Skip to main content

Data & Analysis

Overview​

The Data section provides comprehensive visibility into how users interact with your AI Agent. Here, you can view every question users have asked and see the answers provided by the Agent—empowering you to improve both agent performance and user satisfaction.

Accessing the Data Section​

  • From the Dashboard, click on Data in the left-hand navigation menu.
  • The main Data view displays a list of user questions and the corresponding answers from the AI Agent, along with the date of each interaction.

Reviewing and Analyzing Data​

  • Each entry displays:

    • Question: What the user asked.
    • Answer: The AI’s response, i.e. the answer to the question. It can include any embedded control tokens such as [control agentDataId="..." type="carousel"] or [form, keyName="6356..."]. The data for these controls are passed to the bot for rendering.
    • Date: When the interaction took place.
    • Delete Option: Remove specific data entries if needed.

  • Analyze Button: Click Analyze on any answer to open a dialog where you can describe your concern about the AI’s response. After submitting your analysis prompt, the system will review all context—including the question, answer, tools, and system prompt—to provide actionable suggestions for improvement.

    Analyze Data Dialog

Interaction Overview​

  • User interactions are listed chronologically.

  • Each question/answer pair can be expanded for details or collapsed for easy browsing.

  • Feedback options are available for each entry:

    • Thumbs Up: Indicates the response was satisfactory.
    • Thumbs Down: Marks answers needing improvement.
    • Not Answered: Highlights questions the AI could not answer.

Searching and Filtering Data​

  • Search Bar: Quickly find specific questions or topics.

  • Filter Options:

    • Sort data by feedback (e.g., thumbs down, unanswered).
    • Filter by data source (e.g., web, mobile, test drive) using source keys, the API key
    • Filter by date.

    Data Filters

Utilizing User Feedback​

  • Positive Feedback (Thumbs Up): Confirms helpful answers and builds confidence in the Agent.
  • Negative Feedback (Thumbs Down): Identifies gaps or issues that require attention.
  • Unanswered Questions: Reveal knowledge gaps and opportunities for content or prompt improvements.

Exporting Data​

  • Select data using filters (e.g., by feedback or source) and export for deeper analysis.
  • Exporting thumbs down or unanswered data can help focus improvement efforts and streamline reviews.

Best Practices for Data-Driven AI Improvement​

Content Review and Prompt Tuning​

  • Regularly review user questions and Agent responses to ensure quality and relevance.
  • Update the Knowledge Base and adjust system prompts based on feedback and trends.

SEO and Content Strategy​

  • Analyze user questions to identify missing or unclear content.
  • Align your site’s SEO content with actual user queries to improve both AI accuracy and search engine visibility.
  • Use insights from unanswered or poorly answered questions to enhance your FAQ or documentation.

The Data & Analysis section is your central hub for monitoring, refining, and enhancing your AI Agent. By leveraging real user interactions and feedback, you can continuously improve both the AI’s performance and your digital experience.

Let me know if you want it even more concise or if you want to add examples or screenshots!