GitHub exposes team Copilot metrics

GitHub said on May 14 that its Copilot metrics API now lets enterprise customers analyze usage by team, extending a reporting system that had previously centered on organization- or enterprise-wide totals. The update gives organizations a way to connect licensed users to the teams they belong to and then aggregate Copilot activity at that level, according to GitHub’s changelog and documentation. The release applies to team-level reporting through the REST API, not a new dashboard view. GitHub published the change one day after a separate Copilot metrics update added more detail on code review suggestions. ### Which new data did GitHub add on May 14? GitHub’s May 14 changelog entry said “team-level Copilot usage metrics” are now available through the API. The company said customers can break down usage across IDE completions, chat, Copilot CLI, code review and Copilot cloud agent activity, with cuts by language, IDE, feature or model. The GitHub Docs page on team-level metrics says the company does not publish a single pre-aggregated team report. Instead, customers construct team metrics by joining a user-teams report, which lists each user’s team memberships for a given day, with a per-user usage metrics report for that same day. ### If there is no dashboard, how do companies actually use it? (github.blog) GitHub’s changelog said the new user-teams reports are available “through the REST API only.” The same post said there is “no dashboard surface for team-level metrics in this release,” meaning customers need to retrieve the underlying reports and assemble the breakdown themselves. (docs.github.com) The REST API documentation says the Copilot Metrics API access policy must be enabled for the relevant organization, and that organization owners — along with owners and billing managers of the parent enterprise — can view team metrics. The docs also say OAuth app tokens and classic personal access tokens need scopes including `manage_billing:copilot`, `read:org`, or `read:enterprise` for access. (github.blog) ### What are the limits on the team numbers? GitHub’s May 14 post said teams with fewer than five Copilot-seated users are excluded from the user-teams report. The company added that those users’ individual activity can still appear in the per-user usage report. GitHub’s documentation also says users who belong to multiple teams are counted in each team’s aggregate. (docs.github.com) That means team totals cannot be summed to reproduce an organization-wide or enterprise-wide total, a caveat the company highlighted in both the changelog and docs. ### How does this fit with GitHub’s broader Copilot metrics rollout? (github.blog) GitHub said on February 27 that Copilot metrics had reached general availability after a public preview announced at Universe 2025. In that earlier post, the company said the product was meant to give customers “a single place” to track adoption and usage trends and to build reports that fit their organizations. (github.blog) On May 8, GitHub added another reporting change, saying the usage metrics API now breaks down Copilot code review suggestions by comment type in enterprise and organization reports. That sequence shows GitHub continuing to add more granular reporting fields around how Copilot is being used inside companies, according to the company’s changelog. (github.blog) ### What data is still outside these reports? GitHub’s usage-metrics reference says license and seat management data are not included in Copilot usage metrics reports. The company says customers should use the Copilot user management API as the source of truth for license and seat information. GitHub’s metrics documentation also says many measures rely on IDE telemetry, and users must have telemetry enabled in their IDE to be counted. (github.blog) The same reference says activity from some other surfaces, including Copilot Chat on GitHub.com, is not included in those metrics. GitHub’s next step for customers is set out in its team-level metrics documentation, which includes the join recipe and rolling-window aggregation guidance for daily reports. (docs.github.com) The company’s REST API docs say metrics are processed once per day for the previous day and can return up to 100 days of data. (docs.github.com)