Last updated September 2026.
This glossary defines 40 terms used across AI SEO tools, monitoring platforms, and the wider answer-engine optimization field. Each entry opens with a direct, self-contained definition. Most add one sentence of context or a cross-link to a related term.
Jump to a letter: A · B · C · D · E · F · G · H · I · K · L · M · N · P · R · S · T · V · W · Z
A
AEO (Answer Engine Optimization)
AEO, or answer engine optimization, is the practice of structuring and distributing content so AI answer engines cite your brand, state its facts correctly, and recommend it to users asking relevant questions. It is the direct successor to SEO’s ranked blue links. AEO work centers on citation rate and share of voice, not position on a results page.
Agentic Search
Agentic search describes an AI system that does not just answer a query but takes multi-step action to complete it: comparing options, filling a form, or completing a task with no further human input. A page has to supply facts an agent can act on directly: a price, a spec, a stock status. Not prose a person reads and interprets. Schema markup is one way to supply that.
AI Bot Traffic
AI bot traffic is the set of website requests made by AI crawlers and retrieval systems, such as GPTBot and PerplexityBot, rather than by human visitors or traditional search crawlers. Rising AI bot traffic on a page is an early sign an engine has pulled it into its retrieval window. Scrunch and Temso both report this traffic as a stream separate from human sessions.
AI Crawler
An AI crawler is an automated program an AI company runs to fetch, read, and index web pages for a model’s training data or its live retrieval layer. Blocking a crawler in robots.txt removes a site from that engine’s retrieval pool entirely. Coverage differs by crawler, so a page open to GPTBot can still be closed to PerplexityBot.
B
Brand Mention
A brand mention is any instance where an AI-generated answer names your brand, product, or a defined alias, whether or not it links to your site. Mentions are the rawest AI SEO signal: a simple count, not a rate. Tools track mentions alongside citation rate, because a brand can be named often while earning very few source links.
C
Citation
A citation, inside an AI-generated answer, is a link or explicit source reference that credits a specific URL for a fact, quote, or recommendation. Not every AI answer cites sources. ChatGPT’s default mode often does not, while Perplexity and Google AI Overviews cite far more consistently. Earning a citation is the core goal of most AI SEO work.
Citation Rate
Citation rate is the percentage of tracked prompts in which a specific URL or domain is cited as a source inside the AI-generated answer. A page cited in 12 of 50 tracked prompts holds a 24% citation rate for that set. Otterly.AI and Peec AI both report citation rate at the individual URL level, not only the brand level.
Content Freshness
Content freshness is how recently a page was published or meaningfully updated, a factor several AI retrieval systems weigh when choosing which sources to pull into an answer. A page with accurate but three-year-old figures can lose out to a newer, less authoritative page simply because its numbers read as current. Dating a page’s key facts is a direct freshness signal.
Crawlability
Crawlability is whether a page’s technical setup, robots.txt rules, JavaScript rendering, and server response all allow an AI crawler to fetch and read it successfully. A page can be well written and still invisible to an engine if a blocked directive or a rendering failure keeps the crawler from ever reaching its content.
D
Direct Answer
A direct answer is a self-contained sentence, placed at the top of a page or section, that states a fact or definition without requiring the reader to scroll for context. AI engines quote direct answers more readily than buried explanations, because a synthesized response has to lift a passage whole rather than assemble one from scattered sentences. This glossary’s entry format is built around that principle.
E
E-E-A-T (Experience, Expertise, Authoritativeness, Trustworthiness)
E-E-A-T is Google’s framework for judging content quality: an author’s first-hand experience, subject expertise, site authority, and overall trustworthiness. Google AI Overviews inherits E-E-A-T signals from Google’s underlying index, so a page that fails E-E-A-T for classic search also struggles to surface in Google’s AI answers, even though the two systems otherwise weigh content differently.
Engine Coverage
Engine coverage is the number and identity of AI engines a monitoring tool actually tracks. ChatGPT, Perplexity, Google AI Overviews, Gemini, and Microsoft Copilot are the five most commonly covered. Coverage varies more than buyers expect: Profound’s 2026 analysis of 100,000 prompts found only 11% of the domains cited by ChatGPT also appear in Perplexity’s citations for the same prompts. Tracking one engine tells you little about another.
