Manage statistics

Available in Classic and VPC

Analyzing statistical data and the speech records of incoming users is a critical task because their actual usage patterns may differ from the speech types anticipated during chatbot design, and modeling cannot be said to represent the speech patterns of all users. NAVER Cloud Platform provides various analytics indicators to help you continuously improve conversation models. You can view statistical data by domain and also check detailed statistical data for each domain.

View statistical data

To view statistical data on users or user conversations by domain, in NAVER Cloud Platform console, navigate to > Services > AI Services > CLOVA Chatbot > Management menu.

Overview
You can view the following statistical data regarding users and user conversations by domain.

• Actual users, messages, session counts, and increase/decrease rates for domains in active service
• Average and maximum users, messages, and sessions per hour
• Average active users and average messages per conversation session
• The number of chatbot service conversation models built and the number of services deployed
  

Actual Users
Provides information on the increase and decrease in actual chatbot users.

• When viewed on a daily basis, you can check the actual user information by time slot.
• Provides weekly and monthly trends in active users.
  

Messages
Provides information on the increase and decrease in user messages.

• When viewed on a daily basis, you can check the number of messages by time slot.
• Provides weekly and monthly trends in messages.
  

Sessions
Provides information on the increase and decrease of chatbot conversation sessions.

• When viewed on a daily basis, you can check the total number of sessions by time slot.
• Provides weekly and monthly trends in conversation sessions.
• You can also check the average number of messages of each user per session.
  

Build
Provides information on building conversation models and deploying services for domain administrators.

• When viewed data on a per-day basis, you can check the number of builds and build failures by time slot.
• Provides information on weekly and monthly conversation model builds and service deployments.
  

View integrated metrics

To view the integrated metrics for a specific domain:

1. In the NAVER Cloud Platform console, navigate to > Services > AI Services > CLOVA Chatbot > Domain.
2. Click [Run builder] of the domain you want to run the Chatbot Builder.
3. In the Chatbot Builder, click Integrated Metrics menu.

Integrated Metrics
Provides the following statistical data on chatbot users and user conversations, and displays the time the most recent data was reflected:

• Actual users of the chatbot service, user messages, number of conversation sessions and increase/decrease rates, failed conversations, number of general messages, number of single responses, number of similar responses, number of customer center integrations, number of service integrations (LINE Pay, NAVER Pay, NAVER Knowledge iN Service), etc.
• Information and trends on average active users and average messages per conversation session

Ranking Metrics
Provides the following ranking data on chatbot user conversations, and displays the time the most recent data was reflected:

• Ranking of the most frequently responded conversation names, conversation types, and entities during a specific period

Operational Metrics
Provides the following statistical data on the retraining of failed and dissatisfied conversations, and displays the time the most recent data was reflected:

• The number of failed (dissatisfied) conversations, edited cases, reviews pending, reviews excluded, total number of reviews

Feedback Metrics
Provides the following statistical data on feedback-related conversations, and displays the time the most recent data was reflected:

• YES/NO feedback ratio, feedback satisfaction rate/response rate, total feedback count, feedback count by conversation name
  

Korean domains provide sentiment analysis data that categorizes the emotions embedded in user questions as positive, negative, or neutral. Click the data to view questions classified under a specific sentiment.