GEO & AI Visibility Glossary

GEO — Generative Engine Optimization

The practice of improving how generative AI engines (ChatGPT, Perplexity, Gemini, Google AI Overviews) present and recommend a brand in their answers. GEO is to AI answers what SEO is to blue-link results: it targets the citations, entities and sentiment the models draw on at inference time, not keyword rankings.

AEO — Answer Engine Optimization

The discipline of structuring content so answer engines can extract, quote and cite it — self-contained answers, question-led headings, schema markup. AEO and GEO overlap heavily: AEO emphasizes the content formatting; GEO names the whole practice.

AI Visibility

Whether and how a brand appears in AI-generated answers — the measurement layer of GEO, also called 'AI search' when referring to the surfaces themselves. AI visibility counts appearances (mentions and citations); by itself it says nothing about whether the AI actually recommends the brand, which is why visibility metrics pair with recommendation-level metrics like Recommendation Rate.

LLM SEO / AI SEO / ChatGPT SEO / LLMO

Interchangeable industry names for Generative Engine Optimization — including ChatGPT SEO, the ChatGPT-specific framing of the same discipline. Whichever label you use, the work is the same: improve the sources and signals AI engines draw on so they surface (and ideally recommend) your brand.

AI Share of Voice

How often a brand appears in AI answers relative to named competitors: brand appearances ÷ total category appearances × 100. The most widely used AI-visibility metric — and the most often misread, because an 'appearance' can be a caveated mention rather than an endorsement.

Share of Model

A brand's overall presence within one model's answers for a category — its 'mind share' inside the LLM, measured across many prompts. Observed per engine, because models frequently disagree about the same category.

Answer Inclusion Rate

The percentage of relevant queries where a brand makes it into the AI's final answer at all. An inclusion-level metric: it counts presence, not position or endorsement.

Share of Answer

How much of the answer experience a brand owns — placement, space and prominence within the response, not just whether it appears. A finer-grained cousin of AI Share of Voice.

Recommendation Rate

The percentage of AI answers that actively recommend a brand (endorse it as a choice) rather than merely mention or cite it. Visibility metrics count appearances; Recommendation Rate counts advice. A brand can hold a high AI Share of Voice and a near-zero Recommendation Rate. Not to be confused with Recommendation Share: Rate is how often you're recommended; Share is your slice of all recommendation slots in a category.

Recommendation Share

A brand's proportion of total recommendation slots across a category, measured across repeated runs — AI Share of Voice counted only when the AI actually recommends. Expressed as a percentage; categories typically show concentration in the top 2–3 brands.

Mentioned vs Recommended

The core distinction RecommendHQ measures. A mention is any appearance in an AI answer; a citation means the AI drew on your source but didn't necessarily advise the reader to choose you; a recommendation is when the answer actively endorses the brand as the pick ('go with X'). Visibility tools stop at mentions and citations — Recommendation Rate and the Visibility Spectrum measure whether you cross into 'Recommended'.

R-Score

A 0–100 composite score expressing how likely an LLM is to recommend a specific brand in response to a buying-intent query. Computed per surface, then blended across surfaces using surface-level weights. Scores typically fall in these ranges by Visibility-Spectrum tier (80–100 Recommended, 55–79 Cited, 30–54 Mentioned, 0–29 Ghosted), but the tier itself is set by the evidence in the answer, not by the band: 'Cited' requires your domain among the cited sources and 'Recommended' requires an explicit endorsement.

ARM — AI Recommendation Management

What RecommendHQ does — the platform layer that manages how AI recommends you, unifying GEO, AEO and AI visibility. It runs one loop with recommendation as the outcome: measure whether models recommend you (not just mention you), diagnose the citation gaps behind a competitor's lead, act on the highest-leverage sources, and prove the movement on the next scan.

ARI — AI Recommendation Index

RecommendHQ's live leaderboard of blended R-Scores for top B2B software brands, aggregated across all four surfaces — ChatGPT, Perplexity, Gemini and Google AI Overviews. Refreshed weekly; the definitive snapshot of who AI currently recommends and how confidently.

Visibility Spectrum

The four-tier model describing a brand's position in LLM output, worst to best: Ghosted → Mentioned → Cited → Recommended. Your tier is derived from what the AI actually did, not from your score: 'Cited' means your domain appears among the sources it drew on; 'Recommended' means the answer explicitly endorsed you. Scores typically fall in these ranges (Ghosted 0–29, Mentioned 30–54, Cited 55–79, Recommended 80–100), and each tier carries a distinct recommended action set. A brand that becomes the model's default, near-unchallenged answer may also earn a separate Canonical seal in the ARI — a distinction, not a fifth tier.

