Data available in Copilot usage metrics

The Copilot usage metrics dashboard and APIs display and export data using a consistent set of fields. This reference lists all available metrics and describes how to interpret their values in both dashboard visuals and NDJSON or API exports.

• The Copilot usage metrics dashboards are available at the enterprise and organization level.
• The Copilot usage metrics APIs return reports at the enterprise or organization scope, with aggregated, repository-level, and individual user-level report granularities. Shapes vary by scope and granularity.
• Team-level metrics are not pre-aggregated; you construct them by joining the user-teams report with the per-user usage metrics report. See Team-level Copilot usage metrics.
• Repository-level reports contain one record per repository with pull request activity on the requested day. See Repository-level fields (API only) later in this article.

For guidance on how to read and interpret these metrics, see GitHub Copilot usage metrics.

Copilot usage dashboard metrics

These metrics appear directly in the Copilot usage dashboard. These charts do not include Copilot CLI usage.

Metric	Description
Agent adoption	Percentage of Copilot-licensed active users who tried an agent in the current calendar month.
Average chat requests per active user	Average number of chat requests per active user.
Code completions (suggested/accepted)	Total number of inline code suggestions shown and accepted.
Code completion acceptance rate	Percentage of suggestions accepted by users.
Daily active users	Number of unique users who used Copilot on a given day.
Weekly active users	Number of unique users who used Copilot during a seven-day window.
Total active users	Licensed users active in the current calendar month.
Language usage	Distribution of programming languages used with Copilot.
Language usage per day	Daily breakdown of languages used.
Model usage	Distribution of AI models used for chat.
Model usage per day	Daily breakdown of chat model usage.
Model usage per chat mode	Model usage by Copilot feature (ask, edit, plan, agent).
Model usage per language	Distribution of languages broken down by model.
Most used chat model	The most frequently used chat model in the last 28 days.
Requests per chat mode	Number of chat requests by mode (ask, edit, plan, agent).

Impact dashboard metrics

These metrics appear in the impact dashboard, which groups users into adoption cohorts and connects that adoption to pull request output. For guidance on interpreting adoption cohorts, see GitHub Copilot usage metrics.

Metric	Description
Engagement trends	How the organization's adoption cohort mix and pull request throughput (total pull requests merged per month) have changed over the last six months.
Feature engagement	Number of active users who engaged with each included Copilot feature on at least two distinct days during the inclusive 28-day window.
Adoption cohort distribution	Share of licensed users grouped into each adoption phase (Passive users, Phase 1, Phase 2, Phase 3) for the period. Passive users correspond to the No Cohort value in the API.
Adoption multiplier	Compares engaged users (Phase 1, 2, or 3) against passive users on code shipped (pull requests merged per user per month) and time to merge pull requests, to show the relative impact of deeper Copilot adoption.
Recommendations	Suggested actions, such as configuring Copilot cloud agent or enabling Copilot code review, based on the organization's current cohort distribution.

Code generation dashboard metrics

These metrics appear in the code generation dashboard and provide a breakdown of how code is being generated across user-initiated and agent-initiated activity. All values are derived from lines of code (LoC) added or deleted in the IDE.

Metric	Description
Lines of code changed with AI	Total lines of code added and deleted across all modes in the last 28 days.
Agent contribution	Percentage of lines of code added and deleted by agents (including edit, agent, and custom modes) in the last 28 days.
Average lines deleted by agent	Average number of lines automatically deleted by agents on behalf of active users during the current calendar month.
Daily total of lines added and deleted	Total number of lines added to and deleted from the codebase across all modes for each day.
User-initiated code changes	Lines suggested or manually added by users through code completions and chat panel actions (insert, copy, or apply).
Agent-initiated code changes	Lines automatically added to or deleted from the codebase by agents on behalf of users across edit, agent, and custom modes.
User-initiated code changes per model	User-initiated lines of code, grouped by model used in the IDE.
Agent-initiated code changes per model	Agent-initiated lines of code, grouped by model performing the agent actions.
User-initiated code changes per language	User-initiated lines of code, grouped by programming language.
Agent-initiated code changes per language	Agent-initiated lines of code, grouped by programming language.

API and export fields

These fields appear in the exported NDJSON reports and in the Copilot usage metrics APIs. Most tables below list the field name, its JSON type, whether the value can be null, and a description.

Reports come in different shapes depending on their scope and granularity, so the fields available in a record depend on which report it comes from:

• Per-user reports (*-users-1-day and *-users-28-day) contain one record per user, including user_id, user_login, ai_credits_used, the used_* indicators, and ai_adoption_phase. They do not contain active-user counts, pull_requests, or totals_by_ai_adoption_phase.
• Aggregated reports (enterprise-1-day and organization-1-day) contain one aggregated record per enterprise or organization, including active-user counts, pull_requests, and totals_by_ai_adoption_phase. They do not contain user_id, user_login, or the used_* indicators.
• 28-day aggregate reports (enterprise-28-day and organization-28-day) wrap an array of daily aggregated records in a day_totals field, with the reporting window and optional copilot_feature_engagement object at the top level. The object is not included in per-user reports.
• User-teams reports (*-user-teams-1-day) map users to the teams they belong to, so you can construct team-level metrics.
• Repository-level reports (*-repos-1-day) contain one record per repository with pull request activity for the day, including pull requests created by Copilot cloud agent and reviewed by Copilot code review.

Organization-scope reports also include organization_id alongside enterprise_id. For example schemas of the data returned by the APIs, see Example schema for Copilot usage metrics.

The Type column uses JSON Schema types: string, integer, number, boolean, array, and object. The Nullable column indicates whether a field's value can be null or absent from a record where it would otherwise apply. Activity breakdown arrays are always present but can be empty ([]). Optional arrays are marked as nullable and can be absent.

Report identification fields

These fields identify the scope and date of each record. The exact set present depends on the report shape.

