Skip to main content
Statistics helps your team understand how visitors use the chatbot and where answer quality may need attention. Use these screens after the widget has started receiving conversations. The Statistics area includes:
  • Domain statistics summary
  • Topic Analysis
  • Sentiment Analysis
  • Feedback Analysis
  • Knowledge Gaps
  • All Sessions
  • AI Customer Brain, when available on your plan

​
Domain statistics summary

Domain statistics summary gives you a top-level view of activity for the selected domain. The summary cards show:
  • Total Chats: how many chat conversations have been recorded
  • Unique Users: how many distinct visitor identities appear in chat activity
  • Average Session Duration: how long sessions usually last
  • Usage: the assistant usage volume shown by the product
Use these cards to spot changes in traffic and usage. For example, a sudden increase in chats may be expected after a campaign, while a sudden drop may mean the widget should be checked.

​
All Users table

The All Users table shows visitor-level activity for the selected domain. Common columns include:
  • Channel
  • User
  • Sessions
  • Questions
  • Durations
  • Cost
Use the search field to find a user by email or visitor identity when available. Use the channel filter to focus on a specific source such as the website widget or an integration. Choose View Chat to open that user’s Chat History.

​
Topic Analysis

Topic Analysis groups conversations by the topics visitors ask about. Use this screen to understand:
  • How many topics have been detected
  • Which topic appears most often
  • How many sessions have been analyzed
  • Whether there are sessions still waiting for analysis
  • Example questions for each topic
The All Topics table includes topic name, count, percentage, and example questions. Search by topic when you want to inspect a specific area. If sessions are waiting for analysis, use Analyze Now when you want to refresh topic results.

​
Sentiment Analysis

Sentiment Analysis summarizes the tone of analyzed sessions. Use this screen to review:
  • Positive sessions
  • Neutral sessions
  • Negative sessions
  • Average sentiment score
  • Sessions still waiting for analysis
  • Sentiment trend over recent days
  • Sentiment table details
Use this screen when you want to understand whether visitors are generally finding what they need or whether frustration is increasing. If sessions are waiting for analysis, use Analyze Now when you want to refresh sentiment results.

​
Feedback Analysis

Feedback Analysis uses visitor ratings collected from assistant answers when feedback collection is enabled. Use this screen to review:
  • Rated Responses: how many answers received feedback
  • Helpful Rate: the share of rated answers marked helpful
  • Bad Responses: answers marked unhelpful
  • Feedback Coverage: how much of the answer volume has feedback
  • High Confidence + Bad Feedback: cases where the answer looked confident but visitors rated it poorly
  • Low Confidence + Good Feedback: cases where the answer was cautious but visitors rated it helpful
Feedback Analysis includes several tables:
  • Confidence vs Feedback: compares confidence levels with visitor ratings
  • Confidence Score Distribution: shows how ratings differ across score ranges
  • Recent Miscalibrated Examples: shows cases where confidence and visitor feedback disagree
  • Recent Bad Responses: shows recent examples that received negative feedback
Use these examples to decide whether to update Reference Documents, adjust Widget Setup, or improve Q&A coverage.

​
All Sessions

All Sessions lists every chat session for the selected domain. Common columns include:
  • Channel
  • User
  • Session ID
  • Status
  • Message Count
  • Usage
  • Cost
  • Session Duration
  • Last Activity
Use the table to search by user, sort by activity, and open a specific session with View Session Detail. This is useful when you want to investigate a single conversation after seeing a quality issue in another Statistics screen. Use this routine for regular quality review:
  1. Start with Domain statistics summary to check overall activity.
  2. Review Topic Analysis to see what visitors ask about most.
  3. Review Sentiment Analysis to detect frustration or satisfaction patterns.
  4. Review Feedback Analysis for poor answers and rating mismatches.
  5. Open All Sessions or Chat History to inspect the exact conversation.
  6. Update Reference Documents, Q&A, or Widget Setup when the same issue appears repeatedly.