Core concepts

Brand

The company you are measuring. A brand carries a name, a set of name variations, a website, an industry and a default country for new groups. Every metric is scoped to exactly one brand at a time.

Accounts can hold several brands — for agencies, or for a company with distinct product lines. Switching brands switches the entire workspace: prompts, competitors, metrics and audits. See Brands.

Group

A named set of prompts, usually one theme or product line — "CRM comparisons", "pricing questions". Groups are how the dashboard is sliced: metrics are calculated per group, so you can see that you own one category and are absent from another.

A group also carries the country its prompts are asked in, and it is the unit of scheduling: a group is what gets refreshed on a cycle. Those two facts make the group, not the brand, the thing you create per market — see Geographies & languages.

Prompt

One question, as a customer would ask it. Prompts belong to groups and can be moved between them.

A prompt is only a measuring instrument — it is not something you optimise. You change what assistants say by changing what they read, not by rewording the prompt.

Answer

What one assistant replied to one prompt, at one point in time, in one country. A single prompt produces many answers over a period: one per assistant, per run.

Answers are the denominator of nearly every metric. A prompt with no answer in a period is excluded from calculation rather than counted as an absence — see How metrics work.

Mention

An occurrence of your brand name — or one of its configured variations — in the text of an answer. Mentions drive Visibility (counted once per answer) and Share of Voice (counted per occurrence).

Citation

A source link an assistant used when building an answer. Citations aggregate into the Sources leaderboard, which is the most actionable view in the product: it names the pages that decide what assistants say about your category. See Citations & Sources.

Report

A snapshot of a group at a point in time: the prompts, the answers they produced, the mentions and citations found in them, and the metrics derived from all of it. The dashboard aggregates reports across the period you select; individual runs are opened from the group itself in the Prompts tab.

"Report" is also the name of a tab, but that tab holds the site audits (Schema, Technical, Robots.txt) rather than these snapshots. Those audit your own website, not AI answers — the two live in one product because both describe how AI systems see you.

How they fit together

Brand
└── Group ("CRM comparisons")
    └── Prompt ("best CRM for a 10-person agency")
        └── Answer (ChatGPT, 12 May, Ukraine)
            ├── Mentions  → Visibility, Share of Voice, Position, Sentiment
            └── Citations → Sources leaderboard

Every number in the product is an aggregation of that tree over a period, a set of assistants and a set of brands.