Metrics glossary

Every KPI on the dashboard uses the same definitions org-wide. This page explains what each metric means, what data it includes, and how to interpret it in team and contributor reviews.

Metric	Definition	GitHub source	Pilot
Merge throughput	Merged PRs per period	merged_at	Yes
Lead time	Median open → merge hours	PR timestamps	Yes
Review responsiveness	Median time to first human review	Review events	Yes
Productivity score	0–100 composite of core KPIs	Derived	Yes
Deploy frequency	Production releases	CI/CD	No
Change failure rate (full)	Incidents per deploy	Incidents	No

DeliveryCompass calculates metrics at three levels. The formula is the same; the repositories included differ:

Level	Repositories included
Organization	All ingested, non-excluded repositories in your workspace
Team	Only repositories you assigned to that team on Teams & repos
Contributor	All ingested repositories (author activity org-wide)

Unassigned repositories count toward organization metrics but not toward any team card until mapped.

Merge throughput

Count of pull requests merged during the selected period. Open PRs and PRs closed without merge are excluded.

Scope: Organization uses all ingested repos; teams use mapped repos only; contributors count PRs they authored and merged.

Lead time

Median hours from PR creation to merge for PRs merged in the period. Measures end-to-end delivery time for shipped work.

Formula: median(merged_at − created_at) for merged PRs in the selected range.

Review responsiveness

Median hours from PR creation to the first meaningful human review (approved, changes requested, or commented). Bot-only reviews are excluded by default.

Formula: median(first_review_submitted_at − created_at) for PRs with at least one human review in the period.

Productivity score

Composite score from 0 to 100 combining merge throughput, lead time, and review responsiveness with equal weight. Higher is better.

Each component is normalized within the selected period before combining, so teams and contributors are compared on a consistent scale for that time window.

Contributor views use the same core KPIs as teams, plus code-change statistics from merged PRs — GitHub-reported line counts and filename extension mix, not stored source code:

• Lines added / removed — summed across merged PRs in the period
• Average lines per PR and average files per PR
• File-type mix — breakdown by file extension from changed files on merged PRs
• Period delta — change vs the previous period of the same length

See the dashboard guide for contributor drill-down charts and tables.

The weekly summary and Sunday digest flag teams and contributors as outperforming or lagging compared to the previous period of the same length (for example, this 30 days vs the prior 30 days).

These are comparative coaching signals — not performance ratings. Use them to decide where to dig deeper in charts or 1:1 conversations, especially when sample size is sufficient.

When merged PR count in a period is small, metrics may shift significantly as more work lands. The dashboard shows a confidence label (low / medium / high) based on sample size.

Treat low-confidence periods as directional — useful for spotting trends early, not for final judgments.

Most KPI cards show the current value plus a delta vs the previous period of the same length. A 30-day range compares to the prior 30 days; a calendar year compares to the prior year.

Example: Team A shows 42 merged PRs (+8 vs prior 30d), median lead time 18h (−4h), review response 6h (+1h). Throughput improved while review lag ticked up — a reasonable coaching topic before blaming individuals.

Use deltas together with the confidence label — a +50% throughput jump on three merges is less meaningful than the same change on forty merges.

The org overview adds delivery-focused views beyond the core KPIs:

Revert rate

Percentage of merged pull requests flagged as reverts (title-based) among all merges in the selected period. Useful as a stability signal alongside throughput and lead time.

Work in progress by team

Point-in-time count of open pull requests in repositories assigned to each team. Repositories not mapped to a team appear in an Unassigned bucket.

Merged code type (file extension mix)

Distribution of filename extensions across files touched in merged PRs during the period. Based on extension counts from GitHub metadata — not language analysis or file contents.

Merged lines of code (monthly)

GitHub-reported additions and deletions on merged pull requests, grouped by calendar month of merge. Shown as grouped bars for the months overlapping the selected range.

By default, bot-authored PRs and bot-only reviews are excluded from throughput, lead time, review responsiveness, and productivity score.

Enable Include bots on the dashboard when you want automation (for example Dependabot) reflected in volume metrics. Most teams leave this off for coaching conversations.