Dashboard & metrics
The dashboard is the running picture of how often AI assistants mention your brand, how they talk about it, and who they name instead. Every number is computed from your prompt runs — nothing is modelled or estimated.
How it works
- 1
Visibility Score
Each tracked prompt is sent to each AI platform, and every prompt × platform pair is one cell. The score is the percentage of answered cells that mention your brand. Cells that returned no response are excluded, so a provider outage shrinks your sample rather than your score.
- 2
Mention Rate and Share of Voice
Mention rate is how many of your prompts produced at least one mention. Share of voice compares your mentions against the competitors detected in the same answers — it tells you whether you're losing ground or the whole category is quiet.
- 3
Sentiment
How your brand is characterized when it does appear: positive, neutral, or negative. A high visibility score with weak sentiment is a content problem, not a coverage problem, and needs a different fix.
- 4
Citations
Which sources the models cite when answering your prompts. These are extracted from the responses themselves, so an empty state means no run has completed yet rather than that nobody cites you.
- 5
Action items and page audit score
The page audit grades the technical side — crawler access, structured data, answer-ready content — and turns failures into action items ordered by severity.
Common problems
The failure modes we actually see, what causes them, and the fix.
The score moved but I changed nothing
- Cause
- AI answers are non-deterministic: the same prompt can name different brands on different days. Provider-side model updates cause step changes too.
- Fix
- Read the trend across several runs, not one. Automatic runs fire Monday, Wednesday and Friday at 07:00 UTC on the days you enabled for the brand, which builds a usable series within about two weeks.
“Your data is getting stale”
- Cause
- No prompt run has completed for more than 14 days, so the dashboard is showing an old snapshot.
- Fix
- Run the prompts manually from the Prompts page, or switch on automatic refreshes in brand settings.
Automatic refreshes are paused
- Cause
- The brand tracks more prompts than its plan allows, so scheduled runs are held back rather than producing a partial, misleading series.
- Fix
- Archive prompts until you're inside the limit, or upgrade. The dashboard banner shows your current count and the allowed maximum.
The manual run button is disabled
- Cause
- Automatic refreshes are on for this brand. Manual runs are blocked so a hand-triggered run can't distort the scheduled series.
- Fix
- Turn automatic refreshes off in brand settings if you need to test by hand, then switch them back on.
AI Crawlers shows no data
- Cause
- Crawler traffic is read from Cloudflare, and no Cloudflare zone is connected to the brand.
- Fix
- Connect Cloudflare under Integrations. Snapshots are collected daily at 02:00 UTC, so the first numbers appear the following day.