Missing telemetry is a measurement problem, not evidence that employees are inactive. GitHub’s November 5, 2026 update to Copilot usage reports makes a previously hidden problem visible: some users register as active while their editor usage events never arrive. Enterprises that use usage dashboards to allocate licenses, measure adoption or justify engineering investments need to audit data completeness before drawing conclusions.
What changed in the reporting API
GitHub added two user-level signals: client_telemetry_present identifies whether editor events arrived that day, and usage_metrics_requirements reports whether the user’s IDE and extension versions meet supported minimums. Daily aggregate fields count users missing telemetry and users below minimum versions. The fields are available in enterprise and organization reports, subject to the appropriate permission and enabled usage-metrics policy. This is an observability improvement, not a new claim about developer productivity.
GitHub specifically notes that active users without client telemetry can lack detailed lines-of-code, feature and model breakdowns. Potential causes include outdated software, disabled IDE telemetry, and network intermediaries that block events. A compliant version does not guarantee that telemetry arrives, so version compliance and event delivery must be tested separately.
Why adoption dashboards can mislead finance
Suppose a department has 200 active users and 40 have no editor telemetry on a given day. A report based only on detailed editor events would describe activity for 160 people, not the entire active population. Dividing observed feature counts by 200 might understate activity; dividing by 160 might overstate adoption among all users. Neither denominator is automatically appropriate for a license-renewal decision. This is a hypothetical example, not a measured GitHub customer outcome.
These distinctions matter for the cost accounting described in our Copilot usage metrics analysis and for assessing model-routing economics. An incomplete measurement system can make a lower-cost option appear to be less used or a higher-cost option appear to be more productive, without supporting evidence.
Three separate data-quality questions
Coverage: How many users are in scope, active and reporting detailed telemetry? Preserve all three counts. Compatibility: Which editor and extension versions fall below the published minimum? Continuity: Do event gaps cluster by network, operating system, business unit or day? A single aggregate count hides whether the issue is a local configuration problem or a persistent reporting blind spot.
Active users → Telemetry present? → Version requirements met? → Detailed metrics eligible for interpretation
Track missing telemetry and outdated versions independently; users may appear in both groups.
Six tests before trusting the trend line
1. Reconcile populations. Compare entitled users, active users and users with detailed events. Investigate the differences without labeling every gap as inactivity.
2. Avoid double counting. GitHub warns that the missing-telemetry and below-minimum-version daily counts can overlap. Their sum is not a unique affected-user count. Likewise, summing daily counts across a month does not produce unique monthly users.
3. Preserve missingness. Null or omitted values mean data is unavailable, not false or zero. Historical days before the November 5 release have no values; do not turn these into artificial improvements in a trend chart.
4. Test the network path. Compare affected and unaffected devices behind the same proxy, firewall and security controls. Document approved changes and avoid weakening corporate monitoring policy simply to improve a dashboard.
5. Segment compatibility. Identify IDE and extension versions that fail minimums; use controlled updates and retest event delivery. A version update may not resolve blocked network telemetry.
6. Qualify financial decisions. Report the percentage of active users with observable detailed activity alongside any utilization, code-change or cost metric. Defer punitive individual assessments or major license changes when coverage is materially incomplete.
What a defensible executive dashboard shows
A useful monthly scorecard separates seats assigned, active users, users with valid detailed telemetry, version-compliance gaps, usage by feature and cost per eligible population. Include a clear caveat where detailed metrics cannot represent all active users. Compare like-for-like cohorts and track fixes to telemetry coverage as an operational task, not as a productivity target.
The broader principle applies to other enterprise measurement systems: absence of an event is not evidence of absence of work. For agent workflows, auditability also depends on reliable identity and execution records, as discussed in our agent audit-trail framework.
Sources and limitations
Primary source: GitHub Changelog, November 5, 2026. The field definitions, overlapping daily counts, null semantics and permission conditions above are drawn from GitHub’s announcement. For related metric definitions, see GitHub’s November 5 code-review metrics release. The data-quality framework and numerical example are original analysis rather than claims made by GitHub.
Leave A Comment