Skip to content
Talk to an engineer

Analytics

Dashboard

Path: Analytics → Dashboard

The Dashboard is a customizable canvas of metric widgets. Each widget tracks a specific signal over time. Open it from the first icon in the Analytics left sidebar.

WidgetWhat it measures
ConversationsTotal agent conversation sessions
Conversation HandoffsSessions where the agent escalated to a human
User MessagesTotal messages sent by users across all conversations
Action CallsTotal times the agent executed an action
Resolution RateShare of conversations resolved — each conversation is classified as Resolved, Unresolved, or Escalated
LanguagesBreakdown of conversations by detected primary language
User SentimentAI-estimated emotional tone per conversation — rated Great, Good, or Negative
Signals FrequencyShare of conversations per signal tag — intent signals (Information Request, Action Request, Reported Issue) and agent gaps (Agent Knowledge Gap, Agent Action Gap), sorted by frequency

Each widget shows a trend indicator (up or down) with the percent change against the previous period. Rate widgets like Resolution Rate and User Sentiment show the change in percentage points (pp) rather than percent. Click the date range on any widget to adjust its comparison window.

Click + Add Widget in the top-right. The widget picker lets you browse by category or search by name.

Conversations category:

WidgetDescription
ConversationsTracks overall conversation volume across the selected period
User MessagesTotal number of messages sent by users across all conversations
Conversation HandoffsMeasures how often conversations were handed off to a human
Action CallsShows how often the agent triggered actions
Resolution RateShare of conversations that were successfully resolved. Each conversation is classified as Resolved, Unresolved, or Escalated
User SentimentAI-estimated emotional tone of users’ messages in each conversation. Each conversation gets one rating: Great, Good, or Negative
User IntentAI-classified intent signals from user messages: Information Request, Action Request, and Reported Issue. Signals can overlap — one conversation may appear in multiple intents
Signals FrequencyShare of conversations per signal tag: intent signals (Information Request, Action Request, Reported Issue) and agent gaps (Agent Knowledge Gap, Agent Action Gap). Tags are boolean flags and can overlap; sorted by frequency
LanguagesShare of conversations by detected primary language (mutually exclusive per conversation)

Audience & Revenue category:

WidgetDescription
New UsersTracks the number of new sign-ups over time, indicating user base growth
Active UsersReflects user engagement by showing the number of users interacting with the product
New SubscriptionsTracks the acquisition of new subscription plans within a specified timeframe
Active SubscriptionsDisplays the count of active subscription plans currently in use, for visibility into revenue streams
User RetentionMeasures the percentage of users who continue to engage with the application over time

Select one or more widgets and click Add Widgets to place them on the dashboard.

Each widget has controls in its top-right corner:

  • Date range (calendar icon): set the widget’s time window and comparison period.
  • Filters (funnel icon): scope the widget to a user segment or custom filter — a badge shows the number of active filters.
  • ⋯ More: scope to an Agent (all agents or one), switch the Chart Type, Refresh, Duplicate, or Delete the widget.

Drag the handle on the widget’s left edge to rearrange the dashboard.

Click Switch dashboard next to the dashboard name in the top bar to create additional dashboards or switch between existing ones. This is useful for separating an executive summary view from an operational monitoring view.

  1. Check the Dashboard weekly. Look for drops in Resolution Rate or spikes in Conversation Handoffs as early warning signs.
  2. Build an Agent Gap report. Run it monthly to see if Action Gaps or Knowledge Gaps are growing, then use that to prioritize what to add to Actions or Knowledge.
  3. Build an Action breakdown report. Understand which actions are called most versus least. Low-usage actions may need better agent instructions; zero-usage actions may have a bug.
  4. Use Audience to find power users. Filter to a recent Last Seen Date and sort by Number of visits, then reach out to the top users for feedback to inform your roadmap.
  5. Watch Session Replays for escalated conversations. Filter Session Replay to sessions with a handoff event to understand what context users had before escalating.