# Core concepts

This page explains how Voxoria is organised and defines the terms you'll see throughout the product and these docs. For the full definition of each metric, see the [Metrics reference](/docs/metrics).

## The hierarchy

Voxoria nests from billing down to individual AI answers:

- **Account** — your billing root.
- **Organization** — a workspace inside the account, where your team and projects live.
- **Project** — paired one-to-one with your own brand. A project holds everything Voxoria tracks for you.
- Inside each project: your **brands** (you plus competitors), **prompts** (the questions tracked daily), **engines** (the AI surfaces queried), and the **AI answers** captured from each run.

Everything you measure rolls up from the answers at the bottom of this tree.

## Definitions

**Project.** The container for one brand you want to track. It's paired one-to-one with your own brand and holds its competitors, prompts, engines, and captured answers. To set one up, see the [Quickstart](/docs/quickstart).

**Own brand vs competitors.** Each project has exactly one **own brand** — yours — plus the **competitors** you benchmark against. Competitive metrics like share of voice compare your own brand to this set.

**Prompts.** The buyer questions Voxoria tracks for you, like "best CRM for agencies". Each prompt runs daily against your engines. A prompt belongs to a topic, can be **generated** by Voxoria or **manual** (authored by you), and has a status: **active** (tracked daily), **inactive** (paused — history kept, no new runs), or **suggested** (a candidate waiting for you to accept).

**Topics.** Themes that group related prompts — for example pricing, comparisons, or use cases. Topics let you read your results by buyer need rather than one question at a time.

**Engines (channels).** The AI answer surfaces Voxoria queries on your behalf — the major AI answer engines, including ChatGPT, Perplexity, Gemini and Google AI Overviews. You track up to three per project, and which engines are available depends on your plan.

**AI answers (chats).** One captured answer per prompt, per engine, per day. Every answer is broken down into mentions, sources, and sentiment — these captured answers are the raw material for all your analytics.

**Mentions.** Each time a brand is named in an answer. Mentions power visibility, share of voice, and where your brand sits in the order an answer lists brands.

**Sources.** The URLs an answer drew on. A source can be **retrieved** (the model fetched the page) and/or **cited** (shown as a visible footnote in the answer). These two are **independent** — a page can influence an answer without being cited, and a citation implies the page was used. Source metrics show which pages AI engines treat as evidence.

**Sentiment.** Each claim an answer makes about a brand is scored as **endorsed**, **neutral**, or **undermined**. These roll up into **net sentiment**, a score from −1 to +1 that tracks whether a brand's narrative is helping or hurting it.

**Audits.** An on-demand, AI-driven analysis of a single tracked prompt. An audit produces a verdict, a scored rubric across dimensions, a prioritised action plan ranked by effort and impact, and a phased roadmap. It's the flagship "take action" feature, and each audit costs **100 credits**.

**Credits.** Prepaid units you spend on actions. Today only audits consume credits — at 100 per audit. Your plan includes a monthly credit allowance.