The vocabulary of AI search.
We sample 50,000 buyer questions a week through ChatGPT, Claude, Gemini, and Perplexity to see which brands they recommend. This is the glossary we use, plain-language definitions of every term that comes up in the work, plus how to measure each one and where it shows up in our reports.
A
Answer Engine Optimization (AEO)
Answer Engine Optimization (AEO)
AEO is the practice of optimizing your site so AI assistants like ChatGPT, Claude, and Gemini name your product when buyers ask for a recommendation.
Read the definitionAll entries
Sorted by · Conceptual
- GDefinition
Generative Engine Optimization (GEO)
GEO is the SEO industry term for the same problem AEO describes: getting LLMs to mention you when they generate answers in your category.
Definition·4 min·2026-05-09 - VDefinition
AI Search Visibility
AI search visibility is whether and how often LLM-powered tools name your brand when they answer questions in your category.
Definition·4 min·2026-05-09 - %Metric
LLM Mention Rate
LLM mention rate is the percentage of model answers to a given buyer question that name your brand.
Metric·5 min·2026-05-09 - ≈Methodology
Prompt Sensitivity
Prompt sensitivity is how much an AI answer changes when a buyer phrases the same question slightly differently.
Methodology·5 min·2026-05-09 - σMethodology
Stochastic AI Search
Stochastic AI search means the same question can produce different recommendations each time, even on the same model.
Methodology·5 min·2026-05-09 - ↔Definition
AEO vs SEO
SEO ranks your URL on a list of links. AEO gets your brand into the one-sentence answer the AI gives instead of the list.
Definition·4 min·2026-05-09 - GMetric
ChatGPT Brand Mentions
ChatGPT brand mentions are the share of ChatGPT answers that name your brand for a given buyer question.
Metric·5 min·2026-05-09 - PMetric
Perplexity Citations
Perplexity Citations are the source URLs Perplexity links beneath each answer. Earning them is the most direct way to convert AI search demand into traffic.
Metric·6 min·2026-05-10 - OMetric
Google AI Overviews
Google AI Overviews are the AI-generated summary that appears above traditional results for around half of US searches. Getting cited there is the new top-of-page.
Metric·6 min·2026-05-10 - RMethodology
Retrieval Augmented Generation (RAG)
RAG is the architecture behind every modern AI search engine. Understanding it changes what you optimize on your own pages.
Methodology·7 min·2026-05-10 - KMethodology
Content Chunking for AI Retrieval
Chunking is how AI search engines split your page into passages before retrieval. Bad chunks are why pages with the right answer still don't get cited.
Methodology·6 min·2026-05-10 - VMethodology
Vector Embeddings for AI Search
Vector embeddings are how AI engines decide which of your passages match a buyer's question. Cosine similarity above 0.88 is the practical threshold to clear.
Methodology·6 min·2026-05-10 - EMetric
Entity Density
Entity density is the number of recognized named entities per thousand words on a page. Crossing 15 entities per thousand words is the practical AI Overview threshold.
Metric·6 min·2026-05-10 - DMethodology
Definition-First Content
Lead every section with a one-sentence direct answer to the implied question. It is the single highest-leverage chunk pattern in AI search.
Methodology·5 min·2026-05-10 - SMethodology
Schema Markup for AEO
Schema.org markup (FAQ, HowTo, Article) lifts AI Overview citation odds by 3x. The fastest technical SEO win still on the table.
Methodology·6 min·2026-05-10 - BMethodology
AI Crawler Allowlist (GPTBot, ClaudeBot, PerplexityBot)
If your robots.txt blocks GPTBot, ClaudeBot, or PerplexityBot, you are invisible to every AI search engine. The first thing to check.
Methodology·6 min·2026-05-10 - CMethodology
Entity Consistency Across the Web
Your homepage, G2 profile, LinkedIn, and Crunchbase must describe your company in matching language. Inconsistency is the most common cause of weak AI mention rate.
Methodology·5 min·2026-05-10 - SMetric
Share of Voice in AI Search
Share of voice in AI is the percentage of category answers that name your brand. It is the cleanest single rollup of AEO performance.
Metric·6 min·2026-05-10 - MMetric
Cross-Model Agreement
Model agreement is how aligned ChatGPT, Claude, and Gemini are on a buyer question. Low agreement means the category is contested, which is exactly when AEO matters most.
Metric·6 min·2026-05-10 - QMethodology
Query Universe
A query universe is the curated set of 50 to 150 buyer questions that define what your AEO program actually measures. Without it, every other metric is anecdotal.
Methodology·6 min·2026-05-10 - ↓Methodology
The 3-Month Citation Cliff
Pages that have not been updated in 90 days are increasingly rotated out of AI Overview citations. A citation cliff that surprised most teams in 2025.
Methodology·5 min·2026-05-10 - ±Methodology
Confidence Intervals for AI Mention Rate
A single AI scan is not a measurement. Reporting mention rate without a confidence band is either dishonest or naive.
Methodology·6 min·2026-05-10 - →Methodology
Simulate Before Publish
Before you ship a page edit, predict what mention rate change it should produce. The forecasting step that closes the AEO feedback loop.
Methodology·6 min·2026-05-10 - ✓Methodology
Verify After Deploy (14-Day Attribution)
Two weeks after shipping a page edit, re-scan and compare actual mention-rate change to the predicted lift. Closes the AEO feedback loop.
Methodology·6 min·2026-05-10 - XMethodology
Surrogate Model (XGBoost) for AEO
A surrogate model is the lightweight predictor that maps page features to mention rate. Trained daily, it makes simulate-before-publish feasible.
Methodology·6 min·2026-05-10 - ΦMethodology
SHAP Values for AEO Explainability
SHAP values explain why the surrogate predicts what it predicts. Without them, AEO predictions are oracle outputs nobody trusts.
Methodology·6 min·2026-05-10 - 1Metric
Position Bias in LLM Answers
Where in the answer your brand appears matters as much as whether it appears. Position bias is the second most-actionable AEO metric.
Metric·6 min·2026-05-10 - ↻Methodology
Recency Bias in LLM Citations
AI engines preferentially cite newer pages, often at the expense of older pages with stronger topical authority. The freshness curve every AEO program now plans against.
Methodology·5 min·2026-05-10 - %Metric
Answer Share
Answer share is industry shorthand for share-of-voice in AI search. Used interchangeably with brand mention rate across the AEO platforms.
Metric·5 min·2026-05-10
See what AI says about you, today.
Send your domain. We run 50 buyer questions in your category through ChatGPT, Claude, Gemini, and Perplexity, and email back the answer set, your mention rate, and the page edit that moves the needle. 24 hours, free.
4 models · 50 questions · 24-hour turnaround · no credit card