Analytics
Analytics help you understand how your voice agents are performing. This section covers call logs, metrics, dashboards, and how to use data to improve your agents.
Dashboard Overview
The main dashboard provides a quick overview of key metrics:
Call Volume Chart:
Key Metrics
Call Volume Metrics
| Metric | Description | What It Tells You |
|---|---|---|
| Total Calls | Number of calls in period | Overall usage level |
| Calls per Day/Week | Average call frequency | Demand patterns |
| Peak Hours | Busiest times | Staffing needs |
| Trend | Change vs. previous period | Growth or decline |
Quality Metrics
| Metric | Description | Good Target |
|---|---|---|
| Success Rate | Calls resolved without escalation | Above 70% |
| Avg Duration | Average call length | 2-5 minutes |
| Escalation Rate | Calls transferred to human | Below 30% |
| Sentiment Score | Customer satisfaction | Above 70% positive |
Aim for above 70% success rate and below 30% escalation rate. If your numbers fall outside these ranges, review your agent's prompt and knowledge base for gaps.
Agent Performance
| Metric | Description | Purpose |
|---|---|---|
| Calls per Agent | Distribution of calls | Load balancing |
| Resolution by Agent | Success rate per agent | Identify issues |
| Avg Duration by Agent | Efficiency comparison | Optimization |
Call Logs
Accessing Call Logs
- Click "Call Logs" in the sidebar
- View the list of all calls
- Use filters to narrow results
- Click individual calls for details
Call Log List
Each entry shows:
| Column | Information |
|---|---|
| Date/Time | When the call occurred |
| Duration | How long the call lasted |
| Agent | Which agent handled the call |
| Caller | Phone number (if available) |
| Call Type | PHONE (telephony) or WEB (widget) |
| Direction | INBOUND (customer called) or OUTBOUND (agent called) |
| Status | STARTED, SUCCESS, FAILED, or PENDING |
| Disconnection Reason | Why the call ended (see below) |
Disconnection Reasons
When a call ends, the platform records why it was disconnected:
| Reason | Description |
|---|---|
| USER_HANGUP | The caller hung up normally |
| AGENT_HANGUP | The agent ended the call (e.g., after farewell) |
| CALL_TRANSFER | The call was transferred to a human operator |
| INACTIVITY | The caller was silent too long and the call timed out |
| CONCURRENCY_LIMIT_REACHED | Too many simultaneous calls for the account |
| ERROR_TWILIO | A telephony provider error occurred |
| ERROR_ASR | A speech recognition error occurred |
| ERROR_SIP | A SIP connection error occurred |
| ERROR_AGENT_INIT | The agent failed to initialize |
| USER_LEFT_EARLY | The caller disconnected before the agent could respond |
| NO_CREDITS | The user ran out of credits mid-call |
Filtering Calls
Filter by:
- Date range — Last 7 days, 30 days, custom
- Agent — Specific agent
- Call type — Phone or Web
- Direction — Inbound or Outbound
- Status — Started, Success, Failed, Pending
- Duration — Short, medium, long
Call Details
Click any call to see:
| Field | Value |
|---|---|
| Call ID | call_abc123xyz |
| Date | January 15, 2024 at 14:32 |
| Duration | 3 minutes 42 seconds |
| Agent | Reception Assistant |
| Caller | +43 664 123 4567 |
| Call Type | PHONE |
| Direction | INBOUND |
| Status | SUCCESS |
| Disconnection Reason | USER_HANGUP |
Optional features (if enabled in agent settings):
| Feature | Description |
|---|---|
| Sentiment Analysis | Automatic analysis of customer sentiment (positive, neutral, negative) |
| Call Summary | AI-generated summary of the conversation |
| Call Recording | Audio recording of the full call, with playback controls |
| Data Retrieval Results | Structured data collected during the call (e.g., name, email, phone) |
Transcript:
Agent: Hello! Thank you for calling ABC Company. How can I help you today?
Caller: Hi, I'd like to know your business hours.
Agent: We're open Monday through Friday from 9am to 6pm, and Saturdays from 10am to 4pm.
Caller: Great, thanks! Can I make an appointment?
...
Transcript Features
- Full conversation text
- Speaker identification (Agent vs Caller)
- Timestamps for each message
- Searchable content
Audio Playback
If recording is enabled in agent settings:
- Play full recording
- Skip to specific parts
- Adjust playback speed
- Download audio file
Sentiment Analysis
The platform automatically analyzes customer sentiment:
Sentiment Categories
| Category | Indicators | Score Range |
|---|---|---|
| Positive | Happy, satisfied, grateful | 70-100% |
| Neutral | Business-like, factual | 30-70% |
| Negative | Frustrated, angry, disappointed | 0-30% |
Using Sentiment Data
High negative sentiment signals a problem. Review transcripts of negative calls to find patterns — specific topics causing confusion, time-of-day issues, or agent-specific problems that need prompt adjustments.
