Metrics guide
How velocity, throughput, burndown and workload are calculated, and what they exclude.
Metrics guide
Velocity
Sum of storyPoints on tasks whose status is done, grouped by the sprint they belong to, using updatedAt as the completion time. A task moved out of done stops counting. Tasks with no story points contribute zero, which is why unestimated work makes velocity look flat rather than wrong.
Burndown
Open points at sprint start minus points completed each day, plotted against an ideal straight line from the sprint start date to its end date. The AI-projected line on /dashboard/ai-scrum-master uses the same series and extends it.
Weekly throughput
Count of tasks that reached done in each of the last N weeks. It counts tasks, not points, so it is comparable across teams with different estimation habits and not comparable to velocity.
Cumulative flow
Task counts per status per day. Widening bands mean work is accumulating in that status — a review band that keeps growing is a review-capacity problem, not a coding problem.
Assignee workload
Open tasks and open points per member, from the same filter the dashboard is using. Pair it with the wellbeing page before acting on it: load alone does not tell you whether someone is struggling.
Due-date pressure
Tasks with a dueDate inside the selected range that are not done, grouped by how close the date is.
Quality signals
Blocked-task count and time in review, both derived from status changes rather than from any code analysis.
What is deliberately excluded
- Lines of code, commit counts and keystrokes.
- Per-person velocity rankings.
- Activity or presence scores.
The exclusion list is published on /dashboard/ethical-metrics, which also prints the formula behind each indicator it shows.
Comparing teams
Compare throughput and predictability, not velocity. Velocity is only meaningful inside one team's own history because story points are a local unit.