Overview

When managing your AI Assistant, it's essential to analyze its performance regularly to ensure it meets your users' needs. This article will help you understand the AI Assistant report, including key metrics, filters, and insights available.

Conversations

You can filter the conversations by User IDs, Thread IDs, Likes, Dislikes, Language, Fallback Generated, and Date Created.

You can use the Export CSV function to export conversations within a specified timeframe, including the desired attributes you would like to have in your CSV file.

Conversation ID

It's a unique ID we assign to each conversation.

Topics

Explore the topics discussed in chat sessions to identify common user needs:

• Click on a chat topic to open the full conversation.

• You can review the user's chat history with the AI Assistant in the conversation drawer.

Language

Analyze conversations based on language to track engagement across different user segments.

Status

Ongoing

• The conversation has not ended yet.
• The user is still active in the chat or waiting for a resolution within 24 hours.
  

Resolved

• A resolution is counted in a conversation when the customer confirms any answer provided is satisfactory or exits the conversation without requesting further assistance.
• A user has 24 hours to return to the conversation. If the customer does not request further assistance after 24 hours, this will be considered a resolution.
• After 24 hours from the last message, a conversation will be in read-only mode. Users can read previous messages but not follow up with new queries or interact with the fallback buttons in the chat.
• Customers can confirm an answer is satisfactory by pressing the Thumbs Up ‘👍’ for a response text provided by the AI Assistant in a chat.
• Customers can request further assistance by:
• • Entering a response that suggests their query is unresolved.
  • Asking clarifying questions or
  • Asking to speak to human/agent support.
    

Unresolved

• A conversation is not resolved when:
• • The user receives the Fallback Reply you set.
  • The user clicks on the Thumbs Down Button ‘👎’ for the latest response in the chat and asks no further questions.

Reaction

Measure user satisfaction using the feedback provided after chat interactions:

👍 Indicates positive reaction.

👎 Indicates negative reaction.

Feedback helps you understand users' feelings about the AI Assistant's responses and where improvements are needed.

Performance

Under the Assistant Performance section, you can view a chart tracking chat activity over time. You can filter it daily, weekly, and monthly. You can also opt for a more customized approach by specifying a particular date range to observe results within that specific timeframe.

Conversations Started & Conversations Resolved

This chart displays the number of chats initiated with the assistant. You can click on any data point on the chart to view the list of users who started chats on that specific day and those with resolved conversations.

Metrics

Total Conversation Sessions shows how many chat interactions users initiated with the AI Assistant during the selected timeframe.

Total Resolved Conversations reflects the number of chat interactions that are considered resolved.

Resolution Rate shows the rate of resolved conversations to the total conversations.Performance

Under the Assistant Performance section, you can view a chart tracking chat activity over time. You can filter it daily, weekly, and monthly. You can also opt for a more customized approach by specifying a particular date range to observe results within that specific timeframe.

By regularly analyzing the AI Assistant report, you can gain actionable insights to enhance its performance and improve user satisfaction.