Goldilocks Docs
Insights

Insights Overview

Understand your contact conversations with analytics and AI-powered insights

The Insights section helps you understand what contacts are asking, how your AI is performing, and where you can improve. Everything is presented in a unified dashboard with key metrics, charts, and quick links.

What Are Insights?

Insights turns conversation data into actionable intelligence:

  • Dashboard - Unified view of ROI, performance metrics, and signal distribution
  • Rundowns - AI-generated summaries with actionable recommendations

Why Insights Matter

Improve Your AI

Identify where your AI struggles:

  • Questions without good answers
  • Topics needing more content
  • Frequent escalations

Understand Contacts

Learn what contacts care about:

  • Most common questions
  • Pain points and frustrations
  • Product feature requests

Measure Success

Track support quality:

  • Resolution rates
  • Response accuracy
  • Contact satisfaction

The Insights Dashboard

Navigate to Insights in the sidebar to view the unified dashboard.

Date Filter

At the top, select your time period:

  • Last 7 days - Recent activity
  • Last 30 days - Default view
  • Last 90 days - Quarterly trends
  • Custom range - Pick specific dates

ROI Metrics

Five key cards showing value delivered:

MetricDescription
Tickets DeflectedConversations resolved by AI without human intervention
Cost SavedEstimated savings based on ticket cost (default $15/ticket)
Resolution RatePercentage of issues fully resolved
Deflection RatePercentage not requiring human support
Total ConversationsNumber of chat sessions with signals applied

When significant changes occur, an alert panel appears:

  • Signals trending up or down
  • Pending actions from rundowns
  • Notable changes requiring attention

Click any alert to investigate further.

Signal Distribution

A bar chart showing what topics contacts ask about:

  • Click any signal to view related conversations
  • Trend indicators show if signals are rising or falling
  • Bars are sized relative to each other

Signals are configured in the Training section.

Configure Signals →

Signal Share

A donut chart showing the proportion of each signal:

  • Visual breakdown of topic distribution
  • Click to drill into specific signals
  • Legend shows percentages

Agent Performance

A table comparing your AI agents (personas):

ColumnDescription
AgentPersona name and type
ConversationsTotal handled in period
Resolution RatePercentage resolved without escalation
Escalation RatePercentage requiring human help
SatisfactionContact feedback score
Avg DurationAverage conversation length

Color coding indicates performance (green = good, amber = moderate, red = needs attention).

At the bottom, quick access to:

  • Rundowns - AI-generated summaries
  • Signals - Configure topic tracking
  • View Chats - Browse conversations

Key Metrics Explained

ROI Metrics

MetricCalculation
Tickets DeflectedResolved conversations × deflection rate
Cost SavedDeflected tickets × cost per ticket
Resolution RateResolved ÷ Total conversations
Deflection Rate(Total - Escalated) ÷ Total

Quality Metrics

MetricDescription
Resolution Rate% resolved without escalation
Escalation Rate% requiring human support
Satisfaction ScoreContact feedback ratings

Using Insights

Regular Review

Build insights into your routine:

  • Daily: Check alerts for trending signals
  • Weekly: Review dashboard and complete rundown actions
  • Monthly: Deep dive on agent performance

Team Collaboration

Share insights with your team:

  • Content team: Topics needing documentation
  • Product team: Feature requests and pain points
  • Support team: Common issues to prepare for

Action Loop

Turn insights into improvements:

  1. Identify - Find patterns in data
  2. Analyze - Understand root causes
  3. Act - Make improvements
  4. Measure - Track impact

Export Data

Click Export CSV to download signal data:

  • Signal tags with counts and percentages
  • Trend indicators
  • Useful for reports and presentations

Best Practices

Single data points can be misleading. Look for:

  • Patterns over time
  • Consistent themes
  • Significant changes

Prioritize Impact

Address insights that:

  • Affect many contacts
  • Have high escalation rates
  • Impact contact satisfaction

Close the Loop

After making changes:

  • Monitor the affected metrics
  • Verify improvements
  • Document what worked