How-to
Read the AI monitor overview
Filter the overview, read the KPI tiles and cards, and interpret visibility and share of voice.
The Overview tab of the AI Monitor summarises every answer collected in a date range.
Set the filters
Section titled “Set the filters”- Open AI Monitor → Overview.
- Choose a date range. The default is the last month; you can also pick a custom range.
- Narrow the set with Tags and Models. The Models dropdown appears only when answers exist from more than one AI provider.
All cards below follow these filters.
Read the KPI tiles
Section titled “Read the KPI tiles”- Total answers — how many answers were collected, with the prompt count underneath.
- Models — how many distinct AI models produced those answers.
- Visibility — the share of answers that mention any of your own brands.
- Average sentiment — the average sentiment of those answers, from −1 to +1, with a label.
Read the cards
Section titled “Read the cards”- Rankings — each brand with its visibility percentage and trend badge, an average sentiment dot, and an average position. Your own brands carry an Own badge. Here visibility is each brand’s share of all brand mentions, so the column sums to 100% for the period. The trend compares the range with the previous range of the same length.
- Share of Voice — your brand’s slice of the total brand-mention pool. Each answer counts at most once per brand.
- Top domains — the websites the models cite most often, with the number of answers referencing each one and its share of all answers in the range.
- Visibility trend — one line per brand per day, showing that brand’s share of the mentions that day.
Interpret the numbers
Section titled “Interpret the numbers”Read visibility and share of voice together. High visibility with a low share of voice means the models mention you, but mention competitors more often. A rising average position (a larger number) means the models name you later in their answers.
Empty and warning states
Section titled “Empty and warning states”- If your project has no prompts, a banner tells you so — no answers can be collected until prompts exist.
- If share of voice cannot be calculated, no brand mentions were found in the filtered answers. Widen the date range or clear a filter.