Skip to main content

Monitoring Performance and Agent Analytics

Learn how to review analytics for your AI agents to understand performance and improve outcomes.

Maria Cairns avatar
Written by Maria Cairns
Updated over a week ago

Why it matters

Analytics provide visibility into how your AI agent performs across calls. Monitoring these metrics helps you identify trends, catch issues early, and refine instructions or call settings to deliver more reliable experiences. Although AI agents can also message, analytics are focused on call-based interactions.

Key Concepts

Analytics tab: A dashboard showing aggregated performance data for an agent.

Call volume: Total number of calls handled during a selected period.

Sentiment: Emotional tone of conversations based on transcript analysis.

Call outcomes: High-level results derived from parsing fields or inferred call patterns.

Performance trends: Changes over time in metrics such as call duration, sentiment, or call frequency.

Step-by-Step: Monitor Agent Analytics

  1. Open your agent and select the Analytics tab at the top of the editor.

  2. Review overall call volume to understand how frequently the agent is being used.

  3. Check sentiment metrics to identify general caller response patterns.

  4. Review call duration trends to spot unusually long or short interactions.

  5. Look at outcome summaries or parsed data fields to validate extraction quality.

  6. Compare the current period against previous periods to identify shifts in performance.

  7. Return to the Editor tab to refine instructions, voice settings, or parsing configurations as needed.

Tips and Best Practices

  • Track sentiment trends to gauge improvements or declines in perceived call quality.

  • Compare call duration patterns after major instruction or voice-setting changes to confirm expected results.

  • Use analytics alongside individual call reviews for a complete understanding of performance.

  • If metrics shift unexpectedly, check the recent Version history for changes to the agent’s configuration.

  • Evaluate trends instead of isolated call anomalies to guide refinements.

Troubleshooting

Issue

Possible Cause

Fix

Analytics appear empty or outdated

Limited recent call volume

Run test calls or verify that the agent is assigned to a phone channel

Sudden changes in sentiment or call duration

Recent edits to instructions or call settings

Review version history and adjust configurations

Metrics seem inaccurate

Parsing fields misconfigured

Refine parsing descriptions and examples for more consistent aggregation

Did this answer your question?