Adoption Metrics Guide

Definitions, scope, and interpretation for Phase 1 metrics.

Use this page to answer four questions:

Question	Primary metrics
Reach - how many licensed users are active?	DAU, WAU, MAU, seat utilization
Depth - are users moving beyond simple completions?	Feature mix, chat requests per user, agent adoption %
Quality - are suggestions useful enough to keep trust?	Acceptance rate, lines added with AI
Context - where is adoption strongest or weakest?	IDE, language, and model distribution

Metric Definitions

Metric	What it measures	How to use it	Scope
DAU	Unique users interacting with Copilot on a given day	Best weekly operating signal for day-to-day adoption momentum	Enterprise, Org
WAU	Unique users in a 7-day rolling window	Use with DAU to see whether usage is habitual vs. occasional	Enterprise, Org
MAU	Unique users in a 28-day rolling window	Best high-level view of sustained adoption	Enterprise, Org
Seat Utilization	Active users divided by assigned seats	Flags over-licensing or inactive cohorts that need enablement	Enterprise, Org
Acceptance Rate	Accepted suggestions divided by suggestions shown	Best proxy for suggestion relevance and developer trust	Enterprise, Org, User
Chat Requests / User	Average chat interactions per active user	Shows whether users are moving from completions into conversational workflows	Enterprise, Org
Agent Adoption %	Share of active users using agent or edit modes	Good signal that teams are exploring more advanced Copilot workflows	Enterprise, Org
Feature Mix	Usage split across completions, chat, CLI, PR summaries, and agent workflows	Use to tell whether adoption is broadening or stuck in one mode	Enterprise, Org
Lines Added with AI	Lines accepted into the editor from suggestions	Useful output-volume indicator, but not a productivity metric on its own	Enterprise, Org
IDE Distribution	Usage breakdown by editor	Helps identify unsupported or under-configured IDE cohorts	Enterprise, Org
Language Distribution	Usage breakdown by programming language	Helps spot where Copilot is most relevant or least trusted	Enterprise, Org
Model Distribution	Usage breakdown by AI model	Useful for change management when model usage shifts after policy or feature changes	Enterprise, Org

Interpretation heuristics

Acceptance rate of 25-35% is common, but the trend matters more than the number.
A DAU/MAU ratio above 50% usually indicates repeat usage; 60%+ is a strong Phase 2 readiness signal.
Seat utilization above 70% is generally healthy. Lower rates usually mean licensing or enablement cleanup is needed.
Rising feature mix breadth is often a stronger adoption signal than lines added alone.

How to Read the Native Dashboards

Dashboard	Review each week	Why it matters
Usage & Adoption	DAU, WAU, MAU, seat utilization, feature mix	Tells you whether Copilot is spreading and becoming routine
Code Generation	Agent adoption, lines added with AI, model and language mix	Tells you how deeply developers are using more advanced flows

Weekly review sequence

Check active users first - Is DAU or WAU moving in the right direction?
Check quality next - Is acceptance rate stable or improving?
Check breadth - Are more users trying chat, CLI, or agent workflows?
Check breakdowns - Are low-adoption pockets concentrated in one IDE, language, or team?

Good operating rhythm

Review leading indicators weekly, then summarize lagging indicators monthly for leadership.

Diagnosing Adoption Patterns

Pattern	What it usually means	Action
Low DAU + many assigned seats	Enablement or licensing gap	Run onboarding, reclaim inactive seats, confirm extension rollout
High DAU + low acceptance	Suggestions are visible but not trusted	Review language coverage, coding patterns, and prompt habits
High acceptance + low chat / agent usage	Teams use Copilot only for inline completions	Promote chat, edit, and multi-step workflows
Strong usage in one IDE, weak in another	Platform friction	Audit IDE versions, extension deployment, and policy support
Healthy MAU + weak DAU/MAU ratio	Lots of trial, limited habit formation	Focus on weekly use cases and team-level champions
High lines added + flat feature mix	Output volume is growing, but workflow depth is not	Pair enablement with examples beyond code completion

Scope, Freshness, and Limitations

What IS included

IDE telemetry for completions, chat, agent mode, and PR-related workflows. Users must have telemetry enabled.

What is NOT included

Copilot Chat on GitHub.com
GitHub Mobile activity
License or seat assignment administration data (use the Copilot User Management API)

Freshness and time windows

New activity usually appears within a few hours, but historical data can take up to 24 hours to populate.
WAU is a rolling 7-day view and MAU is a rolling 28-day view - they will move more slowly than DAU.
If you need retention beyond the native API window, set up daily exports or automated collection.

Minimum IDE Versions

IDE	Minimum Version
VS Code	1.101+
JetBrains	2024.2.6+
Visual Studio	17.14.13+
Eclipse	4.31+
Xcode	13.2.1+

Attribution Rules

Org-level metrics are based on org membership, not seat assignment. A user can appear in multiple orgs.
Enterprise-level metrics deduplicate users across orgs.
Org-level data is available from December 12, 2025 onward.
Double-counting is expected when you sum org totals across an enterprise with shared users.

Do not compare unmatched scopes

Enterprise MAU and the sum of org MAUs are answering different questions. Use one scope consistently in executive reporting.

Leading vs Lagging Indicators

Type	Metric	Best use
Leading	DAU/WAU growth	Weekly adoption operations
Leading	Acceptance rate trend	Detect trust or quality issues early
Leading	Agent adoption %	Track advanced workflow uptake
Leading	Feature mix breadth	Confirm adoption is deepening
Lagging	MAU (28-day)	Monthly or quarterly adoption summaries
Lagging	Seat utilization trend	License optimization and renewal discussions
Lagging	Lines added with AI	Output volume context for exec narratives

Tip

Use leading indicators for team actions and lagging indicators for monthly or quarterly business reviews.

Common Analysis Mistakes

Treating acceptance rate as a leaderboard instead of a trend signal
Comparing lines added across languages with very different coding styles
Declaring ROI from adoption metrics alone before adding delivery outcomes
Ignoring feature mix and focusing only on active-user counts
Summing org metrics across an enterprise and assuming the result is deduplicated