Original announcement: GitHub · 2026-11-06. Read the official announcement.
Summary of the Update
On November 6, 2026, GitHub announced enhancements to its enterprise and organization-level Copilot usage metrics API. The update introduces detailed tracking of how developers respond to Copilot-generated code review feedback at the repository level. This includes new metrics to understand the addressed rate of Copilot code review comments and whether further human review requested changes after Copilot’s input.
What Changed in the Copilot Usage Metrics API?
The key addition is a new object called copilot_code_review within each daily repository usage report. This object includes several new fields, such as:
total_comments_evaluated: The number of Copilot code review comments evaluated to determine if developers acted on them.total_comments_addressed: The count of evaluated comments that developers addressed—whether by applying suggestions, asking Copilot for fixes, manually editing code, or committing a fix in a subsequent pull request.- Breakdowns of review comments by severity (high, medium, low), review tier (lite, balanced, max), comment type (bug, security, custom repository guidelines), and customization (custom instructions, skills, managed services).
total_merged_no_human_change_request_after_review: The number of merged pull requests reviewed by Copilot code review that had no subsequent human change requests.
The addressed rate for a repository is calculated by dividing total_comments_addressed by total_comments_evaluated, providing insight into how often Copilot feedback leads to action.
Who Should Care and Why?
This update primarily benefits enterprise owners, organization owners, billing managers, and custom roles with Copilot metrics permissions who want to:
- Measure the effectiveness of Copilot code review feedback within their repositories.
- Understand developer interactions with Copilot’s suggestions beyond just applying ready-to-use recommendations.
- Evaluate the impact of custom configurations such as custom instructions and MCP (Managed Code Protection) servers on code reviews.
- Identify how often human reviewers still request changes after Copilot reviews.
By providing this richer data set, teams can better gauge Copilot’s integration into their review workflows and tailor their usage policies accordingly.
Next Steps for Developers and Teams
To make the most out of these new metrics, organizations should:
- Verify that the Copilot usage metrics policy is enabled for their enterprise or organization.
- Ensure that users accessing the metrics have appropriate permissions (View Copilot Metrics role).
- Review the new
copilot_code_reviewmetrics in daily repository reports to track how code review feedback is addressed over time. - Investigate correlations between types of comments (severity, type) and addressed rates to inform developer training or configuration adjustments.
- Consider using these metrics to optimize custom instructions, skills, or MCP server settings to improve the feedback relevance.
Note that the data on addressed comments is counted when pull requests merge, not when comments are posted, and excludes pull requests opened by bots, except where human change requests are concerned. Access to the updated API documentation will be essential to integrate these data points into existing dashboards or reporting tools.
Important Considerations
- The new metrics appear on each enterprise and organization repository’s daily usage report.
- If a repository lacks Copilot code review activity on a given day, counts default to zero, and arrays are empty.
- These metrics encompass all evaluated comments regardless of whether the developer applied a suggestion directly or indirectly via other actions.
Organizations should confirm these behaviors alongside their actual workflows to ensure the metrics align with their expectations.
