Review your customer service agent's conversations
You can review conversations that your customer service agent has with customers to assess conversation quality, guide your agent’s future responses, and improve your agent’s performance. Conversation review lets you filter and search for specific conversations, see transcripts, rate the conversation, and provide feedback on responses.
You can find Conversation review in the sidebar of your agent settings.
Filtering conversations
You can filter to quickly find conversations such as unresolved cases or those with negative feedback. Use filters to narrow results by:
Conversation type – test or real
Last updated
Customer feedback – good or bad responses
Review status – good, bad, or not reviewed for conversations; reviewed or not reviewed for responses
Review tags – based on tags your team has assigned to responses in conversations
LLM review status – LLM judge’s review (unresolved or resolved)
Viewing conversation details
In the Conversations section, you’ll see a list of all conversations handled by your customer service agent. Each entry shows the customer’s initial question, the current status (Active, Escalated, or Ended), and the date and time. Selecting a conversation opens the full transcript of the conversation. From here you can:
read the customer’s queries and review the agent’s responses
rate the conversation as resolved or unresolved
open the conversation details panel to see the LLM status, conversation type and status, and feedback
review responses from the agent
Rating conversations
The “Did agent resolve the issue?” question applies to the entire conversation and lets reviewers mark if the agent successfully addressed the customer’s need. This is separate from customer feedback, which reflects the customer’s own rating. The resolution question is used for internal review.
Reviewing responses
You can review each agent response within a conversation and provide detailed feedback. When you select Review response in a conversation, a panel will open with the following options:
Guidance – add suggestions to improve how the agent handles similar questions in the future. Read more about guidance.
Tags – annotate responses for accuracy or to flag specific problems. Tags are for internal use and filtering but do not directly impact the agent’s behavior.
Notes – add comments to capture your thoughts about the response. Notes are for internal use and do not directly impact the agent’s behavior.
Was this helpful?