Microsoft 365 Copilot has transformed how organizations approach AI-powered productivity, but measuring its impact remains a challenge for data teams. With Microsoft's new Interactions Export API and the forthcoming Graph 'Copilot Usage' endpoints, organizations can now build comprehensive adoption dashboards that track everything from active users to productivity impact metrics. (Microsoft 365 Copilot APIs Overview)
This hands-on guide walks you through creating a real-time AI usage dashboard using Power BI, combining Microsoft's preview APIs with Worklytics' privacy-safe data processing capabilities. (Microsoft Copilot Sanitized) We'll cover authentication, data mapping, visualization, and security best practices so you can replicate this build in a single weekend.
The Microsoft 365 Copilot APIs provide secure and compliant access to AI capabilities for custom applications and engine agents, eliminating the need for developers to build complex AI systems from scratch or extract sensitive data into external sources. (Microsoft 365 Copilot APIs Overview) By leveraging these APIs alongside Power BI's visualization capabilities, data teams can create comprehensive dashboards that provide actionable insights into AI adoption patterns.
Microsoft's Copilot ecosystem provides several key API endpoints for extracting usage data. The primary endpoints include AIInteraction, AIInteractionAttachment, AIInteractionContext, AIInteractionLink, AIInteractionMention, AIInteractionMentionedIdentitySet, ItemBody, and TeamWorkConversationIdentity. (Microsoft Copilot Sanitized)
These endpoints capture different aspects of Copilot usage:
The data source for Copilot data is represented by the 'appClass' string, which helps categorize different types of AI interactions across the Microsoft 365 ecosystem. (Microsoft Copilot Sanitized)
Microsoft 365 Copilot supports four authentication schemes for API plugins: OAuth 2.0 authorization code flow, Microsoft Entra ID single-sign on (SSO) authentication, API key authentication, and anonymous access. (Configure Authentication for API plugins in Agents in Microsoft 365 Copilot)
For production dashboards, OAuth 2.0 with Proof Key for Code Exchange (PKCE) provides the most secure authentication method. Before starting, you'll need to register with your OAuth 2.0 provider to get a client ID and secret. (Configure Authentication for API plugins in Agents in Microsoft 365 Copilot)
Before building your dashboard, ensure you have:
Worklytics provides access to a Data Loss Prevention (DLP) Proxy that offers full field-level control over sensitive data. (Microsoft Copilot Sanitized) This proxy transforms and pseudonymizes certain fields to ensure compliance with GDPR, CCPA, and other data protection standards.
The registration process requires careful attention to scope permissions, as overly broad access can create security vulnerabilities while insufficient permissions will limit dashboard functionality.
The first step involves establishing secure connections to Microsoft's preview APIs. Here's a sample Power Query M code for connecting to the Interactions Export API:
let
Source = Web.Contents(
"https://graph.microsoft.com/beta/me/insights/copilot/interactions",
[
Headers = [
#"Authorization" = "Bearer " & AccessToken,
#"Content-Type" = "application/json"
]
]
),
JsonResponse = Json.Document(Source),
InteractionsList = JsonResponse[value],
ExpandedInteractions = Table.ExpandRecordColumn(
Table.FromList(InteractionsList, Splitter.SplitByNothing()),
"Column1",
{"id", "timestamp", "appClass", "userPrincipalName", "interactionType"}
)
in
ExpandedInteractions
This code establishes the foundation for extracting interaction data while maintaining proper authentication headers and error handling.
Worklytics' DLP Proxy provides essential privacy safeguards for sensitive AI interaction data. The proxy transforms and pseudonymizes certain fields automatically, ensuring that personal information is protected while maintaining analytical value. (Microsoft Copilot Sanitized)
Key privacy transformations include:
Mapping Microsoft's API responses to a consistent analytical schema requires careful field alignment. Here's a sample DAX measure for calculating active Copilot users:
Active Copilot Users =
CALCULATE(
DISTINCTCOUNT(CopilotInteractions[PseudonymizedUserId]),
CopilotInteractions[InteractionDate] >= TODAY() - 30
)
This measure leverages Worklytics' pseudonymized user identifiers to count unique active users while maintaining privacy compliance.