Canonical (seal)

The gold 'Recommended' / authority marker shown when a brand is actively and consistently recommended by the models (R-Score 80–100). The Canonical seal asserts that AI treats the brand as the default, near-unchallenged answer in its category — a distinction earned at the top of the Recommended tier, not a separate score or a fifth tier.

Surface

A distinct AI destination where a buyer can ask for a recommendation and get an answer. RecommendHQ tracks four grounded surfaces: ChatGPT, Perplexity, Gemini and Google AI Overviews. Google AI Overviews (the AI summary inside Google Search) is a separate surface from the Gemini app even though both use Google models, because each retrieves and ranks sources independently. The R-Score is computed per surface, then blended.

Prompt

A buying-intent query we run against the surfaces on a brand's behalf — the kind of question a real buyer asks, such as 'best payment processing platform'. The prompt is RecommendHQ's core unit of measurement: every R-Score, citation and recommendation is observed in response to one. Plans include a set number of tracked prompts per brand, and you choose the category queries that matter to your buyers.

Locale

A market-and-language combination a prompt is run in — for example us-en (United States, English). Because AI assistants tailor answers to a buyer's region, the same prompt can surface different brands across locales. Add locales when you sell into more than one market and the AI answers differ by region.

Brand (brand domain)

One tracked domain or workspace — typically a single company or product you're monitoring. The brand is RecommendHQ's billing unit: each plan includes a set number of brands (Starter 1, Growth 3, Agency 10, Command 40), with additional brands available at a flat per-brand rate.

AI Recommendation Audit

A point-in-time report on how AI recommends a single brand: its blended R-Score across the grounded AI surfaces, where it sits on the Visibility Spectrum, and the citation gaps competitors own. Agencies use audits as the opening move in a pitch — showing a prospect exactly where AI recommends a rival instead of them. Sold as one-off audits in the Pitch Pack (one audit = one prospect domain) with a white-label-ready PDF, no subscription required.

Citation Gap

The delta between a competitor's citation-source footprint and your brand's footprint, on a per-surface basis. Closing the citation gap on authoritative sources (G2, Reddit, Capterra, industry publications) is the highest-ROI lever in GEO.

Action Engine

The system that turns citation gaps into a prioritized, quantified action plan. Each opportunity is ranked by leverage (citation frequency × achievability × impact × breadth) and sorted into effort and impact bands, so you know what to tackle first and what it's worth. It prescribes legitimate Entity Authority only, never astroturfing or fake reviews.

Grounding Signal

Data originating from a retrieval-augmented generation (RAG) surface where the LLM cites live content at inference time. Grounded signals are immediately influenceable — publishing or updating a cited source affects scores within one scan cycle. Contrast with Brand Equity (parametric weight).

Brand Equity (parametric signal)

The model's built-in, 'parametric' familiarity with a brand — the knowledge baked into its weights during pre-training, before any live retrieval. Brand Equity is why well-established brands can surface even without a fresh citation, and it shifts only slowly. Contrast with grounded Grounding Signals, which you can influence within a single scan cycle.

LLM-as-a-Judge

An evaluation method where a large language model scores or classifies text against a fixed rubric, standing in for manual human review at scale. RecommendHQ uses an LLM-as-a-Judge to grade Semantic Sentiment: it reads each AI answer and classifies how the brand is portrayed — explicit endorsement, secondary option, caveated warning, or absent — on a consistent scale. It is the measurement technique behind the Sentiment signal in the R-Score.

Semantic Sentiment

The tone and confidence of a brand's portrayal in LLM outputs, graded by an LLM-as-a-Judge on a 4-tier scale: explicit endorsement, secondary option, caveated warning, or absent entirely. Semantic Sentiment is one of the four signals that compose the R-Score, and on scan reports the four tiers surface as the matching Visibility Spectrum labels: Recommended, Cited, Mentioned and Ghosted.

Entity Authority

The strength of a brand's structured and unstructured footprint in sources that LLMs retrieve and cite — Wikipedia, authoritative directories, original research, and high-velocity community platforms. Building Entity Authority is the primary organic lever in GEO, analogous to domain authority in classic SEO.

The GEO & AI visibility dictionary.

GEO & AI visibility terms