Conversation Insights

Conversation Insights shows how your customer conversations are going — session volume, agent performance, customer sentiment, and when traffic peaks during the week.

What you can track

• Session counts: Total conversations, unassigned sessions, and agent assignments
• Agent performance: Human agent metrics including response times and handovers
• Label distribution: How conversations are categorized by labels
• Sentiment analysis: Customer sentiment breakdown across conversations
• Traffic patterns: When conversations happen throughout the week

Opening Conversation Insights

1. Click Insights in the left sidebar.
2. Click Conversations (selected by default).

Filters

Time range

Pick the period you want to analyze:

• Last 7 days (default)
• Today
• Yesterday
• This month
• This year
• All time

Channel filter

Narrow the view to specific channels:

1. Click the Channel Filter dropdown.
2. Select one or more channels:
   • Instagram
   • Facebook
   • WhatsApp
   • WebChat
3. Leave empty to view all channels.

Conversation counts

The Conversation Counts section shows key session metrics:

Metrics shown

• Total Sessions: All conversation sessions in the selected time range
• Unassigned Sessions: Conversations not yet assigned to an agent
• AI Agents: Sessions handled by AI
• Human Agents: Sessions assigned to human agents (also shows an "Unreplied" count)

Reading the metrics

• Live updates: Counts update in real time (marked with a "Live" badge)
• Unreplied count: For human agents, shows how many conversations haven't been replied to yet
• Auto-refresh: Data refreshes every 2 seconds

Agent performance

The Agent Performance section shows detailed metrics for your human agents.

Label distribution

The Label Distribution chart shows how conversations are categorized by labels.

Reading the chart

• Bar chart: Visual breakdown of label distribution
• Percentages: Each label shows its share of total conversations
• Sorted by count: Most frequent labels first

How to use it

• Identify common topics: See what customers ask about most
• Track trends: Monitor how label distribution changes over time
• Plan resources: Understand which topics need more attention

Sentiment distribution

The Sentiment Distribution chart analyzes customer sentiment across conversations.

Sentiment categories

Cloodot categorizes conversations as:

• Positive: Customers expressing satisfaction
• Neutral: Standard, factual conversations
• Negative: Customers expressing dissatisfaction or concerns

Reading the chart

• Bar chart: Sentiment breakdown at a glance
• Percentages: Share of conversations in each category
• Automatic analysis: Cloodot analyzes sentiment for each conversation

How to use it

• Track customer satisfaction: See overall sentiment trends
• Spot issues early: Identify negative sentiment spikes
• Measure improvement: Monitor sentiment changes over time

Session traffic

The Session Traffic heatmap shows when conversations happen throughout the week.

Reading the heatmap

• Days of week: Left column shows days (Monday through Sunday)
• Hours of day: Bottom row shows 24-hour periods (0-1, 1-2, etc.)
• Color intensity:
  • Light green: Lower activity
  • Medium green: Moderate activity
  • Dark green: High activity
  • Gray: No activity

Using traffic data

• Peak hours: Identify when you receive the most conversations
• Staffing: Plan agent schedules around traffic patterns
• Response optimization: Focus resources during high-traffic periods
• Hover for details: Hover over any cell to see the exact session count

Spotting patterns

• Weekday vs weekend: Compare activity across days
• Time zones: Factor in customer time zones
• Seasonal trends: Track how patterns shift over time

Best practices

Regular monitoring

• Daily: Review conversation counts
• Weekly: Analyze sentiment and label distribution
• Monthly: Review traffic patterns for staffing decisions

Turning insights into action

• High unassigned count: Add more agents or improve AI responses
• Negative sentiment trend: Investigate root causes and improve processes
• Peak traffic times: Ensure adequate coverage during busy periods
• Slow response times: Provide training or additional resources

Data-driven decisions

• Resource allocation: Schedule agents around traffic patterns
• Training focus: Address topics with high negative sentiment
• Process improvement: Identify bottlenecks from agent performance data

Troubleshooting

No data showing

Check your time range first — expanding to "All time" often reveals data that's outside the default 7-day window. Then verify:

• You have conversations in the selected period
• Channels are properly connected
• Conversations have been created

Data not updating

Refresh the page. If that doesn't fix it:

• Check your internet connection
• Verify conversations are being created
• Wait a few seconds (updates happen every 2-10 seconds)

Missing Agent Performance section

Check that the feature flag is enabled. Then verify:

• You have human agents assigned to conversations
• Conversations exist in the selected time range

Related topics

• Marketing Insights — Campaign analytics
• Intelligence Overview — AI-powered features
• Channels Overview — Channel connections