Agent AI Analytics Feedback (Beta)

Back to Analytics Overview

The Feedback dashboard displays the widget-satisfaction data, suggestion quality metrics, and summary feedback for the Agent AI widget. It gives supervisors and administrators a consolidated view of how agents rate AI-generated suggestions, conversation summaries, and their overall widget experience.

Use the Feedback dashboard to:

• Monitor widget satisfaction rates.
• Identify low-performing suggestions.
• Review the top reasons agents give negative feedback.
• Monitor agent feedback on AI-generated summaries.

Access the feedback dashboard

1. Sign in to AI for Service.
2. Go to Agent AI > Analytics.
3. Select Agent AI > Feedback.
4. Filter data based on the following table:

Filter	Description
Date range	Filters all metrics by the selected time period. Defaults to Today.
Channels	Filters data by channel: Chat, Voice, or Email. Defaults to All Channels.

Session and suggestion metrics

The top of the Feedback dashboard shows four summary metrics for the selected date range. Each metric card includes a delta indicator (↑ or ↓) that compares the current period against the previous equivalent period.

Metric	Description
Agent AI Sessions	Total Agent AI widget sessions—My bot, search, and Agentic Copilot—across the selected time range.
Total Suggestions	Total suggestions shown to agents across all sessions.
Used Suggestions	Total suggestions agents used—copied or sent—across all sessions.
Suggestions Per Session	Average number of suggestions per session.

Overall Agent AI widget feedback

The Overall Agent AI Widget Feedback panel shows how agents rate their overall experience with the Agent AI widget. A donut chart visualizes the distribution of Satisfied, Unsatisfied, and No Feedback responses across all conversations.

Select or clear Satisfied, Unsatisfied, or No Feedback on the right of the donut chart to include or exclude the corresponding responses from the chart.

Metric	Description
Participation Rate	The percentage of conversations in which agents submitted the overall widget feedback. Use this metric to monitor how actively agents submit feedback, and encourage agents who haven’t participated to do so. Higher participation improves widget performance.
Satisfied	Conversations agents rated as satisfied. Shows as a count and percentage of total conversations.
Unsatisfied	Conversations agents rated as unsatisfied. Shows as a count and percentage of total conversations.
Total Conversations	Total conversations for the selected time range.

Details

The Details bar chart shows the top five reasons agents selected Satisfied or Unsatisfied.

• Select Satisfied or Unsatisfied to switch between views.
• Select View All to open the Details panel, which lists all feedback entries for the selected sentiment. The panel contains the following columns:

Column	Description
Conversation ID	Unique identifier for the conversation in which the agent submitted feedback.
Reasons	Feedback reasons the agent selected, such as Inaccurate Search Results or Irrelevant Dialogs. A conversation can have multiple reasons.
Additional Comments	Free-text comment the agent entered with their feedback.
Date & Time	Date and time the agent submitted the feedback.

• Enter a reason in the Search by Reasons box to filter rows by matching words in the Reasons column.
• Select a Conversation ID to view the Conversation Log page. It shows the conversation summary, customer-agent transcripts, agent feedback on widget performance and AI suggestions, and conversation identifiers. For more details, see Conversation Log.
• Select the download icon to export the data in CSV format. In the confirmation window, select Proceed to save the file to your device, or select Cancel to close the window without exporting.

Suggestions feedback

The Suggestions Feedback panel shows how agents rate individual AI-generated suggestions. A donut chart visualizes the distribution of Positive, Negative, and No Feedback responses across all suggestions.

Select or clear Positive, Negative, or No Feedback on the right of the donut chart to include or exclude the corresponding responses from the chart.

Metric	Description
Participation Rate	Percentage of suggestions for which agents submitted feedback.
Positive	Suggestions agents rated positively. Shows as a count and percentage of total suggestions.
Negative	Suggestions agents rated negatively. Shows as a count and percentage of total suggestions.
Total Suggestions	Total suggestions for the selected time range.

Negative feedback reasons

The Negative Feedback Reasons bar chart shows the top five reasons agents gave negative feedback on AI-generated suggestions.

• Select View All to see the complete list of negative feedback reasons.
• Select Negative or Positive to filter the list under those tabs.
• Enter a reason in the Search by Reasons box to filter rows by matching words in the Reasons column.
• Select a Conversation ID to view Conversation Logs.
• Select the download icon to export the data in CSV format. In the confirmation window, select Proceed to save the file to your device, or select Cancel to close the window without exporting.

Feedback Details: Negative and Positive Feedback

Column	Description
Conversation ID	The unique identifier for the conversation in which the agent submitted feedback.
Source	Source from which the suggestion came. For example, Agentic.
Involved App	The application associated with the conversation.
Customer Input	The message or query the customer entered during the conversation.
AI Suggestion	The response or suggestion AI provided to the agent.
Reasons	The reason the agent selected for the negative feedback.
Additional Comments	Any additional comments the agent entered when submitting feedback.

The Positive view excludes the Reasons and Additional Comments columns.

Summary Feedback

The Summary Feedback panel shows how agents rate AI-generated conversation summaries. A donut chart visualizes the distribution of Positive, Negative, and No Feedback responses.

Select or clear Positive, Negative, or No Feedback on the right of the donut chart to include or exclude the corresponding responses from the chart.

Metric	Description
Participation Rate	Percentage of summaries for which agents submitted feedback.
Positive	Summaries agents rated positively. Shows as a count and percentage of total summaries.
Negative	Summaries agents rated negatively. Shows as a count and percentage of total summaries.
Total Summaries	Total summaries for the selected time range and summary type.

Feedback Reasons

The Feedback Reasons bar chart displays the top five reasons agents selected when submitting negative feedback on summaries for the selected summary type.

Select Mid Conversation Summaries, Agent Transfer, or Closing to filter the panel by summary type.

• Select View All to see the complete list of negative feedback reasons.
• Select Negative or Positive to filter the list under those tabs.
• Enter a reason in the Search by Reasons box to filter rows by matching words in the Reasons column.
• Select a Conversation ID to view Conversation Logs.
• Select the download icon to export the data in CSV format. In the confirmation window, select Proceed to save the file to your device, or select Cancel to close the window without exporting.