Analytics & A/B testing: consent rates, banner performance, experiments

What it's for

Dashboard → Analytics turns raw consent events into the metrics you actually act on: how many visitors opt in, which categories they grant, where they're coming from, and — with A/B testing enabled — whether a banner change actually moved the needle before you roll it out to everyone.

Step-by-step: reading the page

1. Pick a property. If your org has more than one site registered, use the site switcher; analytics are always scoped to one property (cbid) at a time, so the numbers stay meaningful — you never see two sites' traffic blended together.

2. Pick a range. The segmented control in the header (7d / 30d / 90d / All) controls the window every chart and KPI on the page reflects.

3. Read the KPI row at the top:

   KPI	What it means
   Consents	Total consent decisions in the selected range
   Opt-in rate	Share of decisions that granted at least one non-necessary category
   Impressions	How many times the banner was shown
   Ratio	Consents ÷ impressions — how many banner views actually convert into a decision

   Each KPI shows a delta arrow against the equivalent prior period (e.g. the previous 30 days) so you can catch a regression — like a copy change that tanked opt-in rate — quickly, without eyeballing raw numbers.

4. Scan the charts, in order:

   • Opt-in trend — daily opt-in rate over time (area chart).

   • Consent volume — daily opt-in vs. opt-out counts, stacked.

   • Implied vs. strict — a donut of decisions made passively (implied) vs. via explicit interaction. If a site has no implied/strict data yet, this card shows an empty state instead of a misleading empty chart.

   • Per-category opt-in — three trend lines (Preferences, Statistics, Marketing), so you can see, for example, that visitors are happy to accept Statistics but consistently reject Marketing.

   • Top countries — a bar chart plus a matching table of consent volume by country.

   • Funnel — impressions → opt-ins → opt-outs, with a percentage bar for each stage relative to impressions, showing drop-off at a glance.

5. Drill into a country. Click a country in either the bar chart or the table to filter every other chart on the page down to that country — the trend, volume, category, and implied/strict cards all recompute against the filtered subset. A chip appears in the page header showing the active filter; click it (or the country again) to clear it. Combine this with the date-range selector — filtering to "DE" and switching to "7d" narrows everything at once.

What the data means

The underlying numbers come from GET /api/v1/:cbid/consent/stats, which returns one row per day with opt-in/opt-out counts, per-category opt-in counts, implied-vs-strict counts, impressions, and a country breakdown. The dashboard does the rest client-side: computing rates, deltas against the prior period, and country-filtered scaling (when you drill into a country, each day's volume is scaled by that country's share of that day's total events, since the API reports country totals per day rather than per individual record).

An important distinction: this page shows aggregated, anonymized trends — it answers "how is my banner performing," not "what did person X consent to." For the latter, you want the consent log, looked up by stamp.

Multi-property orgs

If you manage several sites, remember analytics is always single-site. To compare two properties, switch between them rather than expecting a combined view — this is intentional, since blending traffic from unrelated domains would make the opt-in rate and country mix meaningless.

A/B testing

Under Dashboard → Studio → Settings → Targeting, you'll find an A/B testing toggle (this is where "Banner settings" lives — the old /dashboard/banner route redirects to Studio). Enabling it splits visitors between your current banner (variant A) and a challenger (variant B) using an adjustable split slider (0–100%, in 5% steps, defaulting to a 50/50 split).

Once enabled, impressions and opt-ins are recorded per variant and feed the same consent-stats pipeline analytics reads from, via:

GET /api/v1/:cbid/ab

which returns each variant's impressions, opt-ins, and opt-in rate — letting you judge objectively whether a design or copy change actually improves opt-in rate before you commit to it for all visitors.

Note: A/B testing is a paid ("Scale" plan) capability. If your org isn't entitled, the toggle in Studio shows a plan badge instead of turning on.

Export and compliance value

Analytics isn't itself an audit artifact — it's aggregated and doesn't carry individually-verifiable proof the way a signed receipt does. Its value is operational: showing you, and anyone reviewing your consent program internally, that your banner is actually functioning as intended (visitors are seeing it, categories are being genuinely offered, and opt-in behavior looks plausible for your audience and region mix). If a regulator or auditor needs record-level proof, point them at the consent log or a signed receipt instead — analytics is for you, receipts are for them.

See also: Consent log · Audit export & receipts