Reviewing and exporting chat feedback

Review and export chat feedback to understand user sentiment and improve response quality. This task helps you find trends, diagnose issues by using full conversation history, and share insights through downloadable reports.

1. On the OpenSearch dashboard, open the search interface.
2. In the search bar, enter one of the following queries.
   • Retrieve all feedback: "FeedbackGraph.registerFeedback Feedback"
   • Retrieve positive feedback: "isPositive:true"
   • Retrieve negative feedback: "isPositive:false"

   For more information about OpenSearch query syntax and filtering, see the OpenSearch documentation.
3. Set the date range for the time period that you want to analyze.
4. In the field selector, select the following fields:
   • ext_requestId
   • message
   • Any additional fields such as timestamp, if required.

   The following example shows a typical feedback entry structure.

   FeedbackGraph.registerFeedback Feedback: {
     isPositive: false,
     feedback: "Note was not updated correctly while cancelling the order",
     conversation: [
       {u: "cancel order"},
       {a: "To proceed with the cancellation of the order, please provide the order number."},
       {u: "Y100007740"},
       {a: "### Summary\n\nThe order Y100007740 has been cancelled successfully.\n\n| Order Number | Total Amount |\n|--------------|--------------||\n| Y100007740   | 25.00 USD    |"}
     ]
   }

   In this example,

   • The isPositive field is set to false, indicating negative feedback.
   • The feedback field contains the user's comment: "Note was not updated correctly while cancelling the order".
   • The conversation array shows the complete interaction, including the user's request to cancel an order, the agent's request for the order number, and the final cancellation. confirmation

5. Review the results that are displayed in the Results page.
6. To export the data, click Download as CSV in the Results page.