Share via

Measuring copilot engagement

  • Article

Important

Power Virtual Agents capabilities and features are now part of Microsoft Copilot Studio following significant investments in generative AI and enhanced integrations across Microsoft Copilot.

Some articles and screenshots may refer to Power Virtual Agents while we update documentation and training content.

Tracking conversation engagement rates and outcomes are crucial to measure the copilot performance metrics, and spot areas for improvements.

Understanding conversations and analytics sessions

A single conversation between an end user and a Microsoft Copilot Studio copilot can generate one or multiple analytics sessions, typically happening when a user has new questions after an initial topic conversation path completed.

Note

Analytics sessions and billed sessions are distinct concepts.

Engaged and unengaged analytics sessions

Analytics sessions are either unengaged (default) or engaged.

This notion is used in the Microsoft Copilot Studio analytics dashboard.

Tip

Analytics session engagement is also tracked in conversation transcript records in Dataverse, if you have a custom analytics strategy.

In a conversation transcript record content, the engagement status of a session (either unengaged or engaged) is available as part of the SessionInfo activity, in the value type.

  • A session starts when a user interacts with the chatbot or when the copilot sends a proactive message to the user.
  • The analytics session begins in an unengaged state.
  • An analytics session becomes engaged when either entering a custom topic or the Escalate topic.
  • The last custom topic triggered—or the first system topic triggered, if no custom topics were triggered—is associated with the analytics session.

Chatbot engagement

Note

Having a copilot that proactively starts a conversation or that is placed on a website can greatly increase the session total and the number of unengaged sessions.

Measuring chatbot outcomes