Setup and Configuration
This guide will help you configure your analysis settings to get the most accurate and valuable insights from your data.
AI Insights will not begin collecting or analyzing data until the setup is complete. If no configuration exists, insights will remain inactive and no data will be processed. Setup must be completed separately for each bot. Bots can have their own unique set of topics or share the same topics, but each bot requires its own configuration to activate insights.
Analyzer Settings
The Analyzer Settings section allows you to choose which metrics AI Insights will analyze. Disabling a metric will completely exclude it from analysis and hide the graphs containing it. By default, all metrics are enabled.
Resolution When enabled, AI Insights will analyze how issues are resolved in your data. This helps you track the effectiveness of your resolutions and improve overall service quality.
Sentiment This setting allows AI Insights to assess the tone and emotional content of your interactions. By identifying positive, neutral, or negative sentiments, you can better understand customer satisfaction and address areas that need improvement.
Topic The topic analyzer identifies and categorizes the main subjects discussed in your interactions. This helps you track trending topics, recurring issues, and areas of interest for continuous improvement.
Topic Settings
Under Topic Settings, you can define the specific topics you want AI Insights to identify and analyze. For each topic, you can set:
Topic Name A clear and descriptive label for the topic (e.g., "Account Management").
Description A brief explanation that defines the scope of the topic (e.g., "Includes issues related to account logins, profile updates, and password resets").
Example Phrase A sample customer question or statement that fits the topic (e.g., "How do I reset my password?").
Saving Your Settings
After configuring your analyzer and topic settings, click the Save button at the top right of the page to apply your changes.
Best Practices for Selecting Topics
To ensure accurate and meaningful analysis, follow these best practices when creating and defining topics:
Keep topics distinct and separate:
Choose topics that are clearly different from each other to avoid overlap and confusion. Avoid broad or ambiguous categories, as well as topics that are too similar and could be interpreted as the same thing. Specific topics help the AI classify conversations more accurately.
Examples:
Instead of using a single, broad topic like "Billing," break it down into more specific topics such as "Invoice Queries" and "Payment Disputes."
Avoid creating two topics that are nearly identical, such as "Billing" and "Invoicing." Instead, combine them under a well-defined topic or separate them by distinct use cases, like "Invoice Requests" and "Payment Processing."
Having one "Other" or "General" topic intended for chats that do not fall under any other topic is recommended since if no topics are determined, no topic will be set.
Write clear descriptions:
Provide a concise and precise description for every topic. Make sure the description clearly outlines what should and should not be included.
Example: For "Order Status," use "Questions about order tracking, delays, or cancellations."
Include representative example phrases:
Add one or more sample phrases that demonstrates the type of inquiry or issue belonging to the topic. Use examples that customers are likely to use in real conversations. You can add more than one example question in the same example field. Usually one or two examples are sufficient.
Example: For "Shipping Issues", examples might include: "My package is missing", "I received the wrong item."
Review and update:
Periodically check your topics, descriptions, and examples to ensure they remain relevant. Update them as your business needs and customer interactions evolve.
Last updated
Was this helpful?