Skip to main content
The Analytics Dashboard provides workspace-level metrics across all agents and phone numbers. Use it to understand usage trends, identify performance regressions, and track cost over time. Access it at Monitor > Analytics.

​
Available metrics

Call Volume

Total number of calls initiated in the selected time range. Broken down by day, week, or month depending on the range. Includes inbound and outbound calls.

Cost

Total TTS, LLM, and transcription cost for the selected period. Includes a per-agent and per-phone-number breakdown. Billed in USD.

Average Duration

Mean call duration in seconds. Useful for detecting unexpected changes — a sudden spike in average duration can indicate the agent is looping, getting confused, or callers are waiting on hold.

Success Rate

Percentage of calls that completed with the expected outcome as defined by post-call analysis. Requires post-call analysis to be configured on the agent.
Additional metrics shown in the dashboard:
  • Calls by agent: volume and cost per agent over the selected period
  • Calls by phone number: breakdowns by inbound number for multi-number workspaces
  • Call outcome distribution: breakdown of post-call analysis outcomes (if configured)
  • TTS cost vs. LLM cost vs. transcription cost: component-level cost attribution

​
Filters

FilterOptions
Date rangeLast 7 days, Last 30 days, Last 90 days, Custom range
AgentAll agents, or a specific agent
Phone numberAll numbers, or a specific inbound number
Filters apply to all charts simultaneously. Switching from “All agents” to a specific agent narrows every metric to calls handled by that agent. Sudden drop in success rate: typically indicates a prompt regression (a recent prompt edit broke agent behavior), a change in caller behavior, or a downstream service outage affecting function calls. Cross-reference the date of the drop with recent agent edits in the deployment history. Spike in average duration: callers are spending more time on each call. Common causes: the agent is looping on a particular flow step, a function call is timing out and causing retries, or callers are having trouble providing expected input. Drill into Session History to find representative long calls and identify the pattern. Cost spike with stable volume: cost per call has increased. Check which cost component grew (TTS, LLM, transcription) in the breakdown. LLM cost spikes often indicate longer conversations or more function calls per call. TTS cost spikes can indicate longer agent utterances or a change in voice model. Volume drop on a specific phone number: could indicate a routing change, number suspension, or a caller experience issue driving callers away. Check with your telephony provider if the drop is sudden.

​
Exporting data

Click Export in the top-right of the Analytics Dashboard to download the current view as a CSV file. The export contains the same time series data shown in the charts, with columns for date, metric name, agent (if filtered), and value. For programmatic access to analytics data, use the DialNexa API: 
GET /v1/analytics?start=2024-01-01&end=2024-01-31&agent_id=agent_xxx