Field	Type	Nullable	Description
day	string	No	Calendar day this record represents, in YYYY-MM-DD format. In 28-day reports, day appears within each day_totals entry rather than at the top level.
enterprise_id	string	No	Unique ID of the enterprise. Included in both enterprise- and organization-scope reports.
organization_id	string	Yes	Unique ID of the organization. Included in organization-scope reports only; omitted from enterprise-scope reports.

Per-user report fields

Per-user reports contain one record per user for the reporting period. The 28-day per-user report also includes report_start_day and report_end_day to mark the reporting window.

Field	Type	Nullable	Description
user_id	integer	No	Unique identifier for the user.
user_login	string	No	GitHub username for the user.
ai_credits_used	number	No	Total AI credits consumed by the user in the reporting period. This field is included in per-user reports only and is not broken down by feature, model, or surface. This metric is for consumption analysis, not invoicing totals.
user_initiated_interaction_count	integer	No	Number of explicit prompts sent to Copilot. Only counts messages or prompts actively sent to the model. Does not include opening the chat panel, switching modes (for example, ask, edit, plan, or agent), using keyboard shortcuts to open the inline UI, or making configuration changes.
code_generation_activity_count	integer	No	Number of distinct Copilot output events generated. Includes: All generated content, including comments and docstrings. Multiple blocks: Each distinct code block from a single user prompt counts as a separate generation. Note: This metric is not directly comparable to user_initiated_interaction_count, since one prompt can produce multiple generations.
code_acceptance_activity_count	integer	No	Number of suggestions or code blocks accepted by users. Counts: All built-in accept actions, such as “apply to file,” “insert at cursor,” “insert into terminal,” and use of the Copy button. Does not count: Manual OS clipboard actions (for example, Ctrl+C). Granularity: Each acceptance action increments the count once, regardless of how many code blocks were generated by the initial prompt.
loc_suggested_to_add_sum	integer	No	Lines of code Copilot suggested to add (completions, inline chat, chat panel, and so on; excludes agent edits).
loc_suggested_to_delete_sum	integer	No	Lines of code Copilot suggested to delete (future support planned).
loc_added_sum	integer	No	Lines of code actually added to the editor (accepted completions, applied code blocks, agent and edit mode).
loc_deleted_sum	integer	No	Lines of code deleted from the editor (currently from agent edits).
used_agent	boolean	No	Whether the user used agent mode in the IDE that day. Does not include Copilot code review activity, which is captured separately in used_copilot_code_review_active and used_copilot_code_review_passive.
used_chat	boolean	No	Whether the user used IDE chat that day.
used_cli	boolean	No	Whether the user used Copilot CLI that day.
used_copilot_app	boolean	No	Whether the user used the Copilot app that day.
used_copilot_coding_agent	boolean	No	Whether the user used Copilot cloud agent (previously Copilot coding agent) that day.
used_copilot_cloud_agent	boolean	No	Whether the user used Copilot cloud agent that day. Carries the same value as used_copilot_coding_agent; both names are retained for backward compatibility.
used_copilot_code_review_active	boolean	Yes	Whether the user actively engaged with Copilot code review that day. A user is considered active if they manually requested a Copilot review, or applied a Copilot review suggestion. Null when there is no Copilot code review signal for the user that day.
used_copilot_code_review_passive	boolean	Yes	Whether the user had Copilot automatically assigned to review their pull request that day, without actively engaging with the review. Null when there is no Copilot code review signal for the user that day.
used_vscode_agent	boolean	Yes	Whether the user used the dedicated VS Code Agents window during the reporting period. Omitted when no dedicated Agents-window data is available for the user.
ai_adoption_phase	object	No	The user's AI adoption phase for the day. Always present; defaults to the "No Cohort" phase. See AI adoption phase fields.
distinct_skill_use_count	integer	Yes	Number of different skill identifiers with recorded activity for the user. See Copilot CLI customization fields.
distinct_custom_agent_use_count	integer	Yes	Number of different custom agent identifiers with recorded activity for the user. See Copilot CLI customization fields.
distinct_mcp_use_count	integer	Yes	Number of different Model Context Protocol (MCP) server identifiers with recorded connection activity for the user. See Copilot CLI customization fields.
distinct_slash_cmd_use_count	integer	Yes	Number of different slash command identifiers with recorded activity for the user. See Copilot CLI customization fields.
distinct_plugin_use_count	integer	Yes	Number of different plugin identifiers with recorded activity for the user. See Copilot CLI customization fields.
totals_by_skill	array	Yes	Top skills used in Copilot CLI, ordered by interaction count. See Copilot CLI customization fields.
totals_by_custom_agent	array	Yes	Top custom agents used in Copilot CLI, ordered by interaction count. See Copilot CLI customization fields.
totals_by_mcp	array	Yes	MCP servers with the most connection and reconnection attempts in Copilot CLI, ordered by interaction count. See Copilot CLI customization fields.
totals_by_slash_cmd	array	Yes	Top slash commands used in Copilot CLI, ordered by interaction count. See Copilot CLI customization fields.
totals_by_plugin	array	Yes	Top plugins used in Copilot CLI, ordered by interaction count. See Copilot CLI customization fields.
totals_by_cli	object	Yes	CLI-specific metrics for the user. Omitted when the user had no Copilot CLI usage that day. See Copilot CLI metrics fields.
totals_by_copilot_app	object	Yes	Copilot app metrics for the user. Omitted when the user had no Copilot app usage that day. See Copilot app metrics fields.
totals_by_3rd_party_agent	array	Yes	Per-agent usage metrics for recognized agent apps. Omitted when the user had no recognized agent app activity during the reporting period. See Agent apps metrics fields.
totals_by_vscode_agent	object	Yes	Dedicated VS Code Agents-window metrics for the user. Omitted when both required counts are not available. See VS Code Agents-window metrics fields.
totals_by_ide	array	No	Per-IDE breakdown of the user's activity. See Activity breakdown objects.
totals_by_feature	array	No	Per-feature breakdown of the user's activity. See Activity breakdown objects.
totals_by_language_feature	array	No	Breakdown combining language and feature dimensions. See Activity breakdown objects.
totals_by_language_model	array	No	Breakdown combining language and model dimensions, for chat activity. See Activity breakdown objects.
totals_by_model_feature	array	No	Breakdown combining model and feature dimensions, for chat activity. See Activity breakdown objects.

