The Incident Dashboard Built for Live Response

Dashboards for daily ops are wrong for incidents. The incident-specific dashboard.

Layout

The layout is incident-first. Big state at the top, supporting context below; everything readable in seconds.

Top: current state. Per-incident affected-service status panel; big text, red or green.
Middle: time-series. Per-incident key-metric chart for the last hour; drives "is it getting better?" answer.
Bottom: recent context. Per-incident recent deploys and recent alerts; drives "what changed?" investigation.
Per-dashboard at-a-glance test. Per-dashboard "can I parse this in 5 seconds?" check; catches dashboard sprawl.

Focus

Focus matters more than completeness. Hide non-relevant metrics; let drill-downs handle depth.

Only relevant metrics. Per-incident curated panel set; hide everything else.
Expandable drill-downs. Per-panel in-place expand; detail without leaving the page.
Per-dashboard no-noise rule. Per-dashboard explicit panel-pruning; catches "show me everything" anti-patterns.
Per-dashboard per-incident-class variant. Per-class dedicated dashboard; supports targeted response.

Sharing is the war-room amplifier. URL-shareable view; permanent reference for postmortem.

URL-shareable. Per-dashboard deep-linkable URL; war-room participants see the same view.
Permanent for postmortem reference. Per-dashboard long-lived URL; postmortem links here for the actual data.
Per-dashboard time-window snapshot. Per-incident captured time range; supports later reconstruction.
Per-dashboard access control. Per-dashboard appropriate-audience scope; supports compliance posture.