Entity SEO
Entity SEO is the practice of making a brand, person, or product identifiable to AI systems as a distinct, well-defined thing, usually through consistent naming, structured data, and a stable presence across authoritative third-party sources. A brand that reads as a clear entity is easier for a model to cite correctly and harder to confuse with a similarly named competitor.
F
Fan-Out Query
A fan-out query is an internal sub-query an AI engine generates behind the scenes to gather source material before it writes a response to a user’s actual question. Perplexity and other retrieval-heavy engines rely on fan-out queries that often diverge sharply from the original wording. Optimizing only for the visible query phrase misses a large share of the retrieval window that forms the real answer.
G
GEO (Generative Engine Optimization)
GEO, or generative engine optimization, is the broader discipline of improving how generative AI systems retrieve, describe, and cite a brand across every answer surface, not just Google’s. It overlaps heavily with AEO, and the two terms are often used interchangeably. Where a distinction is drawn, AEO leans toward answer-box optimization, and GEO covers the full range of generative engines, including chat assistants.
Google AI Overviews
Google AI Overviews is the AI-generated summary Google places above standard blue-link results for many search queries, built from Google’s index and, for some queries, live grounding. It is the highest-volume answer surface most brands optimize for, since it appears inside search traffic a brand already receives rather than a separate app a user has to open.
Grounding
Grounding is the technique of connecting an AI model’s response to real, retrievable source documents instead of relying only on patterns learned during training. A grounded answer can name its source. An ungrounded one is reciting from memory, which raises the risk of a hallucination. Google AI Overviews and Perplexity both ground most responses in live web retrieval.
H
Hallucination
A hallucination is a factually wrong statement an AI engine makes with the same confident tone as a correct one, such as an outdated price, a discontinued feature, or a made-up founding date. Hallucinations about a brand persist because they trace back to stale or missing source material. Fixing one usually means correcting the source. See source attribution for how a directory traces it.
I
Index Score
An index score is a single number, typically out of 100, that a directory or platform assigns to a tool by combining several weighted criteria into one figure. This directory’s own index score weighs engine coverage, execution features, ease of use, value, and transparency, and the full breakdown sits on the methodology page. Scores from two different platforms are rarely comparable, since each defines its own weights.
K
Knowledge Graph
A knowledge graph is a structured database of entities, such as brands, people, and products, and the verified relationships between them, which search and AI systems draw on to answer factual questions with confidence. Google’s Knowledge Graph is the best-known example. A populated, accurate knowledge graph entry supports the same goal as entity SEO: fewer chances an AI system confuses a brand with a competitor.
L
Last-Verified Date
A last-verified date is the date a directory or tool profile last confirmed a listed price, feature, or claim was still accurate. This directory prints a last-verified date on every tool entry and re-checks pricing pages monthly and feature claims quarterly. See verification rotation for the full schedule.
LLM (Large Language Model)
An LLM, or large language model, is the type of AI system, trained on large volumes of text, that powers ChatGPT, Gemini, and most other answer engines by predicting and generating natural-language responses. An LLM alone has no live internet access. It is grounding and retrieval-augmented generation that connect it to current, citable facts.
llms.txt
llms.txt is a proposed plain-text file, placed at a site’s root, meant to help AI crawlers find and understand a site’s most important pages faster than crawling alone. Adoption remains uneven, and no major AI engine has confirmed it changes what gets retrieved, so treat it as a helpful signal, not a guarantee.
M
Model Training Cutoff
A model training cutoff is the date after which an AI model has no direct knowledge from its original training data, though live retrieval and grounding can still surface newer facts at answer time. A brand that changed its pricing or name after a model’s cutoff should expect that model to state the old fact until retrieval, not retraining, corrects it.
N
NER (Named Entity Recognition)
NER, or named entity recognition, is the automated process a monitoring tool uses to detect and extract brand names, product names, and competitor names from AI-generated text so it can count brand mentions at scale. NER accuracy varies by tool. An abbreviated brand name or an informal product nickname can be missed, which understates a brand’s real mention count.
P
Prompt
A prompt is the exact question or instruction a user, or a monitoring tool acting on a user’s behalf, sends to an AI engine. Two prompts asking the same real-world question in different words can return different brands, so a single prompt is one sample, not a reliable measurement on its own. See prompt monitoring for how tools account for that.
Prompt Monitoring
Prompt monitoring is the practice of running a defined set of prompts against one or more AI engines on a regular schedule and recording which brands and sources appear in the answers. It is the data-collection method behind every metric in this glossary, from citation rate to share of voice. Temso and Peec AI both run prompt monitoring from the real chat interfaces, not a raw API call alone.
