ChatSignals

Conversation Analytics

Browse, filter, and analyse your bot conversations in ChatSignals to understand customer intent, bot performance, and identify improvement opportunities.

The Chats page lets you browse and analyse every conversation your bot has handled. Use it to audit bot responses, understand customer behaviour, and identify areas for improvement.

Browsing Conversations

Navigate to Chats from the main menu. You'll see a list of all conversations with key information:

  • Summary: An AI-generated one-line summary of the conversation.
  • Participants: The customer, bot, and any human agents involved.
  • Status: Whether the conversation is open, closed, or pending.
  • Channel: Which bot/channel handled the conversation.
  • Date: When the conversation started and its most recent activity.

Filtering Conversations

Use the filters to narrow down what you're looking for:

  • Data Source: Filter by HubSpot connection if you have multiple.
  • Channel: View conversations for a specific bot.
  • Status: Open, closed, or pending conversations.
  • Date Range: Focus on a specific time period.
  • Search: Search conversation content by keyword.

You can also reach filtered conversation lists by clicking on topics from the dashboard or insights from the conversation intelligence page.

Analysing a Conversation

Click any conversation to see the full transcript. The detail view shows:

Message Timeline

The complete conversation thread including messages from:

  • Customers: Their original questions and follow-ups.
  • Bot: AI-generated responses with the knowledge sources used.
  • Human Agents: Any messages sent by your team after a handoff.
  • System: Status changes, handoff events, and ticket creation.

AI Analysis

For each conversation, ChatSignals provides:

  • Conversation Issues: The core questions or problems identified.
  • Resolution Status: Whether the conversation was successfully resolved.
  • Escalation Reasoning: Why the bot chose to respond directly or escalate.
  • Knowledge Used: Which knowledge source chunks were retrieved and their relevance scores.

What to Look For

When reviewing conversations, pay attention to:

  • Response Accuracy: Did the bot answer correctly using the right knowledge sources?
  • Handoff Quality: Did the bot escalate at the right moment, or too early/late?
  • Missing Knowledge: Were there questions the bot couldn't answer that it should have?
  • Customer Satisfaction: Did follow-up messages indicate the customer was satisfied?

Use these observations to refine your bot's knowledge sources and configuration.