Usage metrics API adds pull request review stages

GitHub expanded its enterprise and organization usage metrics reports by adding a new array to daily repository data. This array provides median and 90th percentile timings for three specific review phases of pull requests. The stages cover the time from ready for review to the first review, between the first and final reviews, and from the final review to merging. Teams can now identify where delays occur in the code review workflow rather than just seeing total merge times. The data reveals whether waits stem from initial review waits, iterative feedback loops, or approved code sitting unmerged. The source notes that each stage points to a different fix for workflow bottlenecks. High percentile values help distinguish consistent slowness from a few exceptional pull requests driving the average. The metric excludes bot interactions and author self-reviews, so the total merged count here is typically lower than the overall pull request total. Data does not backfill previous dates, leaving early days with thin data since the release date of September 21, 2026. An empty array signifies no qualifying merges occurred on a given day, while a single review results in a zero for the middle stage duration.