> ## Documentation Index
> Fetch the complete documentation index at: https://docs.callem.ai/llms.txt
> Use this file to discover all available pages before exploring further.
# Analytics
> Visualize call and chat volumes, durations, success rates, and custom metrics with interactive dashboards.
# Analytics
The Analytics dashboard gives you a comprehensive view of your voice and chat AI operations. Track KPIs, identify trends, and build custom charts to monitor what matters most to your business.
## Multichannel Support
Analytics aggregates data from **multiple channels**. Today the platform supports:
| Channel | Description |
| --------- | -------------------------------------------------------------------- |
| **Voice** | Phone calls handled by the voice gateway (SIP / telephony) |
| **Chat** | Text conversations from the web widget or test chat in Callem Studio |
You can view metrics for **Voice only**, **Chat only**, or **both** (combined). The channel selector is a multi-select dropdown at the top of the page. Labels and KPIs adapt: for example, "Total Calls" becomes "Total Chats" when only Chat is selected, or "Total Conversations" when both channels are selected. Additional channels (e.g. WhatsApp and social networks) are planned; the filter is designed to scale as new channels are added.
Latency KPIs (LLM / TTS / Global) are primarily relevant for Voice; chat sessions may show zero or partial TTS latency. Engagement, transfer rate, duration, and call analysis apply to both channels where applicable.
## Filters
All analytics data is filtered by the controls at the top of the page:
| Filter | Description |
| ----------------- | ------------------------------------------------------------------------- |
| **Date range** | Select the time period to analyze |
| **Duration** | Filter by conversation duration range (min/max in seconds) |
| **Agent** | Multi-select to filter by specific agents |
| **Channel** | Multi-select: Voice and Chat (or both). Future channels will appear here. |
| **Call Analysis** | Filter by custom analysis field values (e.g. `sentiment = positive`) |
All KPIs and charts update dynamically when filters change.
***
## Overview Dashboard
The built-in overview dashboard provides two rows of KPI cards and six default charts.
### KPI Cards — Row 1 (Volume & Engagement)
| KPI | Description |
| --------------------------------------- | -------------------------------------------------------------------------------------------------------------------------------- |
| **Total Calls / Chats / Conversations** | Total number of voice calls, chat sessions, or both (depending on channel filter) in the selected period |
| **Engagement Rate** | Percentage of calls that were engaged (meaningful conversations vs. hang-ups, voicemails). Subtitle shows the raw engaged count. |
| **Transfer Rate** | Percentage of calls transferred to a human agent. Subtitle shows the raw transfer count. |
| **Reiteration Rate** | Rate of callers who called back multiple times |
| **Peak Simultaneous** | Maximum number of concurrent calls at any point during the period |
### KPI Cards — Row 2 (Performance & Latency)
| KPI | Description |
| ----------------------- | -------------------------------------------------------------------- |
| **AHT** | Average Handle Time — mean duration of engaged calls, in seconds |
| **Interactions / Call** | Average number of back-and-forth exchanges per call |
| **LLM Latency** | Average time (ms) for the LLM to generate a response |
| **TTS Latency** | Average time (ms) for the text-to-speech engine to produce audio |
| **Global Latency** | Average end-to-end latency (ms) from caller speech to agent response |
LLM and TTS latency are critical metrics for call quality. If Global Latency exceeds 1500ms, callers may perceive the agent as slow or unresponsive. Consider using a faster LLM model or a lower-latency TTS provider.
### Default Charts
| Chart | Type | Description |
| ------------------------------ | ---------- | ------------------------------------------------------------------------------------- |
| **Call / Conversation Volume** | Area/Bar | Total calls or chats per time interval (auto-adjusts granularity based on date range) |
| **Concurrency** | Line | Simultaneous active calls/chats per minute/hour throughout the day |
| **Hourly Distribution** | Bar | Call distribution across 24 hours — identify peak hours |
| **Day of Week Distribution** | Bar | Call distribution across 7 days — identify peak days |
| **KPI Evolution** | Multi-line | Engagement rate / Transfer rate / AHT trends over time |
| **Latency Evolution** | Multi-line | LLM / TTS / Global latency trends over time |
### Per-Agent Table
A sortable table comparing agents side by side:
| Column | Description |
| -------------- | ---------------------------------- |
| **Agent** | Agent name |
| **Calls** | Total call count |
| **Volume** | Visual bar showing relative volume |
| **Engagement** | Engagement rate percentage |
| **AHT** | Average handle time |
***
## Custom Dashboards
In addition to the built-in Overview tab, you can create unlimited custom dashboard tabs.
