Note
Access to this page requires authorization. You can try signing in or changing directories.
Access to this page requires authorization. You can try changing directories.
Understanding an agent's performance is important for enhancing user experience and achieving business goals. Copilot Studio agents include built-in analytics and KPIs, along with the option to create custom analytics and reports. You can continuously optimize agent performance over time by regularly reviewing these metrics and identifying gaps.
Measure engagement rates
The engagement rate for conversational agents measures what percentage of analytics sessions are classified as engaged. A session begins in an unengaged state when the user interacts with the agent, or the agent sends an initial message to the user. The session becomes engaged when a custom topic is triggered, or when the conversation enters one of the following system topics: Conversational Boosting, Escalate, or Fallback.
A high engagement rate means the agent is successful at guiding users into topics. A low engagement rate indicates topic coverage gaps or that users are unclear about agent capabilities. You can monitor the engagement rate on the agent's analytics dashboard.
Learn more in Measure agent engagement.
Track session outcomes
Session outcomes determine what state the conversation ended in. The outcome of an engaged session can be resolved, abandoned, or escalated.
Resolved sessions end successfully.
A confirmed resolved outcome occurs when the End of Conversation topic is triggered and the user confirms their query was answered. The user has the option to fill in a customer satisfaction score (CSAT). For these reasons, it's recommended to design conversations to finish with the End of Conversation topic.
An implied resolved outcome occurs when the session ends without user confirmation, but the agent's logic determines it was resolved.
Escalated sessions occur when the agent enters the Escalate topic or uses the Transfer Conversation node.
Abandoned sessions occur when the session ends without being resolved or escalated.
You can find conversation outcomes in the agent's analytics dashboard.
Learn more in Measure agent outcomes.
Choose an analytics strategy
Copilot Studio provides detailed analytics for each agent, offering insights into usage patterns and other key performance indicators. Learn more in Analytics overview. You can use analytics data generated by Copilot Studio in custom reports. When creating custom reports, plan how and where to store the data. Learn more in Develop a custom analytics strategy.
Define an optimization strategy
Improve an agent built with Copilot Studio over time to increase return on investment and customer satisfaction.
Review agent performance on a monthly basis and focus on these items:
- Deflection rate: Percentage of requests completed through self-service.
- Resolution rate: Percentage of sessions ending with a resolved outcome.
- Engagement rate: Percentage of sessions that were engaged.
- Topics with low resolution: Resolution rate for an individual topic.
- Unrecognized utterances: Phrases that didn't trigger any custom topic.
- Analysis per channel: Metrics to identify performance differences across channels.
Learn more in Deflection overview.
Collect thumbs up and thumbs down user feedback and comments
Users can give a reaction to any message sent by the agent by using the thumbs-up and thumbs-down buttons. When a user selects a reaction button, prompt the user to leave more detailed feedback in a comment. You can view all reactions and comments for the past 28 days in the agent's Analytics tab inside Copilot Studio.
Learn more in Reactions.
Analyze user questions by theme
Copilot Studio analyzes all user questions that trigger a Generative AI response and groups them into common themes. Themes help makers understand what questions users ask and how the agent is performing. You can view all related questions and response quality metrics for each theme.
Learn more in Analyze user questions by theme.
View conversation KPIs in Copilot Studio Kit
Conversation key performance indicators (KPIs) to track agent performance are also provided as part of the Copilot Studio Kit.
Copilot Studio Kit offers these benefits:
- Store aggregated metrics for agents in Dataverse
- Retain KPI data for long-term tracking
- Define specific variables to track as part of the KPIs, such as Net Promoter Score (NPS)
- Save the full conversation transcripts along with KPI records
- Extend the sample Power BI reports to create custom reports
Learn more in Track agent performance with conversation KPIs in Copilot Studio Kit.
Next step
Learn how to interpret signals such as conversation volume, session quality, and return‑use patterns to understand how users are interacting with your agent.