KPIs for software development: which ones to track and how to use them
Software development KPIs are a small set of team-level measures tied to business goals. Most teams start with DORA's five delivery metrics (deployment frequency, change lead time, change fail rate, failed deployment recovery time and deployment rework rate), then add one or two quality, product and team-health measures.
Track five to eight, set targets from your own baseline, and never use them to rank individual developers.
KPIs vs metrics
Every KPI is a metric, but most metrics are not KPIs. A metric becomes a KPI when it shows progress toward a goal someone is accountable for. Commit counts are a metric. Change lead time, tracked because the business wants features in customers' hands faster, is a KPI.
The DORA metrics: the standard for delivery
DORA (DevOps Research and Assessment) has studied software delivery performance for over a decade. Its current model has five metrics. Two are about speed (throughput) and three about stability.
Is Your Engineering Velocity Real, or Just a Reporting Illusion?
Discover whether your engineering velocity reflects real output or hidden inefficiency.
| Metric | What it measures | Type |
|---|---|---|
| Deployment frequency | How often you deploy to production | Throughput |
| Change lead time | Time from commit to running in production | Throughput |
| Change fail rate | Share of deployments that need immediate intervention, such as a rollback or hotfix | Stability |
| Failed deployment recovery time | Time to recover from a deployment that failed (this replaced the older "MTTR" metric) | Stability |
| Deployment rework rate | Share of deployments that were unplanned fixes for production incidents | Stability |
DORA's research has consistently found that speed and stability are not a trade-off: teams that do well on one tend to do well on the other.
Beyond delivery: the other KPI categories
| Category | Useful KPIs | What it tells you |
|---|---|---|
| Flow | Cycle time, work in progress, time waiting for review | Where work gets stuck before it ships |
| Quality | Escaped defects, test coverage on critical paths, review coverage | Whether speed is costing reliability |
| Reliability | Uptime against SLO, incident count, error rate | What customers actually experience |
| Product | Feature adoption, retention, customer satisfaction | Whether shipped work creates value |
| Team health | Developer satisfaction, focus time, attrition | Whether the pace is sustainable |
The SPACE framework for developer productivity
SPACE, published in 2021 by researchers from GitHub, Microsoft and the University of Victoria, argues that productivity cannot be captured by one number. It looks at five dimensions:
- Satisfaction and well-being: how fulfilled and healthy developers are.
- Performance: the outcomes of the work, such as quality and customer impact.
- Activity: counts of actions such as commits or reviews, always read with context.
- Communication and collaboration: how well people and teams work together.
- Efficiency and flow: how easily work moves without interruptions or handoffs.
Use DORA to measure the delivery system and SPACE to make sure you are not improving it at the expense of the people running it.
Need more delivery capacity?
Our embedded engineering squads are measured on your KPIs, not hours billed.
Logiciel's senior engineers join your team, set a baseline for delivery and quality metrics in the first weeks, and report against it.
How to choose your KPIs in five steps
- Start from a business goal. Faster time to market points to lead time and deployment frequency. Better retention points to reliability and escaped defects.
- Pick five to eight. Fewer leaves blind spots; more creates noise nobody acts on.
- Pair speed with quality. For every throughput KPI, track a stability one, so nobody wins by cutting corners.
- Measure a baseline first. Collect a month or two of data before setting any target. Targets copied from other companies rarely fit.
- Review quarterly. Drop KPIs that no longer drive decisions and add ones that match new goals.
Five mistakes that make KPIs backfire
- Ranking individuals. Tie KPIs to individual reviews and people will game them. Measure teams.
- Vanity metrics. Lines of code, story points and commit counts measure activity, not value.
- Speed without stability. Velocity alone rewards shipping more bugs faster.
- No context. A number that moved without an explanation leads to wrong conclusions.
- Too many dashboards. If nobody can name the team's KPIs from memory, there are too many.
KPIs when AI writes part of the code
AI coding assistants raise activity numbers quickly: more commits, more pull requests, bigger diffs. That makes activity metrics even less reliable. Watch time waiting for review, change fail rate and deployment rework rate. If those rise while output rises, the extra code is creating work rather than value.
Tools for tracking KPIs
- Work tracking: Jira, Linear, Azure DevOps for cycle time and work in progress.
- Source control and CI/CD: GitHub, GitLab and your pipeline for DORA metrics.
- Engineering intelligence: tools such as LinearB, Jellyfish or DX that combine both.
- Dashboards: Looker, Power BI or Grafana for a single view shared with the business.
Not sure what to measure?
The Architecture Layer That Decides If Your AI Product Survives Production
Build the architecture layers that make AI products production-ready.
Get a 30-minute engineering metrics review.
A senior engineer will help you pick five to eight KPIs that match your goals and show how to collect them from the tools you already use.