### Managing Dashboard Tabs
* **Create**: click the **+** button next to existing tabs
* **Rename**: right-click a tab or use the dropdown menu
* **Reorder**: drag and drop tabs to rearrange
* **Duplicate**: create a copy of an existing dashboard
* **Delete**: remove a dashboard (the Overview tab cannot be deleted)
### Adding Charts
Each custom dashboard contains a grid of charts that you configure:
1. Enter **Edit Mode** by clicking the edit icon on a dashboard tab
2. Click **Add Chart** to open the chart editor
3. Configure the chart (see below)
4. Save — the chart appears in the dashboard grid
### Chart Sizes
Charts can be resized in the grid:
| Size | Grid Width | Best For |
| ---------- | ---------- | -------------------------------------------- |
| **Small** | 1/3 width | Single KPI numbers, small donuts |
| **Medium** | 2/3 width | Standard charts |
| **Large** | Full width | Detailed time series, complex visualizations |
Charts can be reordered via **drag and drop** in edit mode.
***
## Chart Editor
The chart editor is where you configure what data to visualize and how.
### Chart Types
| Type | Description | Best For |
| ---------- | -------------------------- | --------------------------------- |
| **Column** | Vertical bar chart | Comparing categories, time series |
| **Bar** | Horizontal bar chart | Ranking comparisons |
| **Line** | Line chart | Trends over time |
| **Donut** | Pie chart with center hole | Distribution / proportions |
| **Number** | Single large value | Highlighting one KPI |
### Metrics
#### Built-in Metrics
| Metric | Description |
| ------------------------- | -------------------------------- |
| **Call Counts** | Total number of calls |
| **Average Call Duration** | Mean call duration in seconds |
| **Engagement Rate** | % of engaged calls |
| **Transfer Rate** | % of transferred calls |
| **Voicemail Rate** | % of calls reaching voicemail |
| **LLM Latency** | Average LLM response time (ms) |
| **TTS Latency** | Average TTS generation time (ms) |
| **Global Latency** | Average end-to-end latency (ms) |
| **Interactions / Call** | Average exchanges per call |
#### Custom Analysis Metrics
If you've configured [call analysis fields](/build/agents#call-analysis-tab) on your agents, they appear automatically as available metrics. For example, if you have a `sentiment` field with values `positive`, `neutral`, `negative`, you can create a donut chart showing sentiment distribution.
#### Custom Rate Metric
Create calculated percentage metrics by defining a numerator and denominator:
* **Rate conditions** (numerator): filters that define the "success" subset
* **Base filters** (denominator): filters that define the total population
* **Logic**: combine multiple conditions with AND or OR
**Example**: "Hot lead rate among engaged calls"
* Numerator: `callAnalysis.leadQuality = hot`
* Denominator: `isEngaged = true`
* Result: percentage displayed as a chart
### Chart Filters
Each chart can have its own independent filters (in addition to the global page filters):
| Filter Category | Fields |
| ----------------- | --------------------------------------------------------------- |
| **Call fields** | Duration / isEngaged / callerNum / calledNum / totalInteraction |
| **Agent** | Filter by specific agent(s) |
| **Call Analysis** | Any custom analysis field defined on your agents |
#### Filter Operators
Operators vary by field type:
| Field Type | Available Operators |
| ------------ | ------------------------------------------------------------------------------------------ |
| **Number** | equals / not equals / greater than / greater-or-equal / less than / less-or-equal / Exists |
| **Boolean** | equals / Exists |
| **String** | Equals / Not equals / Contains / Exists |
| **Selector** | Equals / Not equals / Exists |
Multiple filters can be combined with **AND** or **OR** logic.
### Time Granularity
For time-series charts (Column / Line / Bar), choose the aggregation granularity:
| Granularity | Use Case |
| ----------- | --------------------------------- |
| **Hour** | Short date ranges (1-3 days) |
| **Day** | Medium ranges (1 week to 1 month) |
| **Week** | Long ranges (1-6 months) |
| **Month** | Very long ranges (6+ months) |
### Live Preview
The chart editor shows a live preview that updates as you change settings, so you can verify the visualization before saving.
***
## Sustainability Tab
The **Sustainability** tab provides environmental impact metrics for your voice AI operations. It tracks CO2 emissions and energy consumption broken down by AI component (LLM, TTS, STT).
See the dedicated [Sustainability](/monitor/sustainability) page for full details on KPIs, calculation methodology, and CSRD compliance.
## Best Practices
The built-in KPIs and charts cover the most common monitoring needs. Get familiar with them before creating custom dashboards.
Build separate dashboards for different stakeholders: an operations dashboard with volume and latency metrics, a sales dashboard with lead quality and booking rates, a quality dashboard with sentiment and resolution tracking.
The custom rate metric lets you calculate any business-specific percentage (e.g. "appointment booking rate among interested callers") directly in the dashboard without external tooling.
Global filters (date, agent, channel) apply to all charts. Use per-chart filters for specific slices — e.g. a chart showing only calls where `isEngaged = true`.
To compare or isolate performance by channel, use the **Channel** filter: select **Voice** for phone metrics and **Chat** for web widget metrics, or both for combined conversation metrics. Custom dashboard charts respect the same channel selection.