Skip to main content
The Conversations view gives you detailed access to every call your agents handle. This is where you do qualitative analysis: listening to recordings, reading transcripts, checking which tools were called, and reviewing goal results.

​
What You’ll Find

Each conversation in the Conversations tab includes:
Full text transcript showing the back-and-forth between your agent and the caller. See exactly what was said, including:
  • Agent responses
  • Caller input
  • Timestamps
  • Turn-by-turn flow
Use transcripts to spot patterns like repeated clarifications, confusion, or missing knowledge.
Review the actual audio recording of the call. Essential for:
  • Evaluating voice quality and tone
  • Understanding context that text misses
  • Hearing customer sentiment
  • QA and training purposes
Recordings help you catch issues that don’t show up in transcripts alone.
See every action your agent executed during the call:
  • API calls made
  • Parameters passed
  • Responses received
  • Execution timing
Review tool calls when debugging integrations or understanding why an agent took a specific action.
When your agent uses a knowledge base, you can see exactly which content chunks were retrieved for each response. Click the icon next to any agent message that used knowledge retrieval to open the retrieval details modal.What you’ll see:
  • Chunk count: How many knowledge chunks were retrieved
  • Content preview: The actual text that was retrieved and used
  • Similarity score: How relevant each chunk was to the query (displayed as a percentage)
  • Source information: Which knowledge base and item the chunk came from
Understanding similarity scores:
  • 70%+: Highly relevant match
  • 50-70%: Moderately relevant
  • Below 50%: Weak match, may need knowledge base improvements
Common issues to look for:
  • No chunks retrieved: The knowledge base may be missing content on this topic
  • Low similarity scores: Content exists but may need better wording or more detail
  • Wrong chunks retrieved: Review your knowledge base organization and content structure
  • Agent still gave wrong answer: The retrieved content may be accurate but the agent misinterpreted it—consider adjusting instructions
Use this view to debug knowledge-related issues and improve your knowledge base content over time.
After each call, configured goals are automatically scored:
  • Primary goal: Main success metric (Achieved / Not Achieved)
  • Secondary goals: Additional tracked outcomes
  • Reasoning: Why the goal was marked achieved or not
Filter conversations by goal achievement to find successful patterns or identify where your agent is falling short.Goals are configured in Agent Settings → Analytics → Goals. See the Conversation Goals guide for setup.
Custom questions you’ve configured to run after each call:
  • Yes/No questions: Binary outcomes
  • Multiple choice: Categorical data
  • Free-form text: Open-ended analysis
Results appear in the conversation detail pane for quick review.Configure post-call analysis in Agent Settings → Analytics → Post-Call Analysis. See the Post-Call Analysis guide for details.

​
How to Use It

Find specific conversations using filters:
  • Agent: Filter by which agent handled the call
  • Date range: Narrow to specific time periods
  • Status: Filter by call outcome (completed, failed, transferred)
  • Goal achievement: Show only calls where goals were achieved or missed
  • Campaign: For campaign calls, filter by campaign name
  • Phone number: Find calls to/from specific numbers

​
Weekly Review Routine

Make conversation review a regular habit:
1

Pick your focus

Decide what to investigate this week:
  • Calls that failed or transferred
  • Low goal achievement calls
  • Random sample for general QA
  • Specific campaign or time period
2

Review conversations

For 5-10 conversations:
  • Review the call
  • Read the transcript
  • Check tool calls and knowledge retrieval
  • Review goal results and analysis
3

Spot patterns

Look for recurring issues:
  • Agent doesn’t understand common questions
  • Missing knowledge on specific topics
  • Tools failing or returning errors
  • Certain scenarios causing transfers
4

Take action

Based on patterns:
  • Update agent instructions
  • Add missing knowledge
  • Fix or improve actions
  • Adjust goals or analysis questions

​
Conversation Exports

Export conversation data for deeper analysis in your own tools:
From the Conversations tab, export data with flexible options:Scope:
  • Selected conversations only
  • All conversations matching current filters
  • Custom date ranges
Format:
  • XLSX (recommended): Excel format with full formatting
  • CSV comma-separated: Standard CSV
  • CSV semicolon-separated: European CSV format
What’s Included:
  • Conversation metadata (date, duration, agent, phone numbers)
  • Call status and outcome
  • Goal achievement results
  • Post-call analysis answers
  • Tool call summaries
  • Transcript (optional, can be large)
Large exports run in the background:
  • Continue working while exports process
  • Check status in export history
  • Download completed files when ready
Each export shows format and status for easy tracking.

​
Post-Call Email Summaries

Stay in the loop without logging into the dashboard. Configure post-call emails to receive automatic summaries after each call:
  • What happened: Call outcome and duration
  • Goal results: Whether goals were achieved
  • Analysis results: Answers to post-call questions
  • Quick actions: Links to view full conversation details
Set up post-call emails in Agent Settings → Analytics → Post-Call Automation. See the Post-Call Automation guide.

​
Campaign-Specific Reporting

For outbound campaigns, the Campaign → Reporting tab gives you per-contact conversation history:

Contact History

See all conversations for each contact across the campaign lifecycle. Track multiple attempts, callbacks, and outcomes.

Campaign Exports

Export contacts with full conversation data:
  • All attempts and outcomes per contact
  • Goal achievement across calls
  • Analysis results aggregated per contact
  • Complete conversation history
Choose XLSX, CSV comma, or CSV semicolon format.
Campaign exports combine contact data with conversation analytics for complete campaign reporting.

​
Best Practices

Configure automatic email summaries to stay updated without manually checking the dashboard. Get notified on specific conditions like failed goals, transfers, or errors. This is highly recommended for staying informed effortlessly.
When first launching, review conversations daily or after every few calls to iterate quickly. Once stable, shift to weekly reviews to maintain quality without becoming overwhelming.
Start by reviewing failed calls, transfers, or missed goals. These reveal improvement opportunities faster than reviewing successful calls.
For trends across hundreds of calls, export to Excel or your BI tool. The conversation view is best for understanding individual calls.
When you see the same issue 3+ times, update your agent instructions or knowledge base. Don’t wait for perfection.
Keep notes on common problems. This helps prioritize what to fix first and track improvements over time.

​
Next Steps

Dashboard

Switch from qualitative to quantitative: track metrics and trends

Conversation Goals

Configure goals to automatically measure conversation success

Post-Call Analysis

Set up custom questions to extract insights from every call