Improve agents:
- Study positive calls to replicate successful approaches
- Analyze negative calls to fix prompt issues
- Track sentiment trends to measure improvements over time
Cost Breakdown
Track how your credits are being spent across agents and time periods.
Available Views
| View | Description |
|---|---|
| By Agent | See which agents consume the most credits |
| By Date | Track spending over time |
| Custom Grouping | Group costs by your preferred dimension |
Exporting Cost Data
Export cost breakdowns as CSV for accounting and invoicing:
- Go to the cost breakdown view
- Apply desired filters (date range, agents)
- Click Export to download CSV
Exporting Data
Export Options
| Format | Best For |
|---|---|
| Excel (.xlsx) | Analysis in spreadsheets |
| CSV (.csv) | Data processing, import to other systems |
| Reports, documentation |
What's Exported
- Call metadata (date, duration, agent, etc.)
- Transcripts
- Sentiment scores
- Custom fields
Export Steps
- Apply desired filters
- Click "Export" button
- Select format
- Download file
Using Analytics to Improve
Identify Common Issues
- Review calls with negative sentiment
- Look for patterns:
- Same questions causing confusion
- Specific topics leading to escalation
- Time-of-day issues
Optimize Knowledge Base
Add more information to the knowledge base, create FAQ entries for common questions, and clarify confusing content. Small knowledge base updates often produce significant improvements in resolution rates.
Improve Prompts
Agent behaving incorrectly?
- Review transcripts of problem calls
- Identify prompt gaps
- Add specific instructions for scenarios
Adjust Agent Settings
Calls too long?
- Adjust prompt for brevity
- Review if agent is being too verbose
High escalation rate?
- Review escalation triggers
- Add more information to KB
- Train agent for difficult scenarios
Metrics Glossary
| Metric | Calculation | Meaning |
|---|---|---|
| Success Rate | (Completed - Escalated) / Total | % resolved by agent |
| Avg Duration | Total minutes / Total calls | Time per call |
| Escalation Rate | Escalated / Total | % transferred to human |
| First Call Resolution | Resolved without callback / Total | % solved first time |
| Sentiment Score | Weighted average of call sentiments | Overall satisfaction |
Usage Page
The Usage page is a dedicated section (accessible from the sidebar navigation) that gives you a comprehensive view of how your agents are being used and what they cost.
Summary Cards
At the top of the page, four summary cards show high-level numbers for the selected period:
| Card | What It Shows |
|---|---|
| Total Calls | Number of calls in the selected period |
| Total Duration | Combined duration of all calls |
| Average Duration | Average length per call |
| Total Cost | Total credits or cost spent |
Filters
Use the filter bar to narrow down the data:
| Filter | Options |
|---|---|
| Workspace | Select a specific workspace |
| Agent | Filter by a specific agent |
| Period | Last 7 days, 30 days, 90 days, or a custom date range |
| Cost Type | Phone, Web, Messages, Forwarding |
| Direction | Inbound, Outbound |
| Status | Success, Failed |
Charts
Two charts visualize trends over the selected period:
- Daily Call Volume — A bar or line chart showing how many calls were made each day. Useful for spotting busy periods and trends.
- Daily Spending — A chart showing your daily cost. Helps you track spending patterns and catch unexpected spikes.
Breakdowns
Below the charts, several breakdowns let you slice the data in different ways:
| Breakdown | What It Shows |
|---|---|
| By Direction | Split between inbound and outbound calls |
| By Spending | How costs are distributed |
| By Call Type | Phone calls vs. web widget calls |
| By Status | Successful vs. failed calls |
| By Agent | Which agents are handling the most calls and cost |
Sentiment Overview
When you filter down to a single agent, a Sentiment Overview section appears showing the distribution of positive, neutral, and negative sentiment across that agent's calls.
Daily Spending Table
A tabular view of your daily spending — shows cost per day so you can track spending trends and identify spikes.
Recent Calls
A paginated table at the bottom lists recent calls with key details (date, agent, duration, status, cost). Click any row to view the full call details.
Export
Click the Export button to download the current view as a file. Use filters first to export exactly the data you need — for example, a specific agent's usage over the past 30 days.
Best Practices
Regular Review
| Frequency | What to Review |
|---|---|
| Daily | Quick glance at volume and sentiment |
| Weekly | Detailed metrics review, identify trends |
| Monthly | Full analysis, compare to goals |
| Quarterly | Strategic review, plan improvements |
- Set benchmarks — Know what "good" looks like for your business
- Track changes — Monitor metrics after making adjustments
- Test hypotheses — Make specific changes and measure their impact
- Document learnings — Keep notes on what worked and what did not