Aggregated enterprise and organization report fields

Aggregated reports contain one record per enterprise or organization, summarizing all users for the day. Active-user counts include users identified through both client-side and server-side telemetry. Users surfaced only through server-side telemetry are counted in active-user totals but may not appear in the dimensional breakdown arrays (totals_by_ide, totals_by_feature, and similar). The following tables list the active-user counts, then the activity totals and breakdowns.

Active-user counts:

Field	Type	Nullable	Description
daily_active_users	integer	No	Number of unique users who used Copilot on a given day.
weekly_active_users	integer	No	Number of unique users who used Copilot during a trailing seven-day window.
weekly_active_vscode_agent_users	integer	Yes	Number of unique users who used the dedicated VS Code Agents window from Sunday through the report day, inclusive. This calendar-to-date count is not a trailing seven-day window.
monthly_active_users	integer	No	Number of unique users who used Copilot during a trailing 28-day window.
monthly_active_vscode_agent_users	integer	Yes	Number of unique users who used the dedicated VS Code Agents window from the first day of the calendar month through the report day, inclusive. This calendar-to-date count is not a trailing 28-day window.
monthly_active_chat_users	integer	No	Number of unique users who used chat during a trailing 28-day window.
monthly_active_agent_users	integer	No	Number of unique users who used agent mode during a trailing 28-day window.
daily_active_copilot_cloud_agent_users	integer	No	Number of unique users who used Copilot cloud agent on a given day.
weekly_active_copilot_cloud_agent_users	integer	No	Number of unique users who used Copilot cloud agent during a trailing seven-day window.
monthly_active_copilot_cloud_agent_users	integer	No	Number of unique users who used Copilot cloud agent during a trailing 28-day window.
daily_active_copilot_code_review_users	integer	No	Number of unique users who actively used Copilot code review on a given day. Active usage means manually requesting a review or applying a suggestion. When a user has both active and passive signals in the same period, they are counted as active only.
weekly_active_copilot_code_review_users	integer	No	Number of unique users who actively used Copilot code review during a trailing seven-day window. When a user has both active and passive signals in the same period, they are counted as active only.
monthly_active_copilot_code_review_users	integer	No	Number of unique users who actively used Copilot code review during a trailing 28-day window. When a user has both active and passive signals in the same period, they are counted as active only.
daily_passive_copilot_code_review_users	integer	No	Number of unique users who had Copilot code review automatically assigned to review their pull request on a given day, with no active engagement.
weekly_passive_copilot_code_review_users	integer	No	Number of unique users who had Copilot code review automatically assigned to review their pull request during a trailing seven-day window, with no active engagement.
monthly_passive_copilot_code_review_users	integer	No	Number of unique users who had Copilot code review automatically assigned to review their pull request during a trailing 28-day window, with no active engagement.
daily_active_cli_users	integer	Yes	Number of unique users who used Copilot CLI on a given day. This count is independent of IDE active-user counts and is not included in IDE-based active-user definitions. Omitted for enterprises or organizations with no CLI usage that day.
daily_active_copilot_app_users	integer	Yes	Number of unique users who used the Copilot app on a given day. Null when the enterprise or organization has no Copilot app activity that day.
daily_active_vscode_agent_users	integer	Yes	Number of unique users who used the dedicated VS Code Agents window on a given day. Omitted when no dedicated Agents-window data is available for the enterprise or organization.

Activity totals and breakdowns:

Field	Type	Nullable	Description
user_initiated_interaction_count	integer	No	Total number of explicit prompts sent to Copilot across all users for the day.
code_generation_activity_count	integer	No	Total number of distinct Copilot output events generated across all users for the day.
code_acceptance_activity_count	integer	No	Total number of suggestions or code blocks accepted across all users for the day.
loc_suggested_to_add_sum	integer	No	Aggregated lines of code suggested to add for the day. Same definition as the per-user field.
loc_suggested_to_delete_sum	integer	No	Aggregated lines of code suggested to delete for the day. Same definition as the per-user field.
loc_added_sum	integer	No	Aggregated lines of code added for the day. Same definition as the per-user field.
loc_deleted_sum	integer	No	Aggregated lines of code deleted for the day. Same definition as the per-user field.
distinct_skill_use_count	integer	Yes	Number of different skill identifiers with recorded activity across the enterprise or organization. See Copilot CLI customization fields.
distinct_custom_agent_use_count	integer	Yes	Number of different custom agent identifiers with recorded activity across the enterprise or organization. See Copilot CLI customization fields.
distinct_mcp_use_count	integer	Yes	Number of different MCP server identifiers with recorded connection activity across the enterprise or organization. See Copilot CLI customization fields.
distinct_slash_cmd_use_count	integer	Yes	Number of different slash command identifiers with recorded activity across the enterprise or organization. See Copilot CLI customization fields.
distinct_plugin_use_count	integer	Yes	Number of different plugin identifiers with recorded activity across the enterprise or organization. See Copilot CLI customization fields.
totals_by_skill	array	Yes	Top skills used in Copilot CLI across the enterprise or organization. See Copilot CLI customization fields.
totals_by_custom_agent	array	Yes	Top custom agents used in Copilot CLI across the enterprise or organization. See Copilot CLI customization fields.
totals_by_mcp	array	Yes	MCP servers with the most connection and reconnection attempts in Copilot CLI across the enterprise or organization. See Copilot CLI customization fields.
totals_by_slash_cmd	array	Yes	Top slash commands used in Copilot CLI across the enterprise or organization. See Copilot CLI customization fields.
totals_by_plugin	array	Yes	Top plugins used in Copilot CLI across the enterprise or organization. See Copilot CLI customization fields.
totals_by_ide	array	No	Aggregated per-IDE activity breakdown. See Activity breakdown objects.
totals_by_feature	array	No	Aggregated per-feature activity breakdown. See Activity breakdown objects.
totals_by_language_feature	array	No	Aggregated language-and-feature activity breakdown. See Activity breakdown objects.
totals_by_language_model	array	No	Aggregated language-and-model activity breakdown. See Activity breakdown objects.
totals_by_model_feature	array	No	Aggregated model-and-feature activity breakdown. See Activity breakdown objects.
totals_by_cli	object	Yes	Aggregated Copilot CLI metrics for the day. Omitted when there is no CLI usage that day. Unlike the per-user form, it does not include last_known_cli_version. See Copilot CLI metrics fields.
totals_by_copilot_app	object	Yes	Aggregated Copilot app metrics for the day. Null when the enterprise or organization has no Copilot app activity that day. See Copilot app metrics fields.
totals_by_3rd_party_agent	array	Yes	Aggregated per-agent usage metrics for recognized agent apps. Omitted when the enterprise or organization had no recognized agent app activity that day. See Agent apps metrics fields.
totals_by_vscode_agent	object	Yes	Aggregated dedicated VS Code Agents-window metrics for the day. Omitted when both required counts are not available. See VS Code Agents-window metrics fields.
totals_by_ai_adoption_phase	array	Yes	Per-phase aggregates of users and their average activity. Omitted when no adoption-phase data is available. See AI adoption phase fields.
pull_requests	object	No	Daily pull request activity for the enterprise or organization. See Pull request activity fields.

