Playbooks · Teams tasks
Copilot reviews: how many get addressed
Written November 6, 2026
GitHub's usage metrics API now reports code-review details and addressed rates. What the numbers measure, and what they still can't show.
What the API now reports
The enterprise and organization repository-level Copilot usage metrics reports now show what happens to Copilot code review feedback. A new copilot_code_review object on each repos-1-day row reports how many Copilot-reviewed pull requests merged without a later change request from a person, and how many Copilot code review comments developers acted on. With these you can track an addressed rate per repository.
total_comments_evaluated: the number of Copilot code review comments checked to see whether the developer acted on them.total_comments_addressed: the number of evaluated comments the developer acted on — whether by applying the suggestion, asking Copilot to make the fix, editing the code, or replying that the fix will come in a later pull request. Divide bytotal_comments_evaluatedfor the addressed rate.- Breakdowns:
review_comments_by_severity,review_comments_by_review_tier,review_comments_by_comment_type, andreview_comments_by_customization— counts split by severity (high, medium, low), review tier (lite, balanced, max), comment type (bug, security, or custom for your repository's own coding guidelines), and customization (custom instructions, skills, MCP servers, or none). total_merged_no_human_change_request_after_review: the number of merged pull requests that Copilot code review reviewed where no person requested changes after Copilot's first review.
The fields are present on every row of the enterprise and organization repos-1-day reports. Counts are 0 and arrays are empty when a repository had no Copilot code review activity that day. Access needs an enterprise owner, enterprise billing manager, organization owner, or anyone with a custom organization or enterprise role that grants the View Copilot Metrics permission — and the Copilot usage metrics policy must be enabled.
The counting rules change what the numbers mean
- Counted on merge. Comments are evaluated when their pull request merges, so they count on the merge day, not the day Copilot posted them. A comment posted on Monday and merged on Friday shows up in Friday's row.
- "Addressed" is broad. A reply saying the fix will come in a later pull request counts as addressed, the same as applying the suggestion. The rate measures whether developers responded, not whether the code changed in that pull request.
- Bot pull requests are mostly out — with one exception. Copilot usually does not evaluate comments on pull requests opened by bots, such as Copilot cloud agent. But
total_merged_no_human_change_request_after_reviewstill includes those pull requests, so the merged-without-change-request count and the comment counts cover different sets of pull requests. - Different from suggestion counts. The existing
pull_requests.total_copilot_applied_suggestionscounts only ready-to-apply suggestions that someone applied. The new counts include every evaluated comment and count them as addressed however the developer acted on them. The two numbers will not match, and neither is the other one done wrong.
The companion change: telemetry and version gaps
A second changelog the same day adds fields that say how complete the underlying data is. New per-user fields appear in enterprise and organization user reports, and daily affected-user counts appear in enterprise and organization reports.
client_telemetry_present: whether client telemetry was received for the user that day.falsemeans Copilot saw the user's activity, but no usage events arrived from their editor. Common causes include outdated editors or extensions, disabled IDE telemetry, and proxies or firewalls that block events.usage_metrics_requirements: each of the user's IDEs and Copilot extensions listed asmet,unmet, orunknownagainst the minimum supported versions.missing_client_telemetry_user_count: the number of users that day withclient_telemetry_presentset tofalse.below_usage_metrics_minimum_user_count: the number of users that day with at least one IDE or extension below the minimum version.
Three rules govern how to read these. A user can be counted in both counts, so do not add the two counts together. Do not add counts across days to count unique users — each count covers a single day. And null or omitted values mean the data is unavailable, not false or zero; days before this release have no values. A user can also meet every version minimum and still be missing telemetry, for example behind a proxy. Users without client telemetry still count as active, but their reports lack detail such as lines of code, feature breakdowns, and model breakdowns.
What the numbers still cannot show
- Addressed is not correct. The rate says a developer did something in response to a comment. It says nothing about whether the fix was right, whether the review caught a real bug, or whether the author already knew about the problem.
- A high rate can mean easy comments. Breakdowns by severity and comment type exist for exactly this reason: a repository whose addressed rate is high on low-severity style notes and low on security findings is not getting the same value as the headline rate suggests. Read the splits before quoting the headline.
- Gaps in the data look like quiet users. A team with missing telemetry or below-minimum versions will show thin per-user detail. Before concluding a team barely uses Copilot, check the telemetry and version counts — the usage may be there with no events arriving to record it.
Limits
- No metrics API data is reported here. This page follows the two GitHub changelogs; no organization or enterprise usage report was pulled while writing it.
- Comments count on the merge day, so a repository's addressed rate for a given day reflects pull requests that merged that day — not the reviews Copilot posted that day.
- The fields describe repository-level Copilot code review. They do not cover reviews Copilot performed anywhere the usage metrics reports do not reach.
Related playbooks
- Copilot code review now has an API
- Copilot Agent Merge: watch a PR and fix its blockers
- The Copilot app now takes GHES accounts
- Copilot Automations learn from real outcomes
Sources
- Usage metrics API adds code review details and addressed rate — GitHub Changelog
- Usage metrics API flags telemetry and version gaps — GitHub Changelog
Back to: Small-team playbooks hub · Also: Review a pull request with a coding agent · Triage a backlog of issues with an agent · Decide whether Copilot should approve PRs