Analytics

Open Analytics to inspect reviews created in the last 30, 90, or 365 days. The figures come from authorized workspace review records, not PostHog event counts. The full period is aggregated independently of the Reviews list limit.

Metric	Meaning
Activity	Reviews created per UTC day, including zero days.
Completion	Published reviews divided by all reviews created in the period. Cancelled and superseded reviews remain in the denominator.
Findings	Reported or accepted aberrations; observations and dismissed findings are excluded. The same finding in two runs counts in each.
Browser coverage	Targets whose execution status passed, grouped by browser. This does not mean there were no UI findings.
Repositories	Distinct repositories with reviews in the period.
Creators	Explicit requesters, with automatic and unattributed requests separate.
Equivalent manual QA	Completed targets multiplied by ten minutes, expressed in hours.

The manual-QA figure is an uncalibrated estimate, not measured human effort or guaranteed time saved. It excludes unfinished targets, setup, and report writing. Select activity dates and review-count links to inspect matching runs. Reviews can filter by repository, status, UTC dates, completed duration, requester, and PR author. Repository-name links open GitHub. Repeated intentional requests count separately; duplicate delivery of one request does not. Outcomes reflect current state, so a historical period can change as its in-progress runs finish. GitHub and WorkOS requester identities are not merged without a verified link; historical missing identities remain unattributed. In contributor development, analytics can use real local recordings and collected previews, explicitly labeled as local data. Hosted storage failures remain errors; production does not substitute local files or sample analytics.

Get started

Configure reviews

Understand results

Workspace

Reference