Training · Cross-cuttingMoveemergingverified

Proof by Minutes

also known as AI adoption KPI, per-employee AI usage, Copilot assisted hours, weekly Copilot minutes, AI usage time metric, AI adoption dashboard

Tags: KPIadoptionmeasurementCopilotViva-Insightsminutesusage-trackinghabitual-user

Measure how much time per employee is spent actively using AI tools each week — the 'Copilot assisted hours' or 'weekly AI minutes' metric — and use it as the primary Masterpiece KPI for an AI adoption programme. Microsoft's Viva Insights classifies users as power (15+ actions per week for 9 of 12 weeks), habitual, novice, or non-user. Worklytics 2025 benchmarks: low adoption is 30-60 minutes per week; high adoption is 6+ hours per week. The pattern works when minutes are combined with qualitative proof; it fails when minutes alone become a gaming target.

How the learner advances

Intent. Give programme sponsors and managers a single, objective, platform-generated metric that shows whether learners are actually using AI tools at work — not just completing training modules about them.

When to apply. Apply when an AI tool has been deployed to a population and the primary programme question is whether people are actually using it, not just whether they have been trained on it. Apply when the platform provides usage telemetry (Microsoft Viva Insights, Worklytics, or a custom dashboard) so the metric is collected automatically rather than through self-reporting. Apply after the initial training phase to distinguish habitual users from one-time experimenters and identify the novice/non-user population for targeted intervention. Do not apply as the sole measure of success — time spent is a leading indicator of adoption, not a measure of impact. Do not apply when usage telemetry is unavailable and the metric would rely on self-report, which is subject to social desirability bias.

Threshold — earns the next step. At least 60% of the target population has reached habitual-user tier (1+ active AI use day per week for 9 of 12 weeks) and at least one qualitative masterpiece artefact has been produced per learner.

Masterpiece — the artifact that proves it. An adoption dashboard showing the tier distribution across the target population, updated weekly — the programme sponsor's primary evidence that AI tools have become part of how people work, not just a training artefact on a completion report.

Facets

  • Container — async
  • Mode — concept
  • Reach — org
  • Persona — manager-leaderanalyst-ops
  • Craft (AI Fluency) — diligencediscernment

Inputs

  • Usage telemetry — Platform-generated data on the number of AI actions taken per employee per week and the number of weeks in which the employee used the tool. Microsoft Viva Insights, Worklytics, or a custom dashboard are the primary sources.
  • Adoption tier definitions — Clear definitions of what constitutes low, moderate, and high adoption — tied to specific minutes-per-week or actions-per-week thresholds from the platform. Worklytics 2025 benchmarks: low = 30-60 min/week; moderate = 2-4 hours/week; high = 6+ hours/week.
  • Qualitative masterpiece complement — A second measure — a completed work artefact, a peer demo, or a supervisor assessment — that validates whether high-minute users are using AI effectively, not just frequently.

Outputs

  • More capable learner — A learner who has moved from novice or non-user tier to habitual-user tier — not just trained on AI tools but using them regularly enough that the behaviour has become a work habit.
  • Adoption tier distribution — The masterpiece: a programme-level dashboard showing the distribution of power, habitual, novice, and non-user employees — the evidence that the adoption programme is working and the input to the next intervention cycle.

Steps (5)

  1. Enable usage telemetry

    Turn on the platform's usage reporting — Microsoft Viva Insights Copilot Adoption Report, Worklytics, or a custom integration. Define the reporting window (28 days is the Microsoft default) and confirm the tier thresholds before sharing any data with managers.

  2. Set adoption tier thresholds

    Publish the org's tier definitions before the measurement period starts — not after, where thresholds can be adjusted to make the results look better. Use industry benchmarks as a starting point and adjust based on role (a full-time analyst has more opportunity to use AI than a part-time operations staff member).

  3. Report per-team weekly to managers

    Send each manager a weekly digest showing their team's distribution across tiers, with the names of novice and non-user employees highlighted. Make the report actionable: 'these three people have not used Copilot in 4 weeks — a conversation is needed'.

  4. Gate programme thresholds on habitual-user status

    Use habitual-user status (Microsoft: 1+ active day per week for 9 of 12 weeks) as the threshold gate for internal programme completion, not training module completion. A learner who completed the training but never used the tool has not met the threshold.

  5. Combine with qualitative proof

    Require at least one qualitative masterpiece artefact — a prompt library, a completed AI-assisted work product, a peer-reviewed demo — alongside the minutes metric. This catches the edge case of learners who game the metric by running the tool without meaningfully engaging with its output.

Principles

  • Training completion measures exposure; usage minutes measure adoption — only adoption proves the learning transferred to work.
  • Segment by tier before acting — a power user needs a challenge, a non-user needs a barrier-removal conversation, not the same intervention.
  • Minutes alone are a gaming target; minutes plus a masterpiece are a meaningful threshold.

Known uses (2)

Known failure modes (2)

  • [minutes-as-sole-kpi]

    Anti-pattern: treating weekly AI minutes as the only programme success metric. High-minute users can be running AI tools on low-value tasks to hit the metric without improving actual work quality. Minutes must be paired with a qualitative masterpiece.

  • [flat-threshold-across-roles]

    Anti-pattern: applying the same minutes threshold to all roles regardless of how much AI opportunity each role has. A part-time operations worker and a full-time analyst cannot be held to the same weekly minutes target. Role-adjust the thresholds or the comparison is meaningless.

Related trainings (3)

Sources (3)

Provenance

  • Ecosystem: microsoft
  • Added to catalog:
  • Last updated:
  • Verification status: verified