the product
Analytics
Two kinds of numbers live here, and the dashboard never mixes them up: decisions, counted from consent records and exact, and visitor numbers, reported by the banner itself and therefore honest but not flattering.
sourcesWhere the numbers come from
Decisions are counted server-side from consent records, never from pageviews, and roll up daily (with today live). Visitors and banner impressions come from a tiny beacon the banner sends once per pageview; ad blockers remove that beacon, so those numbers understate reality, and they are only counted on production. Whenever a metric depends on the beacon, the page says so.
definitionsEvery metric, defined
| Metric | Definition |
|---|---|
| Visitors | Pageviews reported by the banner's beacon. Best-effort, production only. |
| Banner shown | Pageviews on which the first screen actually rendered. |
| Decisions | Explicit choices: accepted everything, accepted some, rejected everything. |
| Accept rate | Accepted everything, divided by all explicit decisions. |
| Dismissed | Closed the banner where closing counts as refusing (recorded as such, and only when that behaviour is on). |
| Withdrawn | Came back and withdrew consent entirely. |
| Updated | Changed an earlier choice. Not counted as a new decision. |
| GPC opt-outs | The browser's Global Privacy Control said no on the visitor's behalf, under a rule set that honours it. |
| Expired | Stored consents that ran out and will be asked again. |
reading itThe overview
The overview leads with the accept rate, then decomposes every banner impression: accepted everything, accepted some, rejected, and, where the beacon numbers allow it, the visitors who left without choosing. That last group is the one most consent tools cannot see, and it is often where the real story is. When the roll-up counts more decisions than impressions (blocked beacons make that possible), the bar covers decisions only and the page says exactly why.
Ranges are 7, 14, 30 or 90 days, or your own dates, with the environment switchable. Each headline number carries a comparison against the previous equal-length window, and the comparison is only shown when the server really answered for that exact window; a clamped or empty window produces an honest sentence, never a made-up percentage.
deeperBreakdowns and version compare
Breakdowns split decisions by region, country, language, banner layout, rule set or device, each row with its own accept rate; the overview shows the top countries with the full table one click away. Compare puts two published versions side by side, with the config differences between them, so a wording change can be tied to what it did to consent. The same per-version rates appear on the banner's Versions ledger.
On the Free plan, analytics is clamped to the last seven days and the page says so. Impressions cannot be split by country; that split exists only for decisions.