U2U GitHub's Copilot Impact Dashboard Now Shows Which Features Developers Actually Use
GitHub's September 17 changelog adds feature engagement tracking to the Copilot impact dashboard and report APIs — enterprise admins can now see adoption broken down by code completion, agent edit, code review, cloud agent, CLI, and app.
Enterprise AI rollouts have a measurement problem: licenses get bought, but nobody can say which features developers actually fold into their daily work. GitHub’s September 17 changelog takes a step toward fixing that — the Copilot impact dashboard now shows how many active users regularly engage with each key Copilot feature.
Engagement, defined and broken down
The dashboard counts a user as engaged with a feature if they used it on at least two days during a 28-day period — a deliberately low bar that distinguishes “tried it once” from “part of the workflow.” The same data lands in enterprise and organization report APIs, which now include a 28-day feature engagement breakdown with three new objects: copilot_feature_engagement adds the dashboard’s active-user total and engagement counts to 28-day aggregate reports; totals_by_feature splits engagement across code completion, agent edit, passive Copilot code review, active Copilot code review, Copilot cloud agent, Copilot CLI, and Copilot app.
The changelog also draws a useful distinction between the two code review modes: active code review means a user manually requested a Copilot review or applied one of its suggestions, while passive means Copilot was automatically assigned to review the user’s pull request without the user engaging with the review. For an admin trying to tell genuine adoption from ambient automation, that’s exactly the right split.
Better adoption-phase reporting
Alongside feature engagement, GitHub improved AI adoption phase reporting: the new users_in_phase_28d field reports the full rolling 28-day population classified into each adoption phase as of each report day. Previously, the phase breakdown only showed users in each phase who were active on a given day, which made cohorts look smaller than they were and noisy day to day. The existing total_engaged_users field keeps its old day-active-only definition.
A few guardrails: feature engagement appears only in enterprise and organization 28-day aggregate reports, not user-level reports; a user can be counted under more than one feature; and the data is available to enterprise owners and billing managers, organization owners, and custom roles with the View Copilot Metrics permission, with the Copilot usage metrics policy enabled.
The strategic point is straightforward. As AI coding tools move from individual subscriptions to enterprise-wide deployments, the question buyers ask changes from “does it work” to “is it being used, and where.” Usage data down to the feature level is how platform teams decide where to invest in training, which experiences to enable or restrict, and whether the renewal is worth it. GitHub is now selling the instrumentation alongside the tool.