28-day report fields

The 28-day aggregate reports (enterprise-28-day and organization-28-day) are wrappers: they carry the reporting window at the top level and an array of daily aggregated records.

Field	Type	Nullable	Description
report_start_day	string	No	First calendar day of the 28-day reporting window, in YYYY-MM-DD format.
report_end_day	string	No	Last calendar day of the 28-day reporting window, in YYYY-MM-DD format.
created_at	string	No	Timestamp (ISO 8601) when the report was generated.
copilot_feature_engagement	object	Yes	Rolling feature engagement for the 28-day window. Included only in enterprise and organization aggregate reports. Null or absent when the optional calculation is unavailable or the report predates this field. See Copilot feature engagement fields.
day_totals	array	No	Array of daily aggregated records. Each entry has the same fields as an aggregated 1-day report. See Aggregated enterprise and organization report fields.

Copilot feature engagement fields

The optional copilot_feature_engagement object summarizes the report's rolling active-user population and engagement with seven features. For the day_totals entry whose day matches report_end_day, copilot_feature_engagement.active_user_count equals the sum of the mutually exclusive totals_by_ai_adoption_phase[].users_in_phase_28d values.

Counts for individual features overlap because a user can engage with more than one feature. Do not add the engaged_user_count values together. The initial feature set reports active and passive Copilot code review as separate features. It does not include Copilot Chat or VS Code Agent.

Copilot Chat is planned for a later release after 28 consecutive days of data have been collected and report consumers support the additional entry.

Field	Type	Nullable	Description
copilot_feature_engagement.active_user_count	integer	No	Nonnegative number of distinct active users in the inclusive 28-day report window. A valid zero-user report sets this field to 0.
copilot_feature_engagement.totals_by_feature	array	No	Feature engagement counts. The initial array contains exactly one entry for each of these values: code_completion, agent_edit, code_review_passive, code_review_active, cloud_agent, copilot_cli, and github_app. A valid zero-user report includes all seven entries with engaged_user_count set to 0.
copilot_feature_engagement.totals_by_feature[].feature	string	No	Feature identifier. In this object, github_app represents the Copilot app. This differs from the copilot_app value used in activity breakdown arrays such as totals_by_feature.
copilot_feature_engagement.totals_by_feature[].engaged_user_count	integer	No	Nonnegative number of distinct active users who engaged with the feature on at least two distinct days during the inclusive 28-day report window.

User-teams fields

These fields appear in the daily user-teams report (available via REST API at the organization and enterprise scopes) and are used to construct team-level metrics by joining with the per-user usage metrics report. For the full join recipe and the endpoint URLs, see Team-level Copilot usage metrics.

Teams with fewer than 5 seated Copilot users on a given day are excluded from the user-teams report.

Field	Type	Nullable	Description
user_id	integer	No	Unique identifier for the user.
user_login	string	No	GitHub username for the user.
day	string	No	Calendar day this record represents.
organization_id	string	No	Unique ID of the organization the team belongs to. Organization scope only.
enterprise_id	string	No	Unique ID of the enterprise the team belongs to. Enterprise scope only. The enterprise-scoped report includes both enterprise teams and business teams.
team_id	integer	No	Unique ID of the team the user belongs to.
slug	string	No	URL-friendly identifier for the team.

Repository-level fields (API only)

Repository-level reports show daily pull request creation, review, merge, and suggestion activity, plus pull request review stage durations, for repositories in an organization or enterprise, including pull requests created by Copilot cloud agent and reviewed by Copilot code review. Each row represents one repository that had pull request activity on the requested day; repositories with no activity are omitted. These reports focus on pull request lifecycle activity and can contain data even when IDE usage metrics are absent.

These fields are returned by the daily repository-level report (repos-1-day), available via the REST API at the organization and enterprise scopes. For the endpoint URLs, response envelope, permissions, and download workflow, see REST API endpoints for Copilot usage metrics. For example rows, see Example schema for Copilot usage metrics.

Field	Type	Nullable	Description
day	string	No	Calendar day this record represents, in YYYY-MM-DD format.
enterprise_id	string	No	Unique ID of the enterprise. Populated in enterprise-scoped reports, and in organization-scoped reports for organizations owned by an enterprise. Empty for organizations not owned by an enterprise.
organization_id	string	No	Unique ID of the organization that owns the repository. Included in both enterprise- and organization-scoped reports.
repo_id	integer	No	Unique ID of the repository.
repo_owner_name	string	No	Login of the repository owner.
repo_name	string	No	Repository name.
repo_visibility	string	No	Repository visibility. Possible values are PRIVATE, INTERNAL, and PUBLIC.
pull_requests	object	No	Daily pull request activity for the repository. See Pull request activity fields.
pull_request_review_times	array	No	Review stage durations for qualifying pull requests merged in the repository on this day, with one entry per author and reviewer type. Empty when no qualifying pull requests were merged. See Pull request review time fields.

