Skip to main content
Version: 6.0.0

Conversation Analytics

Introduction​

The Conversation Analytics dashboard is a powerful feature of the ACME ACE platform that provides administrators with comprehensive insights into user interactions, engagement patterns, and conversation performance. This tool allows you to monitor conversation volumes, question response rates, user feedback, and temporal patterns through intuitive visualizations. This documentation explains how to access, navigate, interpret, and leverage the Conversation Analytics feature to optimize your platform's conversational performance.

Table of Contents​

Accessing Analytics​

  1. Log in to the ACME ACE platform with Admin or SuperAdmin credentials
  2. Locate the main navigation menu on the left side of the interface
  3. Click on the Analytics icon or text in the navigation menu
  4. A dropdown will appear with two options:
    • Cost Analytics
    • Conversation Analytics
  1. From the Analytics dropdown menu, select Conversation Analytics
  2. You will be redirected to the Conversation Analytics dashboard
  3. The dashboard will load with the default time range (24 hours)

Dashboard Overview​

The Conversation Analytics dashboard is organized into several key components:

  • Page Header - Displays "Conversation Analytics" title
  • Time Range Selector - Allows filtering data by different time periods
  • Summary Metrics - Shows key conversation metrics in individual cards
  • Data Visualizations - Presents various charts, lists, and graphs for detailed analysis

Time Range Filtering​

The dashboard provides flexible time range options to view data across different periods:

FilterDescription
24HData from the past 24 hours (default view)
7DData from the past 7 days
1MData from the past month
3MData from the past 3 months
AllAll historical data available in the system

To change the time range:

  1. Locate the time filter buttons at the top of the dashboard
  2. Click on the desired time range (24H, 7D, 1M, 3M, or All)
  3. The dashboard will automatically refresh to display data for the selected period
  4. All metrics and visualizations will update to reflect the chosen time range

Key Metrics​

The Conversation Analytics dashboard displays five key metrics in summary cards at the top of the page:

Total Conversations​

  • Description: The total number of conversations that occurred within the selected time period
  • Location: First summary card on the left
  • Example Value: 1
  • Icon: Chat/conversation icon
  • Purpose: Provides an overview of platform engagement and conversation volume
  • Calculation Method: Count of all unique conversations initiated during the time period

Total Questions Answered/Unanswered​

  • Description: The ratio of questions that received answers versus those that went unanswered
  • Location: Second summary card from the left
  • Example Value: 0/1
  • Icon: Chart/graph icon
  • Purpose: Measures the effectiveness of the platform in addressing user queries
  • Calculation Method: Count of answered questions compared to unanswered questions

Average Hourly Conversation​

  • Description: The mean number of conversations per hour during the selected time period
  • Location: Third summary card from the left
  • Example Value: 1
  • Icon: User/profile icon
  • Purpose: Provides insight into conversation frequency and timing patterns
  • Calculation Method: Total conversations divided by the number of hours in the selected period

Upvotes/Downvotes​

  • Description: The ratio of positive feedback (upvotes) to negative feedback (downvotes)
  • Location: Fourth summary card from the left
  • Example Value: 0/0
  • Icon: Thumbs up/down icon
  • Purpose: Measures user satisfaction with conversation outcomes
  • Calculation Method: Count of upvotes compared to downvotes

Average Conversation Length​

  • Description: The mean duration or exchange count of conversations
  • Location: Fifth summary card from the left
  • Example Value: 1
  • Icon: Document/text icon
  • Purpose: Indicates the typical depth or complexity of conversations
  • Calculation Method: Sum of conversation lengths divided by the number of conversations

Data Visualizations​

The Conversation Analytics dashboard includes four main visualizations that provide detailed insights into conversation patterns:

Top 10 Asked Questions-Answered​

  • Type: List view
  • Location: Top-left visualization
  • Title: "Top 10 Asked Questions-Answered"
  • Description: Displays the most frequently asked questions that received answers
  • Data Displayed:
    • List of question text
    • Frequency count for each question
  • Example Status: "No Data Found" (in the screenshot)
  • Purpose: Identifies common queries that the system successfully addresses
  • Interpretation: Helps understand what information users most commonly seek and receive

Hourly Number of Conversations​

  • Type: Bar chart
  • Location: Top-right visualization
  • Title: "Hourly Number of Conversations"
  • Description: Shows conversation distribution across hours of the day
  • Data Displayed:
    • X-axis: Hour of day (00-23)
    • Y-axis: Number of conversations (0.0 to 1.0)
    • Purple bars representing conversation volume
  • Example: One conversation at hour 08
  • Purpose: Identifies peak usage times and conversation patterns throughout the day
  • Interpretation: Higher bars indicate hours with more active conversations

