Production RUM Dashboard

Use this when a performance claim matters beyond local benchmarks.

Rule

Design the dashboard even if production telemetry is not available yet. Mark it as a proof gap instead of pretending lab benchmarks are enough.

Required Tags

• interaction name
• cohort
• document size
• visible DOM count
• hidden boundary count
• decoration/comment/annotation count
• custom renderer flags
• mode: off, auto, DOM-present, shell, staged
• mobile/browser/IME
• release/version

Metrics

• p50/p75/p95/p99 interaction latency
• JS heap
• DOM node count
• mounted group count
• listener count
• cached index sizes
• React scheduler priority
• component render/mount counts where available

Dashboard

Datadog or equivalent should answer:

• which cohort regressed
• which interaction regressed
• which mode regressed
• which memory bucket grew
• which release introduced it