Dashboard

What goes in a software delivery dashboard in Metabase?

A software delivery dashboard gives engineering and product leaders one view of flow, throughput, and quality. It's the exec roll-up that sits above per-team boards. Build it from issue-tracker data synced into a database — see Linear or Jira for the connection.

For: Eng leaders, EMs, PMs. Refresh: hourly is plenty. Source: modeled issues, cycles/sprints, projects (+ status history for flow metrics).

Which cards belong on a software delivery dashboard?

Executive KPIs (top row)

  • Open issues
  • Completed last 7 days / created last 7 days
  • Median cycle time (requires status history)
  • High-priority open bugs

Throughput

  • Completed issues by week
  • Created vs. completed by week
  • Completed by team

Flow & quality

  • Median cycle time by week
  • Work-in-progress by status
  • Bugs created vs. resolved by week
  • Oldest open issues (table)

What data does a software delivery dashboard need?

  • A modeled issues table with created_at, started_at, completed_at, state_type, team, priority.
  • For cycle time and time-in-status, a status-change history table. Without it, drop those cards or add a caveat.

How do you build a software delivery dashboard?

  1. Sync your tracker into a database (Linear / Jira).
  2. Model issues + iterations + projects; define state_type and derived cycle_time.
  3. Create one saved question per card (SQL or query builder).
  4. Add dashboard filters: team, project, iteration, priority, date range.

Example card SQL

Created vs. completed by weekPostgreSQL
SELECT
  weeks.week,
  COUNT(c.id) AS created,
  COUNT(d.id) AS completed
FROM (
  SELECT generate_series(date_trunc('week', CURRENT_DATE - INTERVAL '12 weeks'),
                         date_trunc('week', CURRENT_DATE), INTERVAL '1 week') AS week
) weeks
LEFT JOIN issues c ON date_trunc('week', c.created_at)   = weeks.week
LEFT JOIN issues d ON date_trunc('week', d.completed_at) = weeks.week
                  AND d.state_type = 'completed'
GROUP BY weeks.week ORDER BY weeks.week;

Integrations

Dashboards

Metrics

FAQ

Can I build this without status history?
Yes for throughput, open counts, and bug load. Cycle time and time-in-status need history.
Which tools can feed it?
Any tracker modeled onto the shared schema — Linear, Jira, and others.