Activity breakdown objects

The totals_by_* fields are arrays of breakdown objects. The array is always present but can be empty. Within each object, the metric fields (*_count and loc_*_sum) follow the same definitions as the top-level per-user fields, and the dimension fields carry the values documented in Breakdown dimension values.

Object	Dimension fields	Description
totals_by_ide[]	ide	Breakdown by IDE. In per-user reports, each entry also includes last_known_ide_version and last_known_plugin_version.
totals_by_feature[]	feature	Breakdown by Copilot feature (for example, inline chat or chat panel).
totals_by_language_feature[]	language, feature	Breakdown combining language and feature. Does not include user_initiated_interaction_count.
totals_by_language_model[]	language, model	Breakdown combining language and model, for chat activity (not completions).
totals_by_model_feature[]	model, feature	Breakdown combining model and feature, for chat activity (not completions).

In per-user reports, each totals_by_ide[] entry also reports the most recently detected IDE and Copilot Chat extension versions for the user.

Field	Type	Nullable	Description
last_known_ide_version	object	Yes	Most recent IDE version detected for the user, as { ide_version, sampled_at }. Omitted for aggregated breakdown rows, such as the capped "others" IDE bucket.
last_known_ide_version.ide_version	string	No	IDE version string. Present when last_known_ide_version is present.
last_known_ide_version.sampled_at	string	Yes	Timestamp (ISO 8601) when the version was sampled.
last_known_plugin_version	object	Yes	Most recent Copilot extension detected for the user, as { plugin, plugin_version, sampled_at }. Omitted for aggregated breakdown rows.
last_known_plugin_version.plugin	string	No	Extension name (for example, copilot or copilot-chat). Present when last_known_plugin_version is present.
last_known_plugin_version.plugin_version	string	No	Extension version string. Present when last_known_plugin_version is present.
last_known_plugin_version.sampled_at	string	Yes	Timestamp (ISO 8601) when the version was sampled.

Copilot CLI metrics fields

The totals_by_cli object contains the following nested fields when Copilot CLI usage is present; the object is omitted when there is no CLI usage. CLI usage is independent of IDE metrics—it is not reflected in fields such as totals_by_ide or totals_by_feature. The last_known_cli_version object appears in per-user reports only.

Field	Type	Nullable	Description
totals_by_cli.session_count	integer	No	Number of distinct CLI sessions initiated on this day.
totals_by_cli.request_count	integer	No	Total number of requests made to Copilot via the CLI on this day, including both user-initiated prompts and automated agentic follow-up calls.
totals_by_cli.prompt_count	integer	No	Total number of user prompts, commands, or queries executed within a session.
totals_by_cli.token_usage.output_tokens_sum	integer	No	Total number of output tokens generated across all CLI requests on this day.
totals_by_cli.token_usage.prompt_tokens_sum	integer	No	Total number of prompt tokens sent across all CLI requests on this day.
totals_by_cli.token_usage.avg_tokens_per_request	number	Yes	Average number of output and prompt tokens per CLI request, computed as (output_tokens_sum + prompt_tokens_sum) ÷ request_count. Null when there were no requests that day.
totals_by_cli.last_known_cli_version	object	No	Most recent Copilot CLI version detected for the user that day, as { cli_version, sampled_at }. Per-user reports only.
totals_by_cli.last_known_cli_version.cli_version	string	No	Copilot CLI version string. Defaults to unknown if no version was detected.
totals_by_cli.last_known_cli_version.sampled_at	string	Yes	Timestamp (ISO 8601) when the version was sampled.

Copilot CLI customization fields (API only)

These metrics give enterprise and organization administrators visibility into which Copilot CLI automations developers use. Use them to understand adoption, identify enablement gaps, and prioritize the skills, Model Context Protocol (MCP) servers, custom agents, slash commands, and plugins that developers find valuable.

To access the reports, the Copilot usage metrics policy must be enabled. For policy configuration instructions, see:

• Managing policies and features for GitHub Copilot in your enterprise
• Managing policies and features for GitHub Copilot in your organization

These fields appear in the following enterprise and organization reports:

• Per-user 1-day and 28-day reports
• Aggregated 1-day reports
• Each day_totals record in aggregated 28-day reports

Each totals_by_* array contains up to five entries with the most recorded activity for its record, ordered from highest to lowest interaction_count. The values are event counts, not user counts. An entry's absence from an array does not mean that it had zero usage. Do not use these arrays to calculate exact adoption.

Recognized first-party names can appear when available. Customer-defined identifiers are not exposed. Skills, custom agents, MCP servers, and plugins identified only by customer-specific hashes are grouped under other in the arrays. Copilot CLI telemetry reports customer-defined slash commands as custom, so slash command arrays preserve that value.

Each array entry contains the name field listed in the table and an integer interaction_count field.

Array	Name field	What interaction_count measures	Distinct-count field
totals_by_skill[]	skill	Skill invocations.	distinct_skill_use_count
totals_by_custom_agent[]	custom_agent	Custom agent starts.	distinct_custom_agent_use_count
totals_by_mcp[]	mcp	Successful or failed MCP server connection and reconnection attempts. Tool calls through an already connected server do not increase this count.	distinct_mcp_use_count
totals_by_slash_cmd[]	slash_cmd	Slash command invocations.	distinct_slash_cmd_use_count
totals_by_plugin[]	plugin	Skill invocations associated with a plugin. Plugin interactions are a subset of skill interactions.	distinct_plugin_use_count

Every plugin interaction is already included in totals_by_skill, while skill interactions that are not associated with a plugin appear only in skill totals. Do not add plugin and skill interaction counts together.

The distinct-count fields count different item identifiers with activity, not users or uses. In a per-user record, each different identifier that the user used counts once. In an aggregated enterprise or organization record, each different identifier used by anyone in the scope counts once, regardless of how many users used it. Aggregated distinct counts are not sums of per-user distinct counts.

