KPIs — Key Performance Indicators

The Distinction

OKRs and KPIs are often confused, and the confusion produces dysfunction. The cleanest distinction:

• OKRs are ambitions for a defined period. They change quarterly. They describe what the team is trying to improve.
• KPIs are ongoing measures of operational health. They stay stable for years. They describe what the team is trying to maintain.

A retail store's revenue is a KPI — it's measured every period, and the team is responsible for keeping it healthy. A specific revenue target for Q3 (15% growth) is an OKR — it's a time-bounded ambition.

What Good KPIs Look Like

• Stable over time. The same KPI is measured year after year, allowing meaningful trends.
• Lagging or leading. Most KPIs are lagging (they report what already happened), but the best ones include leading indicators (they predict what's coming).
• Tied to the team's responsibility. The team can directly influence the number.
• Aggregable. KPIs roll up cleanly across teams, departments, and the organization.
• Bounded. A team usually has between three and ten KPIs. Above that, attention dilutes.

Common KPI Categories for Agile Teams

Delivery

• Cycle time (median, p85).
• Throughput (items per week).
• Deployment frequency.
• Lead time for changes.

Quality

• Defect escape rate.
• Production incident frequency.
• Mean time to recovery (MTTR).
• Change failure rate.

Product

• Daily/weekly/monthly active users.
• Retention curves.
• Feature adoption rates.

Customer

• CSAT or NPS (with caveats — see NPS for Agile Teams).
• Customer support ticket volume.
• Churn rate.

Team health

• Happiness pulse score (see Team Happiness Metrics).
• Retention rate (employee).
• Velocity stability (if used).

The DORA Four

One of the most widely-adopted KPI sets for agile delivery is the DORA metrics — the four measures identified by the State of DevOps research as predictive of high-performing teams:¹

• Deployment Frequency. How often the team ships to production.
• Lead Time for Changes. Code commit to production.
• Change Failure Rate. Percentage of deployments that cause incidents.
• Mean Time to Recovery (MTTR). How fast the team recovers from incidents.

These four work as a balanced set: the first two reward speed; the second two prevent the speed from coming at the cost of quality. Together, they form a useful KPI baseline for any delivery team.

How KPIs Go Wrong

• Vanity KPIs. Numbers that move but don't reflect what matters. Pageviews, signups (without activation), follower counts.
• Dashboard creep. Thirty KPIs on a dashboard. Nobody looks at any of them carefully.
• KPI as performance review. Same problem as OKR-as-performance-review. Sandbagging follows.
• Output KPIs. "Number of features shipped" is an output. Outcomes matter; outputs don't.
• KPIs without thresholds. Tracking a number with no sense of what good or bad looks like produces wallpaper.
• Stale KPIs. A KPI that mattered three years ago and no longer matches the business. Audit and prune.

Using KPIs With OKRs

The pairing of KPIs and OKRs is often what produces the most useful structure:

• KPIs are the health dashboard. They report what the team has and what it's protecting.
• OKRs are the quarterly bet. They describe what the team is trying to change about those KPIs (or about something the KPIs don't yet measure).

This separation prevents two common failures: KPI dashboards that produce no action (because nothing is targeted) and OKRs that ignore the rest of the system (because they live in their own ambition-shaped silo).

Coaching Tips

Start with the DORA four.

For any delivery team, the baseline. Add other KPIs only after these are instrumented and trending.

Set thresholds, not targets.

"Healthy / watching / unhealthy" beats "must be 95%." Thresholds enable action; targets enable gaming.

Audit KPIs annually.

Some matter less than they did. Pruning matters as much as adding.

Track leading and lagging together.

Lagging indicators tell you what happened; leading indicators tell you what's coming. The pairing makes the dashboard actionable.

Don't confuse KPI with OKR.

KPIs are stable; OKRs are quarterly bets. Conflating them produces ambiguous management of both.

Watch for output KPIs.

"Features shipped" is an output, not a result. Reward outcomes; output KPIs reliably distort behavior.

Summary

KPIs are the stable measures of operational health that complement the ambition-driven changes captured in OKRs. The discipline lies in choosing a small set of KPIs that genuinely reflect what the team is responsible for, instrumenting them well, and reviewing them often enough that trends become visible. The DORA four are the modern baseline for delivery teams; everything else should be additive only if it earns its place. Most KPI dashboards have too many numbers and too little attention to any of them — the fix is pruning, not adding.

Back to Agile Metrics