Feedback Analytics

Analyze user feedback to identify issues and improve your AI agent’s responses

The Feedback Analytics dashboard helps you understand why users are dissatisfied with agent responses, enabling you to identify patterns and make targeted improvements to your knowledge base and agent configuration.

Accessing Feedback Analytics

Navigate to Dashboard in the sidebar and select the Feedback tab.

You need Admin or Account Manager permissions to access Feedback Analytics.

Summary Metrics

The dashboard displays three key metrics at the top:

Metric	Description
Total Feedback	Total number of feedback submissions during the selected period
Unique Threads with Feedback	Number of distinct conversations that received feedback
Threads with Multiple Feedbacks	Conversations where users submitted feedback more than once

A high number of threads with multiple feedbacks may indicate conversations where the agent struggled to provide satisfactory answers.

Feedback Types

When users provide negative feedback, they can categorize the issue into one of five types:

Negative Rating

The user gave a thumbs-down rating to an agent response. This is the most general form of negative feedback.

Inaccurate

The agent provided information that was factually incorrect or misleading. This often indicates:

• Outdated information in the knowledge base
• Conflicting content across documents
• Misinterpretation of the user’s question

Incomplete

The response was partially correct but missing important information. This may suggest:

• Knowledge base gaps
• Documents that need more detail
• Agent instructions that are too narrow

Information Not Found

The agent couldn’t find relevant information to answer the question. This indicates:

• Missing documentation topics
• Poor keyword coverage in existing documents
• Questions outside the agent’s intended scope

Technical Issue

Users encountered a system error or technical problem, such as:

• Slow response times
• Failed message delivery
• Integration errors

Dashboard Components

Issues Distribution Chart

A pie chart showing the breakdown of feedback by type. This visualization helps you quickly identify which category of issues is most common.

Issue Type Frequency Chart

A bar chart displaying the count for each feedback type, making it easy to compare the volume of different issue categories.

Recent Feedbacks Table

A detailed table showing the most recent feedback submissions, including:

Column	Description
Thread ID	Link to the conversation where feedback was given
Issue Types	Badges showing which feedback categories were selected
Additional Feedback	Free-text comments provided by the user
Created At	Timestamp of the feedback submission

Click the thread ID to open the full conversation and review the context.

Analyzing Feedback Patterns

High Inaccuracy Rate

If you see many “Inaccurate” feedbacks:

1. Review the flagged conversations to identify incorrect information
2. Update or correct the source documents
3. Consider adding explicit corrections to agent instructions

High Incomplete Rate

If you see many “Incomplete” feedbacks:

1. Identify which topics lack sufficient documentation
2. Expand existing documents with more detail
3. Add new documents to cover missing subtopics

High Information Not Found Rate

If you see many “Information Not Found” feedbacks:

1. Analyze the questions users are asking
2. Identify documentation gaps
3. Consider expanding the agent’s scope or adding redirect responses for out-of-scope topics

High Technical Issue Rate

If you see many “Technical Issue” feedbacks:

1. Check system logs for errors
2. Review integration configurations
3. Monitor response times and system health

Filtering Options

Use the dashboard header filters to focus your analysis:

Filter	Description
Time Range	Select predefined ranges or custom dates
Organization	Filter feedback from a specific organization
User Type	Filter by user permission level
Agent	Filter by the agent that handled the conversation

Taking Action on Feedback

View All Feedbacks

Click the “View all feedbacks” link at the bottom of the table to access the complete feedback list with advanced filtering options.

Investigate Individual Issues

Click on any Thread ID to open the conversation and:

1. Read the full context of the interaction
2. Identify where the agent went wrong
3. Update knowledge base or agent instructions accordingly

Best Practices

1. Review feedback daily: Regular monitoring helps catch issues early

2. Track trends over time: Use period comparison to see if feedback is improving or worsening

3. Prioritize by impact: Focus on the most common feedback types first

4. Close the loop: After making improvements, monitor if the same types of feedback decrease

5. Use additional comments: The free-text feedback often contains specific details that help diagnose issues

Satisfaction Analytics

Satisfaction feedback data (response rates, satisfaction trends) is displayed in the Evaluation tab of the Dashboard, not in this Feedback tab. Navigate to Dashboard > Evaluation and scroll to the Satisfaction section to view response counts, response rate, satisfaction rate, and daily trends.

Satisfaction analytics appears in the Evaluation tab because it measures overall service quality alongside CX Score and thread evaluation data.

Related Features

• Thread Evaluator - Evaluate conversation quality with LLM scoring
• CX Score - Measure customer experience quality
• Conversations - Search and filter conversations