Tracking user adoption requires multiple dimensions of analysis. Power BI implementation planning emphasizes the importance of adoption tracking for understanding user engagement patterns. (Power BI implementation planning: Adoption tracking)
Key adoption metrics include:
Here's a DAX measure for calculating feature adoption rates:
Feature Adoption Rate =
DIVIDE(
CALCULATE(
DISTINCTCOUNT(CopilotInteractions[PseudonymizedUserId]),
CopilotInteractions[FeatureType] = "Document Summarization"
),
DISTINCTCOUNT(CopilotInteractions[PseudonymizedUserId])
)
Microsoft Copilot Studio released new features in February 2025, including the ability to publish custom agents to Microsoft 365 Copilot Chat and enhanced analytics capabilities. (What's new in Copilot Studio: February 2025) These updates provide richer data for feature-level analysis.
Custom agents created in Copilot Studio can now be published directly to Microsoft 365 Copilot Chat, and their performance can be monitored with analytics. (What's new in Copilot Studio: February 2025) This creates new opportunities for tracking specialized AI agent usage alongside standard Copilot interactions.
Measuring productivity impact requires correlating Copilot usage with business outcomes. Worklytics explores new ways to model work, focusing on areas like workday intensity and work-life balance. (4 New Ways to Model Work)
Hybrid work has changed the shape of the workday, elongating the span of the day and changing the intensity of work. (4 New Ways to Model Work) Workday intensity can be measured as time spent on digital work as a percentage of the overall workday span, providing context for AI adoption patterns.
Outlook calendar analytics provides crucial context for understanding AI usage patterns. The average executive spends 23 hours a week in meetings, nearly half of which could be cut without impacting productivity. (Outlook Calendar Analytics: The Hidden Driver of Productivity in the Modern Workplace)
In hybrid and remote work environments, calendars have become battlegrounds where collaboration clashes with focus time, leading to overbooked teams and burnt-out employees. (Outlook Calendar Analytics: The Hidden Driver of Productivity in the Modern Workplace) Integrating calendar data with Copilot usage patterns reveals when AI tools are most effective.
Worklytics requires access to Outlook Mail API endpoints to extract metadata fields for comprehensive analysis. (Outlook Mail Sanitized) This integration enables correlation between meeting load and AI tool adoption.
Modern organizations use multiple collaboration platforms, requiring comprehensive data integration. Worklytics leverages Slack API endpoints to extract metadata fields for cross-platform analysis. (Slack Sanitized)
The platform also provides access to Google Meet activity data through GMeetActivity endpoints, enabling analysis of video collaboration patterns alongside AI usage. (Google Meet Sanitized)
Data protection remains paramount when analyzing AI usage patterns. Worklytics' DLP Proxy provides full field-level control over sensitive information while maintaining analytical value. (Microsoft Copilot Sanitized)
Key compliance considerations include:
Effective dashboard design balances comprehensive data presentation with user-friendly navigation. Power BI's implementation planning guidance emphasizes the importance of user experience in adoption tracking systems. (Power BI implementation planning: Adoption tracking)
Key design principles include:
| Section | Metrics | Visualization Type |
|---|---|---|
| Executive Summary | DAU, WAU, Adoption Rate | KPI Cards |
| Usage Trends | Daily interactions over time | Line Chart |
| Feature Adoption | Usage by Copilot feature | Stacked Bar Chart |
| User Segments | Power users vs. occasional users | Scatter Plot |
| Productivity Impact | Time saved, tasks automated | Gauge Charts |
| Geographic Distribution | Usage by office location | Map Visualization |
Modern dashboards require interactive elements that enable self-service analytics. Key interactive features include:
Real-time dashboards require careful consideration of data refresh strategies. Microsoft's APIs have rate limits that must be respected to avoid service interruptions. Implement exponential backoff and retry logic for robust data extraction.
Last Refresh Time =
MAX(CopilotInteractions[RefreshTimestamp])
This measure helps users understand data freshness and plan accordingly for time-sensitive decisions.
Robust error handling prevents dashboard failures and ensures consistent user experience. Implement comprehensive logging for:
As Copilot adoption grows, dashboard performance may degrade without proper optimization. Consider:
Worklytics' DLP Proxy transforms and pseudonymizes certain fields for data protection while maintaining analytical value. (Microsoft Copilot Sanitized) This integration ensures that sensitive information is protected throughout the analytics pipeline.
Key DLP features include:
Worklytics uses data anonymization and aggregation to ensure compliance with GDPR, CCPA, and other data protection standards. The platform's privacy-first approach enables organizations to gain insights while maintaining regulatory compliance.
