Home › Docs › Glossary Glossary Common terms used across the PeakGEO docs and console.
On this page This page collects the terms used across the PeakGEO docs and console. Each term has a short definition and a link to the page that covers it in detail.
Basics
Term Definition Learn more GEO Generative Engine Optimization: the work of getting a brand accurately mentioned, understood and recommended in AI-generated answers. Product overview Brand diagnosis A complete measurement for one brand: asking each question on the selected models, computing metrics, and producing key findings and actions. Quickstart Initial diagnosis (baseline) The first diagnosis after a brand is created; the starting point for later retests and comparisons. Quickstart AI platform / model An AI model that receives the questions, such as Doubao, DeepSeek, Kimi, ERNIE or Zhipu Qingyan; selectable platforms depend on the console catalog. Product overview Valid answer An answer that came back and can be counted. Timeouts, failed samples and empty answers are excluded. The unit of measurement is one answer from one model to one question. Metric definitions Brand profile The brand name, industry, category, products and positioning you confirmed, with their sources, plus any uploaded reference materials. Quickstart Target market The market a diagnosis is aimed at; it sets the language and region of questions and competitors. Question sets and measurement types Buyer persona The user point of view questions are written from, such as first-time or price-conscious buyers. Not a verified customer profile. Question sets and measurement types Competitors Competitors you have confirmed for the formal comparison. Unconfirmed candidates are not counted. Cited sources and share of voice
Question set
Metrics and sources
Term Definition Learn more Organic mention rate The share of valid organic answers that mention your brand. Metric definitions Organic recommendation rate Answers to organic recommendation questions that recommend your brand, divided by all valid organic answers. Metric definitions Risk mention rate The share of valid organic answers that express caution, negativity or risk about your brand, counted once per answer. Metric definitions Confidence interval The 95% interval attached to each rate, showing its likely range. With fewer than 20 valid answers the interval is wide, so treat the rate as indicative. Comparing runs and confidence intervals No valid answers Shown when the denominator is zero. It means "not measured", which is different from a real 0%. Metric definitions Share of voice Your brand's share of all organic mentions of your brand and tracked competitors. Not the same as market share. Cited sources and share of voice Answer citation rate The share of valid answers that cite at least one source. Cited sources and share of voice Official-site citation rate The share of valid answers citing your official website or its subdomains. Cited sources and share of voice Domain citation rate The share of valid answers citing a given domain, counted once per answer. Cited sources and share of voice Key findings The opportunities, gaps and risks on the Overview; each one traces back to its questions and original answers. Product overview Evidence gap Content not covered by the materials collected this round that may affect AI answers; shown on the Evidence page. Action recommendations
Actions, retests and monitoring
Term Definition Learn more Actions To-dos derived from the diagnosis, in two kinds: verification and improvement. Action recommendations Verification action An action that first establishes the real cause; its finding is Issue found, No issue found or Insufficient material. Action recommendations Improvement action An improvement carried out once the evidence is clear; its effect is observed through a retest. Action recommendations Topic (content) A content task created from an action, which moves through writing, fact check and logged publication. Content loop Fact check Checking every statement in a draft against its source and for risks. Content loop Retest Measuring again with the same questions and models and comparing against a baseline. Monitoring, retests and alerts Baseline The measurement a retest is compared against. Monitoring, retests and alerts Monitor now A one-off run that only samples and records answers, without a before/after comparison. Monitoring, retests and alerts Indistinguishable The before and after confidence intervals overlap, so the change is within sampling error and no rise or fall is reported. Comparing runs and confidence intervals Report A report you can read online or download; it is a snapshot of that moment. Report export GEO assistant The assistant at the bottom right of the console, for interpreting results and moving work forward. GEO assistant
Didn't find an answer? Email us: [email protected]