an entrance in the top-left, results fully generated by Gemini with 3–10 citations. Unlike AI Overviews, AI Mode is a standalone conversational SERP with no blue links mixed in.
• AEO (Answer Engine Optimization): AEO (Answer Engine Optimization) is a subset of GEO focused on getting content extracted verbatim by answer engines like ChatGPT, Perplexity, AI Overviews. The core is writing self-contained capsule paragraphs plus FAQ schema.
• Answer Engine: An Answer Engine is a search system that returns generated answers instead of link lists — Perplexity, ChatGPT Search, Google AI Mode. They use RAG to fetch live web pages and cite 3–10 sources per answer.
• Generative Search: Generative Search is the paradigm of using LLMs to synthesize search results in real time, replacing the classic 'retrieve-then-rank' model. Examples: Google AI Overviews/Mode, Perplexity, Bing Copilot, Baidu AI Search. By 2026 it covers >60% of mainstream queries.
• Grounding: Grounding is the practice of basing LLM answers on verifiable sources (instead of free-form generation) via RAG, knowledge graphs, or citation injection. Google AI Overviews' 'search-grounded' mode is a textbook example.
• Knowledge Graph: A Knowledge Graph is a structured entity-relation network (company → founder → product → competitor). Google's KG powers AI Overviews and Knowledge Panels. Keys to inclusion: a Wikidata QID and consistent sameAs links.
• Knowledge Panel: A Knowledge Panel is the entity card on the right of a SERP (company/person/product), sourced from Google's Knowledge Graph. Having a panel means AI engines are more likely to recognize your brand entity correctly.
• Hallucination: Hallucination is when an LLM produces factual errors or fabricated citations. Brand risk: wrong numbers cited, founders misattributed, fictional features invented. Grounding + official llms.txt + Wikidata calibration are the main antidotes.
• Cross-Engine Citation: Cross-Engine Citation is when the same prompt sees the same source cited across ChatGPT, Perplexity, Gemini, Claude, and AI Overviews. 3+ engines citing = strong authority signal, often inherited by new engines on launch.
• Agent Readiness: Agent Readiness measures how callable a site is to AI agents (Google Search Agent, WebMCP, ChatGPT Agent) — semantic forms, price schema, callable-action JSON-LD, policy docs. A new 2026 dimension.
• ChatGPT Search: ChatGPT Search is OpenAI's real-time web search integrated into ChatGPT since 2024, available to Plus/Pro/free users by default. Crawler UA: OAI-SearchBot. SEO implication: robots.txt must Allow OAI-SearchBot, or you permanently vanish from ChatGPT answers.
• Doubao (ByteDance AI): Doubao is ByteDance's AI assistant, a 2026 top-3 Chinese-mainland AI search by share. Crawler UA: Bytespider. Chinese GEO must Allow Bytespider in robots.txt and ship Chinese-language schema or you lose AI answer exposure in China.
• Kimi (Moonshot AI): Kimi is Moonshot AI's Chinese assistant, known for ultra-long context (2M chars) and high-quality citations. GEO preference: long-form Chinese + academic/government/authoritative-media sources + explicit dates. Crawler UA: Moonshotbot.

