Analytics dashboard

Edit online

Use the analytics dashboard to monitor agent performance, identify issues, and make data-driven improvements.

Note: This feature is currently not supported in IBM Cloud, AWS GovCloud (US) and On-premises deployments.

The analytics dashboard provides a centralized view of agent activity, performance metrics, and user interactions. Use it to identify trends, evaluate feedback, review conversations, and improve agent behavior.

Note: The legacy Analytics experience is deprecated and will eventually be removed. It is recommended that you start using the new analytics experience.

1. Start with the dashboard

Begin by reviewing overall performance metrics:

  1. From the main navigation menu, select Analyze.

  2. Review the key metrics at the top of the page, such as total conversations, unique users, and average conversation duration.

This overview provides a quick understanding of overall system performance.

2. Filter the data

Refine the data to focus your analysis:

  1. Select a time range (for example, past 7, 15, or 30 days).

  2. Choose specific agents from the All agents filter.

  3. Apply additional filters if available.

Filtering helps isolate trends and investigate specific scenarios.

3. Identify trends and patterns

Use charts and metrics to detect patterns:

  • Look for high or low conversation volumes.

  • Identify changes in performance over time.

  • Compare metrics across agents.

Focus on trends rather than individual data points to understand overall behavior.

4. Review user feedback

Evaluate how users experience your agents by analyzing the feedback they provide in response to the agent's outputs:

  • Review thumbs up and thumbs down feedback.

  • Search for specific agents in the feedback table.

  • Identify patterns in negative feedback.

User feedback highlights quality issues and improvement opportunities.

5. Investigate specific agents

In the User Feedback table, click an agent name to drill down into the single-agent dashboard and view its detailed performance metrics.

  • Identify agents with unusually high or low usage.

  • Compare key metrics such as conversations, users, and duration.

  • Look for inconsistencies across agents.

This helps identify which agents are performing well and which may require improvement.

6. Review conversations

Examine individual conversations to better understand agent behavior:

  • Select the Conversations tab from the analytics page.

  • Review conversation transcripts to identify issues or patterns.

  • Analyze how agents respond to different types of user queries.

Reviewing conversations provides context for metrics and helps identify specific areas for improvement.

7. Debug agent responses

Analyze and troubleshoot issues in agent responses to improve accuracy and reliability:

  • Identify responses that are incorrect, incomplete, or irrelevant.

  • Trace the flow of the conversation to understand how the agent arrived at a response.

  • Check prompts, inputs, and intermediate steps for errors or ambiguity.

  • Validate integrations, tools, or data sources used by the agent.

  • Iterate on prompt design and configurations to correct issues.

Debugging helps ensure that agents produce consistent and accurate responses across scenarios.

8. Take action

Use insights from analytics to improve performance:

  • Optimize agents with low engagement or poor feedback.

  • Investigate and resolve slow or inefficient responses.

  • Update prompts, workflows, or configurations.

  • Monitor changes over time to measure impact.

Continuous analysis and iteration help ensure consistent, high quality agent performance.