Overview
Chats Insights provides all the essential metrics to evaluate the performance of your Widget conversations. It serves as a central hub where you can view all your conversations in one place. With advanced search and sorting features, you can easily find specific conversations, such as those with CSAT responses, ones that utilized particular workflows, or chats that include specific keywords from either the user or the Widget.
Access Insights Overview
Click Solve > Widget Insights > Chats.
Top Level Metrics
The top-level metrics show you the key performance metrics of Widget conversations during the pre-selected time window.
-
Deflections
Counts chat sessions where the user was not transferred to a human agent (handoff) or did not issue a ticket. Even if a single chat session involved multiple handoffs or generated several tickets, it counts as one non-deflected chat. -
CSAT
The average Customer Satisfaction (CSAT) scores from chats. -
Quick Feedback
Direct feedback from users about the usefulness of the article when it was surfaced. -
Relevance (only visible when Discover is enabled)
Uses AI to rate how well the response addresses the user's inquiry. Each deflected chat receives one relevance rating. Non-deflected chats and some historical chats won’t have a relevance rating. -
Engagement (only visible when Discover is enabled)
Uses AI to determine if the user stayed engaged throughout the chat or dropped off midway.
Search by Chat ID, Query, or Chat content
Use the dropdown to select either Chat ID, Query (user query), or Chat, then type in the search field to find the conversation you're looking for.
Sort by various attributes
With the filter function, you can select multiple attributes to locate the conversations you're interested in quickly. The available attributes include workflow types (e.g., classic, autoflows), specific workflows used, relevance score, relevance, user engagement, CSAT, and whether the conversation includes user queries.
Details View - Intent Statistics
Clicking on either row of the chats table will slide over the detail view, which conveniently displays fundamental workflow-level statistics.
Details View - Create a new intent
You can use the button to create a new intent. When you click it, the current user query will be added automatically. However, you still need to include other training phrases and save the intent.
Details View - Add a user query to an existing intent
You can also add the current user query to an existing intent. Simply select the intent and click "assign". This will automatically add the user query to the training phrase list associated with the selected intent.
Important: Assigning the query to a new intent does not remove it from any previous workflows. To avoid conflicts between workflows, it is highly recommended that you manually remove the user query from the old intent.
Details View - Full Chat Transcript
To view the full chat transcript between the user and the bot, simply scroll down in the details view.
Advanced Filters
The advanced filter functionality within Chat Insights provides a powerful tool for in-depth analysis of past conversations between your AI agent and end-users. This feature allows for a more granular understanding of these interactions by enabling you to filter information based on various criteria, including:
- Chat ID: Isolate and examine specific conversations using their unique identifiers.
- Query: Filter by the specific text input provided by the end-user to the AI agent.
- Chat: Search and analyze the full content of individual chat sessions.
- Channels: Filter by the specific channel used for the interaction (e.g., widget, email, and Slack).
- Tags: Use tags associated with chats to categorize and filter conversations based on predefined criteria.
- Context Variables: Filter conversations using specific context variables. The supported Context variable (CV) types are Short Text, List, Email, Phone, Long Text, Date, Checkbox, and Number.
Additional Filtering Options
Beyond these basic filtering criteria, the advanced filter also supports a range of additional options to refine your insights further:
- All Workflow Types: Filter by the specific type of workflow associated with the conversation (e.g., AutoFlows, and Classic).
- All Workflows: Filter by the specific workflow instance associated with the chat.
- Relevance: Analyze conversations based on their relevance to specific topics or intents (e.g., Relevant, Somewhat relevant, and Irrelevant).
- User Engagement: Filter chats based on whether the end-user was engaged during their interaction with the AI agent.
- CSAT (Customer Satisfaction): Filter by customer satisfaction scores associated with the conversation to identify areas for improvement in customer service.
- Query: Analyze conversations based on whether there was a user query.
- Deflection: Filter conversations based on whether the conversation was deflected.