Prompt Set
A prompt set is the defined group of query variants a monitoring program tracks for a given brand, chosen to reflect how real buyers actually phrase the same underlying question. A prompt set built from three or four wordings of one intent produces a far steadier read on share of voice than a single phrasing run once.
R
RAG (Retrieval-Augmented Generation)
RAG, or retrieval-augmented generation, is the architecture in which an AI system first retrieves relevant documents from the live web or a stored index, then uses those documents as context to write its answer. Perplexity, Google AI Overviews, Microsoft Copilot, and the browsing modes of ChatGPT all rely on RAG. A page has to enter the retrieval window before it can ever be cited.
Retrieval Window
The retrieval window is the set of documents an AI engine’s retrieval system pulls as candidate context for a prompt, before the model decides which of them, if any, to reference in the final citation. A page can enter the retrieval window and still be dropped from the answer if a competing source reads as more authoritative or more current.
S
Schema Markup
Schema markup is structured code, most often in JSON-LD, added to a page’s HTML to label its content in a machine-readable vocabulary that search and AI systems can parse directly instead of inferring meaning from prose. FAQ, Product, and Article schema are the types most directly useful for AI citation, since they hand an engine a ready-made fact instead of a paragraph to interpret.
Sentiment Analysis
Sentiment analysis is the classification of an AI engine’s language about a brand as positive, neutral, or negative, covering how it describes pricing, ease of use, support, and competitive standing. AI sentiment is not customer sentiment. A single retrieved article can swing an engine’s framing even when most real reviews are positive.
Share of Voice
Share of voice in AI search is the percentage of relevant prompts, across a defined engine set and competitive group, in which a brand is named at all. A brand appearing in 30 of 100 tracked prompts holds 30% share of voice for that set. It is the closest AI-era equivalent to a traditional search ranking.
Site Audit
A site audit, in an AI SEO context, checks a domain’s technical readiness for AI crawlers and citation: robots.txt rules, schema markup, page load behavior, and content structure. Brightedge and Scrunch both run site audits as a standing module rather than a one-time report, flagging new blockers as a site changes instead of only at signup.
Source Attribution
Source attribution identifies which domains are actually generating a brand’s citations inside AI-generated answers: owned pages, press coverage, review sites, or forum threads. A brand whose citations concentrate on two or three domains is more exposed to a single outage or removed article. See third-party authority for how that concentration risk gets earned back.
T
Third-Party Authority
Third-party authority is the degree to which independent sources, such as review sites, journalists, and forums, describe a brand in ways an AI engine treats as credible enough to cite. A separate Ahrefs study found only 13.7% overlap between the domains cited in Google AI Overviews and Google AI Mode for the same queries. Authority has to be earned engine by engine, not once.
Training Data
Training data is the large body of text an AI model learned from before release, which shapes its default assumptions about a brand until grounding or retrieval supplies newer information. A brand with little presence on the public web during a model’s training window starts every conversation about it from a thin, generic baseline.
V
Verification Rotation
A verification rotation is the fixed schedule a directory follows to re-check listed facts: pricing pages checked monthly, feature and engine claims checked quarterly, in this directory’s case. It exists because vendor pricing and feature sets change faster than most static content gets updated. See the methodology page for the full schedule.
W
Workflow Automation
Workflow automation, in an AI SEO platform, is a feature that turns a diagnosed gap directly into a drafted fix: a rewritten paragraph, a schema block, or a content brief, without a separate tool. Temso includes an AI SEO agent that can execute some recommendations directly inside the platform, alongside its tracking and content modules, though most teams still review a change before it goes live.
Z
Zero-Click Search
A zero-click search is a query answered fully inside the search or chat interface, so the user never clicks through to a website. Google AI Overviews and direct chat answers both drive zero-click behavior, which is why citation rate and brand mention volume, not click-through traffic, have become the primary success metrics for AI SEO work.
Put these definitions to work
Reading a definition is the easy part. The next step is running your brand against a real prompt set and checking which of these terms actually describe a gap you have: thin engine coverage, a low citation rate on your pricing page, or a hallucination nobody has corrected yet.
Compare how the tools in this directory measure each concept above on the AI SEO tools index, where every entry carries a last-verified date and a full index-score breakdown.