Distinct counts retain the full identifier cardinality, including items outside the top-five array. When customer-defined artifacts are grouped under other, each hidden identifier still contributes separately to the corresponding distinct count.

When these fields are present, empty arrays and distinct counts of 0 mean that no matching activity was recorded. The fields can be null or absent when Copilot CLI customization data isn't available during rollout.

Copilot app metrics fields

The totals_by_copilot_app object contains the following nested fields when Copilot app usage is present. It appears in per-user reports and in aggregated enterprise reports; it is omitted or null when the user or enterprise had no Copilot app activity that day. Unlike totals_by_cli, Copilot app coding activity is also reflected in totals_by_feature, totals_by_model_feature, totals_by_language_feature, and totals_by_language_model under the copilot_app feature value, and in the top-level code activity and lines-of-code metrics. There is no last_known_app_version field.

Field	Type	Nullable	Description
totals_by_copilot_app.session_count	integer	No	Number of distinct Copilot app sessions initiated on this day.
totals_by_copilot_app.request_count	integer	No	Total number of requests made to Copilot via the Copilot app on this day, including both user-initiated prompts and automated agentic follow-up calls.
totals_by_copilot_app.prompt_count	integer	No	Total number of user prompts, commands, or queries executed within a session.
totals_by_copilot_app.token_usage.output_tokens_sum	integer	No	Total output tokens generated across all Copilot app requests on this day.
totals_by_copilot_app.token_usage.prompt_tokens_sum	integer	No	Total prompt tokens sent across all Copilot app requests on this day.
totals_by_copilot_app.token_usage.avg_tokens_per_request	number	Yes	Average of output and prompt tokens per Copilot app request, computed as (output_tokens_sum + prompt_tokens_sum) ÷ request_count. Null when there were no requests that day.

VS Code Agents-window metrics fields

The dedicated VS Code Agents-window fields appear in enterprise, organization, enterprise-user, and organization-user 1-day and 28-day reports. In aggregated 28-day reports, they appear within each applicable day_totals entry, and weekly and monthly counts are calculated relative to the day of each entry. In per-user 28-day reports, they appear at the top level of each user record.

These fields are distinct from editor-window agent mode, including used_agent, monthly_active_agent_users, and the chat_panel_agent_mode and agent_edit feature values. Dedicated Agents-window activity is not included in generic top-level activity totals or dimensional breakdowns such as totals_by_feature, totals_by_model_feature, or totals_by_ai_adoption_phase.

The daily count and per-user fields can be omitted when no dedicated Agents-window data is available. The weekly and monthly fields are both omitted when there is no qualifying activity in either period. If one period has qualifying activity and the other does not, the inactive period has a value of 0. The totals_by_vscode_agent object appears only when both nested counts are available, including when both counts are zero.

Field	Type	Nullable	Description
daily_active_vscode_agent_users	integer	Yes	Number of unique users who used the dedicated VS Code Agents window on the day. Aggregated enterprise and organization reports only.
weekly_active_vscode_agent_users	integer	Yes	Number of unique users who used the dedicated Agents window from Sunday through the report day, inclusive. Aggregated enterprise and organization reports only.
monthly_active_vscode_agent_users	integer	Yes	Number of unique users who used the dedicated Agents window from the first day of the calendar month through the report day, inclusive. Aggregated enterprise and organization reports only.
used_vscode_agent	boolean	Yes	Whether the user used the dedicated VS Code Agents window during the reporting period. Enterprise-user and organization-user reports only.
totals_by_vscode_agent	object	Yes	Session and user-message totals for the dedicated VS Code Agents window.
totals_by_vscode_agent.session_count	integer	No	Number of distinct Agents-window sessions represented in the record. Required when totals_by_vscode_agent is present.
totals_by_vscode_agent.total_user_messages	integer	No	Number of user messages sent in the Agents window represented in the record. Required when totals_by_vscode_agent is present.

Agent apps metrics fields

The totals_by_3rd_party_agent array contains one entry for each recognized agent app used during the reporting period. Entries are grouped by agent_id, and integrations that map to the same agent are combined. Use agent_id as the stable grouping key and agent_name for display. The array is omitted when no recognized agent app activity is available during the reporting period.

These metrics come from server-side job activity. The nested user_initiated_interaction_count counts agent app job starts and is distinct from the top-level field with the same name, which counts explicit prompts from other supported telemetry.

Field	Type	Nullable	Description
totals_by_3rd_party_agent[].agent_name	string	No	Display name of the agent app. The name can change, so use agent_id for grouping.
totals_by_3rd_party_agent[].agent_id	string	No	Stable identifier for the agent app.
totals_by_3rd_party_agent[].user_initiated_interaction_count	integer	No	Number of user-initiated jobs started for the agent app during the reporting period. Each job start increments the count once.
totals_by_3rd_party_agent[].session_count	integer	No	Sum of distinct, non-empty session counts for integrations mapped to the agent during the reporting period. Included only in aggregated enterprise and organization reports; omitted from per-user reports.

Pull request activity fields

The pull_requests object appears in aggregated enterprise and organization reports, and in repository-level reports. It captures daily pull request creation, review, merge, and suggestion activity, including activity performed by Copilot cloud agent and Copilot code review.

