GitHub Copilot Metrics

This article outlines the metrics available for GitHub Copilot, how they are calculated, and how to interpret them to drive adoption and value.

Data Configuration

Before analyzing your metrics, it is important to understand how your connection type (GitHub App vs. GitHub Enterprise PAT) and other settings impact data availability.

  • Connection Requirements: Individual developer data is only available if you connect via a GitHub Enterprise Personal Access Token (PAT). See more about connection methods. Likewise, data is only available for custom groups created within Software.com when a GitHub PAT is added.
  • GitHub App Limitations: If you connect using only our GitHub App, data availability is restricted, per GitHub Copilot’s documentation, to GitHub teams with five or more active Copilot licenses on a given day. 
  • Historical Data: Individual data is only available starting on October 10, 2025 (GitHub documentation). 
  • Telemetry Settings: User metrics rely on IDE telemetry. This is a user-level, not a global setting. If developers turn off telemetry in their IDE, it may affect total AI-assisted developer counts.
  • Group Constraints: Groups containing repositories but no members will not display data. However, groups that contain both members and repositories are not bound by the repository constraint when viewing GitHub Copilot data.
  • Scope: User metrics reflect IDE telemetry data only; they do not include web chat activity or PR summaries generated on the web.

License Utilization

These metrics help you understand your financial efficiency and the distribution of licenses across your organization.

  • Total Licenses: The total number of licenses currently provisioned in your organization.
  • Active Licenses: The count of users who have installed the GitHub Copilot extension.
  • Inactive Licenses: The total number of licenses assigned to users who have not yet installed the Copilot extension in their IDE.
  • Unengaged Users: Users who have installed the Copilot extension, but have not yet engaged with the tool.
  • Seat Utilization Rate: The daily average of seats utilized versus seats available. This is a direct measure of financial efficiency and the first step in understanding adoption. If license utilization is low, you should investigate why developers aren’t using their licenses and consider reallocating them.

Adoption & Engagement

These metrics distinguish between users who merely have the tool and those who are getting value from it.

  • Active Developers: Anyone who created a pull request in the last 90 days.
  • Active Users: Anyone who had any GitHub Copilot activity (including passive activity, such as receiving suggestions, even if they were not accepted).
  • Engaged Users: Users who took a specific, intentional action, such as accepting a suggestion or using IDE chat.
  • Unengaged Users: Users who did not engage with a Copilot feature during the period.
  • Adoption Rate: The percentage of developers who were active in any capacity on a given day. Calculated as: Daily Active Users / Daily Active Developers. A high adoption rate indicates that licenses are being utilized and developers are at least seeing suggestions. A low rate may signal issues with installation, awareness, or perceived value.
  • Engagement Rate: The percentage of active users that were engaged with GitHub Copilot features.
  • Team Engagement Level: High (>80% of active users are engaged), Medium (50-80%), Low (<50%), or Unengaged (0%). This measures deeper value than simple adoption. A high engagement rate means developers are not just seeing suggestions, but actively using them and leveraging other features like IDE chat.
  • User Engagement Level: The percentage of work days a user was engaged: High: >80%, Medium: 50-80%, Low: <50%, Unengaged: 0%.

Feature Usage

These metrics provide granular insight into how developers are interacting with specific AI features.

  • Total IDE Interactions: The total number of times a developer actively engages with GitHub Copilot (e.g., accepting a code suggestion). This is a primary measure of user engagement and perceived value. It indicates how often developers are actively turning to AI for help, rather than just passively receiving suggestions.
  • IDE Chats: The total number of chat sessions initiated by users in their IDE, a raw measure of conversational AI usage.
  • Web Chats: The total number of chat sessions initiated on github.com.
  • PR Summaries: The total number of pull request summaries generated, a raw measure of how much documentation and code review work is being automated.
  • Agent Interactions: The total number of unique days a user initiated at least one interaction with a specialized chat agent. This measures the adoption of advanced, high-leverage AI features.
  • AI Code Acceptance Rate: The percentage of code suggestions that were accepted by a developer. Calculated as: Total Code Acceptances / Total Code Suggestions. This is a direct measure of the quality and relevance of Copilot's code suggestions. A high rate indicates that Copilot is generating useful, high-quality suggestions that align with your team's needs.
  • AI Lines Acceptance Rate: The number of AI lines accepted versus the number of lines suggested.

Still need help? Contact Us Contact Us