Compliance considerations include:
Organizations often require custom metrics that align with their specific business objectives. Here are sample DAX measures for advanced analytics:
Copilot ROI =
DIVIDE(
[Time Saved Hours] * [Average Hourly Rate],
[Copilot License Cost]
)
Feature Stickiness =
DIVIDE(
[Daily Active Users],
[Monthly Active Users]
)
Adoption Velocity =
DIVIDE(
[New Users This Month],
[Total Licensed Users]
)
These measures provide insights into return on investment, user engagement depth, and adoption speed.
Comprehensive AI analytics often require integration with additional enterprise systems. Worklytics provides connectors for various platforms including Atlassian Jira Cloud for project management insights. (Atlassian Jira Cloud Sanitized)
Additional integrations might include:
Advanced implementations can incorporate machine learning models for predictive analytics. GitHub Developer Metrics projects demonstrate how to use Power BI for data visualization while showcasing metrics around AI tools like GitHub Copilot. (GitHub Developer Metrics)
The project includes metrics related to GitHub Copilot, comparing before and after figures to showcase the impact and value of Copilot in the development process. (GitHub Developer Metrics) This approach can be adapted for Microsoft 365 Copilot analytics.
Before deploying your dashboard to production, ensure:
Successful dashboard implementations require ongoing maintenance:
As Microsoft continues to evolve the Copilot platform, dashboard implementations must adapt accordingly. Worklytics cannot guarantee the indefinite accuracy of field descriptions, requiring ongoing monitoring and updates. (Microsoft Copilot Sanitized)
Establish processes for:
Building an end-to-end AI usage dashboard for Microsoft 365 Copilot requires careful planning, robust security implementation, and ongoing maintenance. By combining Microsoft's preview APIs with Worklytics' privacy-safe data processing capabilities and Power BI's visualization strengths, organizations can create comprehensive dashboards that drive AI adoption and measure productivity impact. (Microsoft Copilot Sanitized)
The integration of multiple data sources, from Copilot interactions to calendar analytics and cross-platform usage patterns, provides a holistic view of AI adoption in the modern workplace. (Outlook Calendar Analytics: The Hidden Driver of Productivity in the Modern Workplace) This comprehensive approach enables data-driven decisions that maximize the value of AI investments while maintaining strict privacy and security standards.
As Microsoft continues to enhance the Copilot platform with new features and capabilities, organizations with robust analytics foundations will be best positioned to adapt and optimize their AI strategies. (What's new in Copilot Studio: February 2025) The dashboard framework outlined in this guide provides that foundation, enabling continuous improvement and strategic AI adoption across the enterprise.
Microsoft has released new Interactions Export API and the forthcoming Graph 'Copilot Usage' endpoints that allow secure access to Microsoft 365 Copilot capabilities in custom applications. These APIs align with Microsoft 365 compliance standards and enable enterprise developers to leverage AI power while maintaining secure access to organizational knowledge without compromising governance.
Organizations can build comprehensive adoption dashboards that track active users, productivity impact metrics, and usage patterns by connecting Power BI to Microsoft's new Copilot APIs. The implementation includes authentication setup, DAX measures for calculating adoption rates, and real-time monitoring capabilities that provide insights into how teams are utilizing AI-powered productivity features.
Microsoft 365 Copilot supports four authentication schemes for API plugins: OAuth 2.0 authorization code flow, Microsoft Entra ID single-sign on (SSO) authentication, API key authentication, and anonymous access. The OAuth 2.0 flow with optional PKCE support is recommended for secure bearer token access, requiring registration with the OAuth 2.0 provider to obtain client ID and secret.
Worklytics provides privacy-safe data processing by sanitizing Microsoft Copilot data to remove personally identifiable information while preserving analytical value. Their approach focuses on aggregated usage patterns and productivity metrics without exposing individual user content or sensitive organizational data, ensuring compliance with privacy regulations while enabling meaningful insights.
Key security practices include implementing proper authentication flows with Microsoft Entra ID, using role-based access controls for dashboard viewing, encrypting data in transit and at rest, and following Microsoft 365 compliance standards. Additionally, organizations should implement data governance policies, regular security audits, and ensure that sensitive Copilot interaction data is properly anonymized before visualization.
Yes, custom agents published to Microsoft 365 Copilot Chat can be monitored with analytics capabilities. These agents can utilize features like agent handoff, Azure AI Foundry integrations, and autonomous triggers, and their performance metrics can be tracked through the dashboard framework to measure business process automation effectiveness and user engagement.
