All insights
2 min read

GitHub’s new review metrics show where pull requests wait

GitHub now separates review waiting times into three stages. Here is how to read the numbers without turning them into a productivity score.

Where code reviews get delayed — AtmosGate Insights editorial cover

A change can be ready for days while a team waits for a reviewer, resolves feedback or schedules the final merge. Those are different problems, even when they produce the same long delivery time.

GitHub’s September 25, 2026 update adds review-stage timings to its repository-level Copilot usage reports. The new breakdown is useful context for a conversation about how work moves through a team.

Three stages, two useful views

The reports separate ready-for-review to first review, first to final review, and final review to merge. Each stage includes a median and a 90th percentile, which helps show the difference between typical waits and the slower end of the distribution.

These fields are available in enterprise and organization repos-1-day reports. Access requires the appropriate metrics permission and the Copilot usage metrics policy to be enabled.

Sources and release notes: [1]

Understand which changes are counted

This release times human reviews of qualifying human-authored pull requests. Bot reviews and the author’s own reviews are excluded from the timing. Historical data is not backfilled; pull requests ready for review before September 21 are excluded from this section.

A day without qualifying merges has an empty array. That should not be reported as a measured zero-minute review time.

Sources and release notes: [1]

Use the data to improve the process

Our recommendation is to discuss one slow stage with the team. Waiting for a first review might call for clearer ownership. Repeated review cycles might point to oversized changes or unclear requirements. A long wait after approval may need a release-process adjustment.

Avoid ranking individual engineers with these numbers. Pair them with the type of work, reviewer availability and the team’s quality checks. The update describes elapsed review time; it does not establish that AI caused a productivity improvement.

Sources and release notes

Original analysis by AtmosGate, based on the sources below. Vendor claims and release status are attributed to their publishers.

  1. Usage metrics API adds pull request review stages

    GitHub · 25 September 2026