Top 5 Asked Questions-Unanswered​

  • Type: List view
  • Location: Bottom-left visualization
  • Title: "Top 5 Asked Questions-Unanswered"
  • Description: Displays the most frequently asked questions that went unanswered
  • Data Displayed:
    • List of question text
    • Frequency count for each question
  • Example: "what is pdc" with count of 1
  • Purpose: Identifies common queries that the system fails to address
  • Interpretation: Highlights knowledge gaps or areas needing improvement

Hourly Questions Answered - Unanswered​

  • Type: Bar chart
  • Location: Bottom-right visualization
  • Title: "Hourly Questions Answered - Unanswered"
  • Description: Shows the distribution of answered and unanswered questions by hour
  • Data Displayed:
    • X-axis: Hour of day (00-23)
    • Y-axis: Number of questions (0.0 to 1.0)
    • Color-coded stacked bars:
      • Green: Unanswered questions
      • Purple: Answered questions
  • Example: One unanswered question at a specific hour
  • Purpose: Identifies times when the system is more or less effective at answering questions
  • Interpretation: Helps pinpoint temporal patterns in system performance

Interpreting Results​

Conversation Analysis Insights​

  • Engagement Monitoring
    • Track conversation volumes over time
    • Identify trends in user engagement
    • Monitor peak usage times
  • Performance Assessment
    • Analyze question-answer success rates
    • Identify knowledge gaps based on unanswered questions
    • Evaluate user satisfaction through upvote/downvote ratios
  • Temporal Analysis
    • Identify high and low activity periods
    • Detect patterns in question answering efficiency by time
    • Plan resource allocation based on usage patterns

Common Patterns to Look For​

  • Usage Spikes
    • Identify unusually high conversation volumes during specific hours
    • Investigate potential causes of engagement anomalies
    • Correlate spikes with specific events or marketing activities
  • Knowledge Gaps
    • Analyze unanswered questions to identify systematic gaps
    • Prioritize content development based on frequently unanswered queries
    • Track improvement in answer rates over time
  • Satisfaction Indicators
    • Monitor upvote/downvote ratios as indicators of user satisfaction
    • Identify conversations or topics with particularly positive or negative feedback
    • Track improvement in satisfaction metrics over time

Use Cases​

For Administrators​

  • Performance Monitoring
    • Track conversation volumes and response rates
    • Identify times when additional support might be needed
    • Measure the impact of system improvements
  • Content Optimization
    • Identify knowledge gaps based on unanswered questions
    • Prioritize content development for frequently asked questions
    • Refine responses for questions with low satisfaction scores
  • Resource Planning
    • Allocate support resources based on peak usage times
    • Schedule maintenance during low-activity periods
    • Plan training and development based on performance metrics

For Content Managers​

  • Knowledge Base Enhancement
    • Identify topics needing additional content
    • Prioritize updates based on question frequency
    • Measure the impact of content additions
  • Query Analysis
    • Understand common user inquiries
    • Group similar questions for comprehensive coverage
    • Identify trending topics or concerns

User Permissions​

Access to the Conversation Analytics dashboard is restricted to authorized roles:

RoleAccess LevelDescription
SuperAdminFullComplete access to all conversation analytics data
AdminFullComplete access to all conversation analytics data
Standard UserNoneNo access to Conversation Analytics
GuestNoneNo access to Conversation Analytics

Cost Analytics​

The Cost Analytics dashboard complements Conversation Analytics by providing financial insights:

  • Navigation: Access by selecting "Cost Analytics" from the Analytics dropdown
  • Key Metrics: Includes total cost, average cost, total tokens, average tokens per request
  • Visualizations: Features data on cost distribution by request type, hourly costs, and user costs
  • Permissions: Accessible only to Admin and SuperAdmin users

For detailed information about Cost Analytics, please refer to the Cost Analytics documentation.

Troubleshooting​

Common Issues​

  • Data Not Loading
    • Refresh the browser page
    • Check your internet connection
    • Verify you have the required permissions (Admin or SuperAdmin)
    • Clear browser cache and cookies
  • "No Data Found" Messages
    • This is normal for sections like "Top 10 Asked Questions-Answered" if there isn't sufficient data in the selected time period
    • Try expanding the time range (e.g., from 24H to 7D)
    • Verify that conversations are being properly logged in the system
  • Visualization Rendering Issues
    • Ensure your browser is updated to the latest version
    • Try a different browser if charts aren't rendering properly
    • Check if JavaScript is enabled in your browser