Field	Type	Nullable	Description
pull_requests.total_created	integer	No	Total number of pull requests created on this specific day. Creation is a one-time event. Each pull request is counted only on the day it is created.
pull_requests.total_reviewed	integer	No	Total number of pull requests reviewed on this specific day. The same pull request may be counted on multiple days if it receives reviews on multiple days. Within a single day, each pull request is counted once, even if multiple review actions occur.
pull_requests.total_merged	integer	No	Total number of pull requests merged on this specific day. Merging is a one-time event. Each pull request is counted only on the day it is merged.
pull_requests.median_minutes_to_merge	number	Yes	Median time, in minutes, between pull request creation and merge for pull requests merged on this specific day. Median is used to reduce the impact of outliers from unusually long-running pull requests. Null when no pull requests were merged that day.
pull_requests.total_suggestions	integer	No	Total number of pull request review suggestions generated on this specific day, regardless of author.
pull_requests.total_applied_suggestions	integer	No	Total number of pull request review suggestions that were applied on this specific day, regardless of author.
pull_requests.total_created_by_copilot	integer	No	Number of pull requests created by Copilot cloud agent on this specific day.
pull_requests.total_reviewed_by_copilot	integer	No	Number of pull requests reviewed by Copilot code review on this specific day. A pull request may be counted on multiple days if Copilot code review reviews it on multiple days.
pull_requests.total_merged_created_by_copilot	integer	No	Number of pull requests created by Copilot cloud agent that were merged on this specific day. Each pull request is counted only on the day it is merged.
pull_requests.total_merged_reviewed_by_copilot	integer	No	Number of pull requests that were both merged and reviewed by Copilot code review during the reporting period.
pull_requests.median_minutes_to_merge_copilot_authored	number	Yes	Median time, in minutes, between pull request creation and merge for pull requests created by Copilot cloud agent and merged on this specific day. Null when no such pull requests were merged that day.
pull_requests.median_minutes_to_merge_copilot_reviewed	number	Yes	Median time, in minutes, between pull request creation and merge, calculated only for pull requests reviewed by Copilot code review. Null when no such pull requests were merged that day.
pull_requests.total_copilot_suggestions	integer	No	Number of pull request review suggestions generated by Copilot code review on this specific day.
pull_requests.total_copilot_applied_suggestions	integer	No	Number of pull request review suggestions generated by Copilot code review that were applied on this specific day.
pull_requests.copilot_suggestions_by_comment_type	array	No	Aggregated counts of Copilot code review suggestions, broken down by the assigned comment type (for example, security or bug_risk). Each entry includes comment_type, total_copilot_suggestions, and total_copilot_applied_suggestions. Always present but can be empty.

Pull request review time fields

The pull_request_review_times array appears only in repository-level reports. It breaks down how long merged pull requests spent in three stages of review, so you can see whether pull requests are waiting for a first review, waiting on back-and-forth between reviewers, or waiting to be merged after the final review.

Each entry covers pull requests merged in the repository on this specific day. All durations are attributed to the merge day, even when the reviews happened on earlier days.

Which pull requests are counted

A pull request is included when a person opened it, it reached ready for review, and at least one other person reviewed it. Only human reviews are timed. Reviews from Copilot code review, from other bots, and from the pull request author are ignored, so a pull request reviewed by both a person and Copilot code review is still included, with its stages timed from the human reviews. Time spent in draft before the first review is not counted.

A merged pull request is excluded from pull_request_review_times when any of the following is true:

• It was created by a bot, including Copilot cloud agent.
• It never received a qualifying human review, including pull requests merged without any review.
• The time it became ready for review is unknown.
• Its timestamps are out of order. For example, its final review is recorded after its merge.

Exclusions are not reported for individual pull requests. Because pull_requests.total_merged counts every pull request merged that day, pull_request_review_times[].total_merged is usually lower. Compare the two values to see how many merged pull requests the durations cover.

Data availability

• Data builds forward from the release of these fields and is not backfilled.
• Pull requests that became ready for review before September 21, 2026 are excluded from this array, but still count toward pull_requests.total_merged.
• The array is empty ([]) on days when the repository merged no qualifying pull requests. An empty array means no data for that day, not zero minutes.

Field	Type	Nullable	Description
pull_request_review_times[].authored_by	string	No	Who opened the pull requests in this entry. Currently always human.
pull_request_review_times[].reviewed_by	string	No	Who reviewed the pull requests in this entry. Currently always human.
pull_request_review_times[].total_merged	integer	No	Number of qualifying pull requests merged in the repository on this specific day. This counts only the pull requests described above, not all merged pull requests.
pull_request_review_times[].median_minutes_ready_to_first_review	number	No	Median time, in minutes, from the pull request becoming ready for review to its first qualifying review.
pull_request_review_times[].p90_minutes_ready_to_first_review	number	No	90th percentile time, in minutes, from the pull request becoming ready for review to its first qualifying review.
pull_request_review_times[].median_minutes_first_to_final_review	number	No	Median time, in minutes, from the first qualifying review to the final qualifying review. A pull request with a single qualifying review contributes a duration of 0, because its first and final reviews are the same event.
pull_request_review_times[].p90_minutes_first_to_final_review	number	No	90th percentile time, in minutes, from the first qualifying review to the final qualifying review. A pull request with a single qualifying review contributes a duration of 0, because its first and final reviews are the same event.
pull_request_review_times[].median_minutes_final_review_to_merge	number	No	Median time, in minutes, from the final qualifying review to merge.
pull_request_review_times[].p90_minutes_final_review_to_merge	number	No	90th percentile time, in minutes, from the final qualifying review to merge.

Medians and 90th percentiles are exact values calculated with linear interpolation and rounded to two decimal places.

AI adoption phase fields

Copilot groups users into AI adoption phases based on their activity. Phase information appears in two places: the per-user ai_adoption_phase object, and the aggregated totals_by_ai_adoption_phase array. For the phase values, see Breakdown dimension values.

Classification logic

A user is classified into a phase based on the Copilot features they are engaged with. A user is engaged with a feature when they use it on at least two distinct days within the trailing 28-day window. Engagement is evaluated from the feature-level activity signals in the table below, not directly from the single-day used_* per-user fields or from code_acceptance_activity_count. Phases are evaluated from the top down, and a user is assigned the highest phase they qualify for.

