Export AI cost metrics from the Consumption Dashboard (preview)

You can download user-level credit usage (AI cost) metrics using the Consumption Dashboard's data export feature. Personal identifiers are available with hashed IDs, unless your administrator has enabled identifiable export for you. If your administrator has enabled identifiable export, the export also includes identifiable fields.

Exports are available at the daily and weekly level, to help you build custom cost reports, reviews, and analyses outside the dashboard.

For example, you can:

• Identify top-consuming users, services, and spending policies to see where credits are being spent and who is driving usage.

• Keep a close grasp on spend and who might be close to hitting their policy or user-level limit through deeper analysis - sorting, filtering, and joining credit data with your own business context.

• Identify low-usage employees so admins can target enablement, training, or nudges.

• Support recurring spend reviews and executive summaries for analysts who need person- and policy-level data.

You can also use Power BI's Get Data tool to pull the data directly. For multi-table output queries, you must provide the Table Name parameter along with the Partition ID and Query ID. Learn more.

Prerequisites

To use the export data feature, you must have access to the Consumption Dashboard.

Data export is currently only available to the following roles:

• Analysts with global partition access.
• Microsoft 365 Global Administrators.

The feature is on by default for eligible tenants and users. Data export for Managers will be available by the end of July.

How to export data

1. At the top right of the Consumption Dashboard, select the download button.

   

2. Select either Export by week or Export by day. Use the guidance below if you're not sure. The download starts automatically in your browser.

   Option	Scenario
   Export by week	Best for tracking usage trends over weeks or months for long-term planning. This export includes complete data for the previous five months and month-to-date data for the current month. Data usually reflects activity from 2 - 8 days before the export date, depending on weekly processing cycles.
   Export by day	Best for tracking day-to-day costs and detecting emerging usage trends. This export includes data from the start of the previous month through the export date. Data usually reflects activity up to two days before the export date, but timing might vary.

Key details to consider

• Only complete days or weeks are included. Unlike data in the dashboard, partial-period data is never exported. If a day's or week's data is incomplete, that period is omitted. Headers are localized as a standard treatment in the output.

• The export downloads as a .zip file named based on the export type, date, and time. For example, AICostDashboard-Daily_Jun23_2026_1845Hours.zip. (The weekly export file follows the same pattern with Weekly instead of Daily).

What's in the export

Each .zip file contains two .csv files, which join on PeopleHistoricalId:

• PersonServiceCreditsMetrics.csv - credit consumption metrics
• PeopleMetaData.csv - people metadata

File 1 - Credit consumption metrics (PersonServiceCreditsMetrics.csv)

One row per person, service, and date. Includes both licensed and unlicensed users.

File 2 - People metadata (PeopleMetaData.csv)

One row per person. Contains the join key plus the Organization HR attribute and Job function if available.

Related topics

• How to use the Consumption Dashboard
• How to run a custom consumption query