Skip to main content
This page walks through the four screens you will use most. Every number on them can be traced back to the AI answers behind it, so you can check a figure against what the engines actually said.

​
Overview

The Overview is the summary for one market. It shows:
  • Visibility: the share of recorded answers that name your brand over the selected window, next to your five most-mentioned competitors. The headline rounds to a whole percent; the Visibility page shows one decimal.
  • Trend chart: one point per day. It appears after a few days of answers; until then the panel says so.
  • Rankings: the brands the AI engines name most for your questions.
  • Top cited domains: the share of AI answers citing each domain.
  • What moved: the notable changes since last week, or “all quiet” when nothing notable moved.
  • Top opportunities: the biggest topic gaps to close.
  • Biggest opening: the question type (comparison, best-of list, how-to…) where the AI names you least, for the window and filters you selected.
  • Recent AI responses: every answer your questions produced, with the engine that gave it.
Use the filters at the top to pick a window (a preset or a date range you draw), an engine and a theme. The headline score follows the range you draw. The week-over-week change is hidden while a filter is on, because it always covers all engines and themes and would not match your filtered score. Run scan asks every active question again. Only Admins can use it, and one scan runs per workspace at a time.

​
Visibility

The Visibility page breaks the same measurement down.
  • Visibility over time: the share of answers naming your brand, per day.
  • Segments breakdown: per topic, per persona and per platform, who leads and where you stand.
  • Reach by platform: the share of answers naming you on each engine. A full bar is the most-named brand on that engine.
  • Prompt-level evidence: one line per question, with the engines that answered it, so you can check any figure against the answers behind it.
Questions that contain your own brand name are left out of the competitor comparison, because an answer to them always names you. You still see them in the per-question breakdown.

​
Competitors

  • Leaderboard: every tracked brand ranked by the share of AI answers naming it, on one scale, with your distance to the leader.
  • Where you lose: the five topics where the gap to the leader is widest right now. A topic shows up once both your share and a rival’s have been measured on it.
  • Suggested: brands that appear in the answers but that you do not track yet. Nothing is added until you confirm.
  • Per competitor: head-to-head, a breakdown by engine and by topic (biggest losses first), and the share of answers where the engines recommended the brand rather than just mentioning it.
Competitors are tracked per market. When you add a market, you can copy the competitors of another one, but review the list: competitors differ by country. Each competitor has two lists you can edit:
  • Other spellings: every way the AIs write the brand, such as an abbreviation or a spelling without the accent. Each one counts as a mention.
  • Never count as this brand: phrases that contain the name but mean something else. Useful when a brand name is also an ordinary word.
The spellings of your own brand are under Brand names in Settings → Domains & markets.

​
Reports

Reports builds one PDF per market for the period you choose, and can email a weekly summary.
  • The weekly email goes out every Monday to the people who turned it on in Settings → Notifications. It attaches a limited number of reports; any other markets are named in the email but not attached.
  • Reports and the weekly email are in English, whatever language the dashboard is set to.
  • A market with nothing collected in the period is listed as “nothing collected” instead of producing an empty report.

​
How to read the numbers

“We asked and you were not named” and “we have not asked yet” are different findings. A rate only appears once enough answers sit behind it; until then you see a dash, with the number of answers shown next to it. A measured rate that would round to 0.0% is shown as <0.1%.
Each share of answers naming you is printed with the number of answers behind it. Two answers are not enough to say “50%”, so a question with too few answers shows a dash.
When an engine cannot be reached for a question, the attempt is recorded with a reason and left out of every rate: it counts neither as an answer nor as a miss. You can see these failures per engine in Settings → Scan history.
A brand tracked in three countries is three markets, each measured on its own answers. The workspace switcher at the top of the sidebar selects the market every screen shows.