September 18, 2026: GitHub announced Copilot feature engagement reporting on September 17. Its impact dashboard now counts active users who used a feature on at least two days within a 28-day window. The same breakdown reaches enterprise and organization aggregate report APIs.
What changed
The copilot_feature_engagement object includes feature-level counts through totals_by_feature, covering experiences such as code completion, code review, CLI and cloud agent. This breakdown is not included in user-level reports. A person may appear under several features.
GitHub also added users_in_phase_28d for the full rolling population in each AI adoption phase. The existing total_engaged_users still describes users active that day.
Missing or null engagement data means the calculation is unavailable. For phase populations, zero means measured with no users; an omitted field means unmeasured. Access requires an eligible administrative or metrics-viewing role and an enabled usage metrics policy. See the September 17 GitHub announcement.
Why this matters for QA engineers
QATechTools analysis: Repeated use can help QA leads assess whether an AI assistant is becoming part of daily work. It does not establish that generated tests catch more defects. Compare adoption trends with reviewed test changes, flaky-test rates and escaped defects before judging a rollout.
For teams testing an internal analytics dashboard, useful checks include preserving missing values, avoiding sums that double-count people across features, and clearly labeling daily activity versus rolling populations. These are suggested validation checks, not new GitHub testing features.
Where to start
Ask your reporting owner to inspect the latest 28-day aggregate report. GitHub documents daily report generation and download links with limited validity in its Copilot usage metrics API reference. Verify the report dates before comparing periods.
Sources checked September 18, 2026. Illustration is editorial, not a product screenshot.