SEO

• SEO (Search Engine Optimization): SEO (Search Engine Optimization) uses three pillars — technical, content, and links — to earn higher organic rankings on Google and Bing. By 2026 SEO has merged with GEO: the same article must satisfy both crawlers and AI citation engines.
• E-E-A-T: E-E-A-T are the four quality signals from Google's Search Quality Rater Guidelines: Experience, Expertise, Authoritativeness, Trustworthiness. Trust is the foundation — without it, the other three count as zero.
• Topical Authority: Topical Authority is Google's overall judgment of how comprehensively you cover a topic. The more complete the coverage, the tighter the internal links, and the higher the originality, the stronger the authority. It has overtaken single-page PageRank as a core ranking factor since 2024.
• Internal Linking: Internal Linking refers to the link structure between pages on the same site. It governs PageRank flow, crawler discovery efficiency, and topical relatedness. In the GEO era it also drives entity coreference — AIs use anchor text to decide whether two pages describe the same entity.
• Anchor Text: Anchor text is the visible text of a hyperlink. Google uses it to infer the target page's topic; over-optimization (a single keyword anchor exceeding ~30%) can trigger a Penguin-style demotion. In GEO it also influences AI entity resolution.
• Backlinks: Backlinks are inbound links from other sites — the original PageRank signal. Quality beats quantity: one DR-80 media link is worth more than a thousand directory links. In the GEO era backlinks also serve as authority endorsements that make AIs more willing to cite you.
• SERP (Search Engine Results Page): A SERP is the page a search engine returns. By 2026 it mixes 10 blue links with AI Overviews, Featured Snippets, People Also Ask, video carousels, local packs, shopping packs — 20+ elements in total. The more SERP features, the lower the blue-link CTR.
• Featured Snippet: A Featured Snippet is the 'position 0' answer box at the top of a SERP, extracting a 40–60 word direct answer from the top 10 results. Optimization: place a 'definition + list / table' structure near the top of the article that directly answers the main keyword.
• Title Tag: The Title Tag lives inside <title> and shows on the SERP blue link and browser tab. Google displays about 60 characters, truncating beyond. Best structure: 'main keyword | value prop | brand'. As of 2026, Google still rewrites titles ~60% of the time.
• Meta Description: Meta Description sits in <meta name="description"> and shows beneath the SERP blue link, about 160 characters. It doesn't directly affect ranking but materially affects CTR. Google rewrites it from body content roughly 70% of the time.
• Semantic Search: Semantic Search returns results by understanding intent and context, not literal keyword matches. Google BERT (2018) and MUM (2019) were the watershed. Content implication: cover complete concepts and synonyms, never keyword-stuff.
• Entity Salience: Entity Salience measures how 'central' an entity is to a document, computed from frequency, position (first-paragraph weighting), and contextual co-occurrence. Google's NLP API outputs a salience score (0–1) — a core topical-relevance signal.
• Brand Mentions: Brand Mentions are textual references to your brand on other sites (link optional). Google treats 'linkless mentions' as implicit link signals; AI engines use them to train entity co-occurrence, boosting citation rate.
• Passage Indexing: Passage Indexing is Google's 2020 capability to index individual passages of long pages, surfacing deep-buried answers for long-tail queries. In the GEO era it mirrors Chunking thinking: write for passages, not just pages.
• Helpful Content System: Google's core system since 2022 that demotes content 'written for search engines, not people'. Signals: first-hand experience, original insight, clear audience, real reader value. AI-generated content farms are the primary target.
• Spam Policies: Google's 2024 spam additions: Scaled Content Abuse (mass low-quality), Site Reputation Abuse (parasite SEO), Expired Domain Abuse (recycled aged domains). A hit = manual sitewide penalty, 3–9 month recovery.
• Scaled Content Abuse: Scaled Content Abuse is mass-producing pages with no unique value (AI or human) to manipulate search. Formally a policy violation since Google's Spring 2024 update. Core test: does each page add Information Gain?
• Keyword Cannibalization: Keyword cannibalization describes overlapping pages that make the intended search destination or its performance unclear; it is not a Google-defined penalty. Several URLs receiving impressions for one query is not enough to diagnose harm. Compare intent, canonical selection, clicks and conversions before differentiating or consolidating pages.
• Digital PR: Digital PR earns media coverage + backlinks via news releases, data reports, expert commentary. The mainstream compliant way to acquire quality links and brand mentions, and a key signal source for AI engine entity training.

Content

• Topic Cluster: A Topic Cluster is a Pillar Page plus N subtopic pages connected by two-way internal links. It signals to Google that you cover a topic comprehensively and is the standard tactic for building Topical Authority.
• Search Intent: Search Intent is the real purpose behind a query. Four buckets: Informational (learn), Navigational (locate), Commercial (compare), Transactional (buy). If you mismatch intent, no amount of links will rank you.
• Keyword Research: Keyword Research is the process of discovering what the target user searches for, at what volume, with what competition, and what intent. The output is a Keyword Matrix: term + volume + KD + intent + pillar/subtopic mapping.
• Long-tail Keywords: Long-tail keywords are 4+ word, low-volume (<1000/mo), high-intent, low-competition queries. Individually small, collectively 70% of all search traffic. In the GEO era they matter even more — AI-bound queries are typically question-style long-tails.
• People Also Ask: People Also Ask is the expandable related-questions box on a SERP — each click dynamically loads 3–5 fresh questions. It's a free keyword goldmine; our Keyword Matrix module scrapes it in one click.
• Content Decay: Content Decay is the gradual traffic drop a few months after publishing, caused by stale info, fresher competitors, or SERP feature shifts (AI Overviews stealing clicks). Diagnose via GSC: a >30% drop comparing month 3 vs month 6 = decay.
• Information Gain: Information Gain is the score from Google's 2022 patent: how much 'new information' a page adds relative to already-indexed content. Generic rewrites score near 0; first-party data, exclusive cases, and original experiments score high and rank first.
• YMYL (Your Money or Your Life): YMYL covers topics that may affect a reader's finances, health, safety, or legal status. Google enforces the strictest E-E-A-T here: real bylined authors, credentials, verifiable sources. AI engines are also more conservative when citing YMYL content.
• Author Bio: Author Bio is the author block at article end — since 2024 a mandatory E-E-A-T and GEO signal. Elements: real name, photo, 3–5 line credentials, Person schema + sameAs to LinkedIn/X/Wikidata.
• Programmatic SEO: Programmatic SEO uses templates + databases to mass-generate long-tail pages (industry × city × tool). Post-2024 it must be paired with Information Gain or it triggers Scaled Content Abuse. Safe ratio: ≥1 exclusive data paragraph per template + 10% human review.
• Topical Map: A Topical Map is a site's topic coverage blueprint — a Pillar Page at the center radiating N subtopics. A complete Topical Map signals to Google and AI engines that you are 'the authority on this domain', not 'someone who wrote a few posts'.
• Pillar Page: A Pillar Page is a long survey covering a broad topic (3,000–8,000 words), linking to N subtopic pages and back-linked by them. The standard structure for building Topical Authority and GEO entity salience.
• AI Content Detection: AI Content Detection uses tools like GPTZero, Originality, Copyleaks to spot AI-generated text. Google officially does not penalize 'AI-generated' per se but indirectly demotes low-quality AI mass production via Helpful Content and Scaled Content systems.

