DORA & SPACE — Measuring Engineering Health

DORA measures delivery performance. SPACE measures developer productivity. Neither tells the full story alone — and neither should be used to evaluate individuals.

When to use

• Quarterly engineering health reviews
• Making the case for tooling or process investment
• NOT for individual performance evaluation

Tradeoffs

• DORA metrics can be gamed (more small deploys artificially improves frequency)
• Activity metrics (PRs merged, commits) are vanity metrics — avoid using them alone

---

DORA 4 Metrics

Metric	Measures	Elite threshold
Deployment Frequency	How often you deploy to prod	Multiple times/day
Lead Time for Changes	First commit → prod deploy	< 1 hour
Change Failure Rate	% of deploys causing incidents	< 5%
Time to Restore	MTTR when an incident occurs	< 1 hour

SPACE Dimensions

Dimension	What it captures
Satisfaction	Developer wellbeing, burnout risk
Performance	Outcomes delivered, quality
Activity	Commits, PRs, reviews (context-dependent)
Communication/Collaboration	Knowledge sharing, unblocking
Efficiency	Flow state, interruptions, toil

---

Go

// DORA metrics — SQL/Prometheus patterns

// Deployment Frequency: deploys per day
// SELECT DATE(deployed_at), COUNT(*) AS deploys
// FROM deployments
// WHERE env = 'prod'
// GROUP BY DATE(deployed_at)

// Lead Time: avg minutes from first commit to deploy
// SELECT AVG(TIMESTAMPDIFF(MINUTE, first_commit_at, deployed_at))
// FROM deployments WHERE env = 'prod'

// Change Failure Rate: failed deploys / total deploys
// SELECT
//   SUM(CASE WHEN caused_incident THEN 1 ELSE 0 END) * 100.0
//   / COUNT(*) AS failure_rate_pct
// FROM deployments WHERE env = 'prod'

// Time to Restore: avg minutes incident open
// SELECT AVG(TIMESTAMPDIFF(MINUTE, opened_at, resolved_at))
// FROM incidents

Python

# DORA metrics — SQL/Prometheus patterns

# Deployment Frequency: deploys per day
# SELECT DATE(deployed_at), COUNT(*) AS deploys
# FROM deployments
# WHERE env = 'prod'
# GROUP BY DATE(deployed_at)

# Lead Time: avg minutes from first commit to deploy
# SELECT AVG(TIMESTAMPDIFF(MINUTE, first_commit_at, deployed_at))
# FROM deployments WHERE env = 'prod'

# Change Failure Rate
# SELECT
#   SUM(CASE WHEN caused_incident THEN 1 ELSE 0 END) * 100.0
#   / COUNT(*) AS failure_rate_pct
# FROM deployments WHERE env = 'prod'

# Time to Restore
# SELECT AVG(TIMESTAMPDIFF(MINUTE, opened_at, resolved_at))
# FROM incidents

Gotcha: Never use DORA metrics to evaluate individual engineers. They measure team and system health. An engineer in a slow org with manual deploys will look worse than an engineer in a fast org with CD — even if they're better.