Skip to main content

Analyze conversations

Review agent conversations, orders, and performance filters.

This narrated walkthrough demonstrates the same workflow documented in the written, step-by-step guide directly below.

Review the AI conversations dashboard​

The AI Conversations dashboard shows conversations, associated orders, and the performance metrics available for review.

Step-by-step feature breakdown​

1. High-level performance metrics​

At the very top of the interface, the dashboard presents a snapshot of your operational health. These "at-a-glance" cards track Total Chats, Active Sessions, and Completed Orders. By monitoring the Success Rate and Abandoned percentages, you can immediately identify if your AI is performing optimally or if there are emerging bottlenecks in the customer journey.

2. Precision filtering​

To manage high volumes of data, the dashboard offers three primary filters that allow you to slice and dice information:

  • Status Filter: Toggle between 'Active', 'Closed', or 'Writing' to see which conversations are currently live and which have reached a conclusion.
  • Reasons Filter: This is a diagnostic powerhouse. It allows you to isolate 'Abandoned' or 'Failed' chats to understand exactly where a customer might have lost interest or where the AI encountered a logic gap.
  • Business Selector: For enterprises managing multiple locations (e.g., "Pizza Pro’s" or "HR Music Store"), this dropdown filters the entire dashboard to show data for a specific branch, allowing for localized performance reviews.

3. Granular order transparency​

Select View Order on a completed chat card to review the associated customer information, delivery address, items, ingredients, and modifiers.

4. Conversation auditing​

By clicking "View Chat," administrators can jump directly into the transcript of any interaction. This allows you to review the AI’s tone, check for accuracy in responses, and ensure the "persona" of your brand is being maintained across every digital touchpoint.

Tips, tricks, and best practices​

  • The "Abandonment" Audit: Once a week, filter by "Abandoned" conversations. Look for patterns—are users dropping off when asked for their address? If so, consider simplifying the prompt or offering more clear instructions within the AI's logic.
  • Response Time Monitoring: Review the Response Time metric on individual chat cards and investigate material changes alongside API and connection data.
  • Compare Locations: Use the Business Selector to compare branches and review configuration differences when their displayed success rates vary.

Conclusion​

Use the dashboard filters, conversation transcripts, and order details together to investigate how an agent handled a customer interaction.