Technical

• llms.txt: llms.txt is a plain-text manifest at a site's root that lists key pages, APIs, and docs in Markdown for AI crawlers to index efficiently. Proposed by Anthropic in 2024, it has become the de-facto 'robots.txt for the LLM era'.
• Core Web Vitals: Core Web Vitals are Google's three UX performance metrics: LCP (Largest Contentful Paint ≤2.5s), INP (Interaction to Next Paint — Good ≤200ms / Needs Improvement 200-500ms / Poor >500ms), CLS (Cumulative Layout Shift ≤0.1). All three must be green to qualify as 'Good' and affect mobile ranking.
• Schema.org Structured Data: Schema.org is a structured-data vocabulary jointly maintained by Google, Bing, Yahoo, and Yandex. Once annotated via JSON-LD, search engines and AI models can directly understand whether a page is an Article, Product, FAQ, or Organization — powering rich results and precise citations.
• JSON-LD: JSON-LD (JSON for Linking Data) is Google's recommended format for structured data, embedded via <script type="application/ld+json">. Compared with Microdata / RDFa, it is decoupled from HTML and the easiest to maintain.
• robots.txt: robots.txt is a crawler-instruction file at the site root using User-agent / Allow / Disallow to control who can fetch which paths. In the GEO era you must explicitly Allow GPTBot, PerplexityBot, and Google-Extended — otherwise AI engines will not cite you.
• Crawl Budget: Crawl budget is the set of URLs Googlebot can and wants to crawl, influenced by crawl capacity and crawl demand rather than a fixed quota. It mainly matters for large or frequently updated sites. Check discovery and server health first: more crawling does not guarantee indexing or rankings.
• Indexation: Indexation is the process by which a search engine adds a page to its searchable index. Discovered ≠ Crawled ≠ Indexed ≠ Ranked. Google Search Console's Pages report is the diagnostic starting point; the IndexNow protocol (Bing/Yandex) lets you push URLs for faster indexing.
• IndexNow: IndexNow is an open protocol from Bing and Yandex that lets sites POST a list of changed URLs to search engines for near-instant crawling. Google does not yet participate, but Cloudflare, Wix, Yoast, and others have integrated it natively.
• Canonical Tag: A canonical tag uses <link rel="canonical" href="..."> to tell search engines which URL is preferred among a group of similar pages, consolidating ranking signals and avoiding duplicate-content dilution. Common for multilingual sites, parameterized URLs, and AMP mirrors.
• hreflang: hreflang is a <link rel="alternate" hreflang="..."> annotation telling Google about language/region variants of the same content, so an English page isn't served to a Chinese user. Pairs must be bidirectional with an x-default fallback.
• Sitemap.xml: Sitemap.xml is the standard format for submitting URL lists to search engines, with lastmod / changefreq / priority per entry. It doesn't guarantee indexing but accelerates discovery, especially for new sites, weakly-linked pages, and freshly published content.
• Page Speed: Page Speed measures the time from request to interactive, quantified by the three Core Web Vitals (LCP / INP / CLS). Mobile thresholds are stricter — on 3G, LCP should be ≤4s; each extra second adds ~7% to bounce rate.
• Duplicate Content: Duplicate Content refers to multiple URLs that share identical or near-identical body text. Google does not penalize per se, but picks one canonical and ignores the rest, scattering traffic. Common sources: parameterized URLs, www / non-www, HTTP / HTTPS, scraped copies.
• RAG (Retrieval-Augmented Generation): RAG (Retrieval-Augmented Generation) is the core architecture of answer engines: user query → vector-retrieve relevant pages → feed retrieved chunks to an LLM to synthesize the answer. This means content must be crawlable, chunkable, and embeddable.
• Vector Search: Vector Search maps queries and documents into high-dimensional embeddings and retrieves by cosine distance — the mainstream recall layer in RAG and AI engines, complementary to keyword inverted-index search.
• sameAs Property: sameAs is a Schema.org property that links the same entity's multiple web identities via URLs (homepage, Wikidata, LinkedIn, X, Crunchbase). It is the strongest structured signal for AI engines to confirm entity consistency.
• Chunking: Chunking is the preprocessing step where RAG slices a page into 200–500 word fragments for vector indexing — usually along H2/H3, paragraph, and sentence boundaries. Chunker-friendly writing means clear subheadings + complete paragraphs = clean citable blocks.
• llms-full.txt: llms-full.txt is the extended sibling of llms.txt: it concatenates each page's core paragraph into one file so an AI can 'read the whole site' in a single fetch. Ideal for glossaries, API docs, and product manuals.
• Person Schema: Person Schema is the Schema.org JSON-LD type for real people: name, jobTitle, worksFor, knowsAbout, sameAs. The core structured signal Google Knowledge Panels and AI engines use to identify author entities.
• FAQPage Schema: FAQPage is Schema.org's Q&A structured type using Question + Answer pairs. Google narrowed FAQ rich results for non-authority sites in 2023, but AI engines still favor FAQPage as an answer source — a GEO must-have.
• HowTo Schema: HowTo is Schema.org's step-by-step tutorial type. Google retired the desktop rich result in 2023, but AI engines (Perplexity, ChatGPT) still heavily extract HowTo for step-by-step answers. Pair with numbered steps + image placeholders.
• Product Schema: Product Schema annotates a product's name, image, price, availability, aggregateRating, review. Both 2026 ChatGPT Ads and AI shopping results depend heavily on it — missing fields drop you from the product feed.
• Organization Schema: Organization Schema is the homepage's must-have company entity structured data: name, logo, url, sameAs, founder, foundingDate. It is the entry point for Google Knowledge Panels and AI engines to ingest your brand entity.

