Metrics Overview

Metrics are key to organizational improvement. If we do not measure, then any attempt at improvement is aimless. Metrics, like any tool, must be used correctly to drive the improvement we need. It’s important to use metrics in offsetting groups and to focus improvement efforts on the group of metrics as a whole, not as individual measures.

The Metrics Cheat Sheet has a high-level view of the key metrics, their intent, and how to use them appropriately.

Goodhart’s Law

CD Execution

When measuring the performance of continuous delivery, we are measuring our ability to reliably and sustainably deliver high-quality changes. We do this by focusing on very frequent small batches of high-quality delivery.

  • Change frequency is important to make sure that waste is driven out of the process. This reduces costs, improves the sustainability of flow, and ensures there is a verified quality process for emergency changes.
  • Small batches are easier to inspect for quality and limit the impact of any quality issues.
  • Change success is an important offsetting metric. If we only focus on change size and change frequency, quality will suffer. If we only focus on reducing the number of defects and eliminating impacting changes, batch size and frequency suffer. The data shows that this results in more defects and more costs.


  • Development Cycle Time: Time from when a task is started until it is “Done”. The suggested definition of “Done” is delivered to production. KPI for how big a unit of work is. Indicator of possible upstream quality issues with requirements definition and teamwork.
  • Delivery Frequency: KPI for batch size, cost, and efficient quality process.


  • Change Failure Rate: Percentage of changes that require remediation. KPI for the effectiveness of the quality process.
  • Defect Rate: Rate of defect creation over time relative to change frequency, generally P1 and P2.
  • Mean Time to Repair: KPI for the maturity of our disaster response preparations and the forethought to design for recovery instead of just for delivery.

CI Execution

Continuous delivery stands on the bedrock of continuous integration. If code is not continuously integrating, it cannot be safely delivered.

The focus of CI is to amplify quality feedback. The more frequently code is integrated and tested, the sooner any quality issues will be found and the smaller those issues will be.

Integration Frequency: Frequency of code integrating to the trunk. KPI for efficacy of

refining requirements, quality process, and teamwork.

  • When a team is mob programming, this should occur several times a day.
  • When a team is pair programming, this should occur several times a day per pair.
  • When the team is working on individual tasks, this should occur several times a day per developer.

Build Cycle Time: Time from commit to production deploy. KPI for the stability of the

pipeline and efficiency of the quality process. Long build cycle times have a direct negative impact on MTTR, and batch size. This encourages abandoning defined quality checks in emergencies. This makes emergency changes the riskiest changes to make.

Workflow Management

Velocity / Throughput: Planning metric to allow the team to predict date ranges for delivery. The

standard deviation of this metric is a KPI for predictability. The average value of the metric has no meaning outside the team.

Lead Time: Total time from when a request is made until it is delivered. KPI for team over-utilization

As the team’s utilization approaches 100%, this metric approaches infinity.

Work In Process (WIP): Key flow metric. Excessive WIP results in rework and delivery delays

Last modified February 7, 2024: Update metrics page for RAG (e9a7b26)