Phase	Engagement criteria
No Cohort (displayed as "Passive users" in the impact dashboard)	The user isn't engaged with any feature that qualifies for a phase. This doesn't mean the user is inactive. A user in this group can still be using Copilot regularly and consuming premium requests, but hasn't used a qualifying feature on at least two distinct days within the trailing 28-day window. The most common case is conversational use of Copilot Chat or agent mode, recorded as chat_panel_agent_mode, that never produces code_completion or agent_edit activity. Copying code out of a chat response also doesn't produce either signal.
Phase 1: Code first	Engaged with inline code completions (the code_completion value of the feature dimension) or with agent edits written directly to files (the agent_edit value of the feature dimension).
Phase 2: Agent first	Engaged with exactly one GitHub-based agent surface: Copilot CLI (the copilot_cli value of the feature dimension), Copilot cloud agent (used_copilot_cloud_agent), or Copilot code review (used_copilot_code_review_active or used_copilot_code_review_passive). Active and passive Copilot code review count together as a single surface.
Phase 3: Multi-agent	Engaged with the Copilot app (the copilot_app value of the feature dimension), or engaged with two or more of the agent surfaces listed for Phase 2.

A user only needs to be engaged with a phase's own signals to reach that phase. A user doesn't need to independently meet Phase 1 criteria to reach Phase 2 or Phase 3, although in practice agent-surface usage often co-occurs with completions or agent edits.

Of the agent surfaces, only the first-party surfaces listed in the table affect a user's AI adoption phase. Activity from partner-built agent apps doesn't affect the phase. The same classification logic applies to enterprise and organization reports, including 1-day and 28-day reports.

Phase assignment is recalculated each day using the trailing 28-day window, so a user's phase can change from one day to the next as their activity within the window shifts. This is expected behavior and does not indicate a data error.

The per-user ai_adoption_phase object contains:

Field	Type	Nullable	Description
ai_adoption_phase.phase_number	integer	No	Numeric phase identifier (for example, 0 for "No Cohort").
ai_adoption_phase.phase	string	No	Human-readable phase name.
ai_adoption_phase.version	string	No	Version of the adoption-phase model used (for example, v1).

Each entry in the aggregated totals_by_ai_adoption_phase array contains two separate user counts:

• total_engaged_users is the number of users classified into the phase who were active on that individual day.
• users_in_phase_28d is the full population classified into the phase using the rolling 28-day window, as of that day.

Use users_in_phase_28d as the population denominator for rolling phase-level calculations. For example, to calculate pull requests merged per user over a 28-day period:

1. Add the daily total_pull_requests_merged values.
2. Add the corresponding daily users_in_phase_28d values.
3. Divide the first total by the second, then multiply the result by 28.

Do not use total_engaged_users as the denominator because it includes only users active on an individual day.

The field does not change the existing avg_* fields or their calculations, the phase classification rules, or a user's assigned phase. Neither user-count field identifies individuals.

When a phase is present in the rolling snapshot but has no activity that day, the report can include a synthesized entry with total_engaged_users set to 0, a positive users_in_phase_28d, and all activity metrics set to zero.

Field	Type	Nullable	Description
phase	string	No	Human-readable phase name.
phase_number	integer	No	Numeric phase identifier.
total_engaged_users	integer	No	Number of users classified into this phase who were active on the day.
users_in_phase_28d	integer	Yes	Complete rolling 28-day population classified into this phase as of the day. A positive value is the measured phase population, and 0 means the phase was measured and had no users. The field is omitted when the day predates this field or the enterprise or organization was absent from the phase snapshot.
avg_user_initiated_interactions	number	No	Average user-initiated interactions per user in this phase.
avg_code_generation_activities	number	No	Average code generation activities per user in this phase.
avg_code_acceptance_activities	number	No	Average code acceptance activities per user in this phase.
avg_loc_added	number	No	Average lines of code added per user in this phase.
avg_loc_deleted	number	No	Average lines of code deleted per user in this phase.
avg_pull_requests_reviewed	number	No	Average pull requests reviewed per user in this phase.
avg_pull_requests_created	number	No	Average pull requests created per user in this phase.
avg_pull_requests_merged	number	No	Average pull requests merged per user in this phase.
total_pull_requests_merged	integer	No	Total pull requests merged by users in this phase. Applies to the organization day and enterprise day aggregated reports.
avg_pull_requests_median_minutes_to_merge	number	No	Average of the per-user median minutes to merge for users in this phase.
avg_pull_requests_minutes_to_review	number	No	Median time in minutes from pull request creation to first review, scoped to merged pull requests and attributed to the pull request's merge day.
avg_pull_requests_review_cycles	number	No	Median number of review submissions (review cycles) per pull request, scoped to merged pull requests and attributed to the pull request's merge day.

Breakdown dimension values

The breakdown objects above use dimension fields whose values come from fixed sets. These are field values, not separate fields. For example, agent_edit and the chat_panel_*_mode values are values of the feature field, not top-level fields.

The feature dimension identifies the Copilot feature or mode an activity is attributed to:

Value	Description
code_completion	Inline code completions.
chat_inline	Inline chat in the editor.
chat_panel_ask_mode	Chat panel interactions with ask mode selected.
chat_panel_edit_mode	Chat panel interactions with edit mode selected.
chat_panel_agent_mode	Chat panel interactions with agent mode selected.
chat_panel_plan_mode	Chat panel interactions with plan mode selected.
chat_panel_custom_mode	Chat panel interactions with a custom agent selected.
chat_panel_unknown_mode	Chat panel interactions where the mode is unknown.
agent_edit	Lines added and deleted when Copilot (in agent and edit mode) writes changes directly into your files in the IDE. Counts edits from custom agents as well. agent_edit is not included in suggestion-based metrics and may not populate suggestion-style fields, such as user_initiated_interaction_count.
copilot_cli	Activity attributed to Copilot CLI.
copilot_app	Activity attributed to the Copilot app.
others	Any feature not covered by the values above.

The ide dimension identifies the IDE an activity occurred in. This is not an exhaustive list, but examples of observed values include vscode, visualstudio, intellij, eclipse, xcode, neovim, vim, emacs, and zed.

The model dimension identifies the AI model used for chat activity. Values include specific model identifiers (for example, gpt-5.4 or claude-sonnet-4.6), auto, unknown, and others. The auto value represents activity where auto model selection was used and the request was not attributed to a specific model.

The phase dimension identifies the AI adoption phase a user is grouped into: No Cohort, Phase 1, Phase 2, or Phase 3.