Metrics

• AI Citation: AI Citation refers to source links or quoted sentences that ChatGPT, Perplexity, and Google AI Overviews surface in their answers. Citation Frequency has become the core brand-visibility metric in the GEO era, replacing rank-and-click as the headline KPI.
• Domain Authority / Domain Rating: DA (Moz) / DR (Ahrefs) are third-party 0–100 authority scores driven mainly by backlink quality and quantity. Not an official Google metric, but widely used as a quick benchmark. A new site reaching DR 20 within 6 months and DR 40 within a year is healthy.
• CTR (Click-Through Rate): CTR = clicks / impressions. In GSC, slice by position: position 1 healthy CTR ≈ 28%, position 5 ≈ 6%, position 10 ≈ 2.5%. Below the industry baseline means Title / Description need a rewrite.
• Citation Frequency: Citation Frequency measures how often a brand/page is cited inside AI engine answers — the core GEO-era metric replacing 'rank' and 'sessions'. Tracked by periodically sampling ChatGPT/Perplexity output via Citation Tracker tools.
• AI Share of Voice: AI SOV measures your brand's share of citations across a tracked prompt set. E.g., for the 'best CRM' cluster with 100 total cites and your brand at 12 = 12% SOV.
• Prompt Stability: Prompt Stability is the consistency of an AI's response to the same prompt across repeated calls. Low stability means brand citations are 'random'; tighter entity signals (Wikidata, official docs, consistent schema) are needed to lock in.
• INP (Interaction to Next Paint): INP is the third Core Web Vital, replacing FID in 2024. It measures the delay between a user interaction and the next paint, covering all clicks/inputs. Good ≤200ms, poor >500ms. Common bottlenecks: heavy JS, third-party scripts, long tasks.
• TTFB (Time To First Byte): TTFB measures request-to-first-byte latency. Good ≤800ms. Driven by CDN edge distance, origin response, TLS handshake, SSR render time — the upstream bottleneck for other Core Web Vitals.