Customer Kernel testing

What is Kernel testing? 

We take customer questions and search through their content to see if there are any relevant pieces of information (Kernels) that match. If the system finds relevant Kernels, the report will show where the answers would come from. If relevant Kernels are missing, this indicates a gap in the content that needs to be filled. 

Perform a Kernel test

When gathering questions for Kernel testing, ensure that the questions accurately represent current user interactions and expectations. This allows you to create a comprehensive and representative set of questions for Kernel testing, ensuring a thorough evaluation of the ExpertAI system.

To formulate the initial set of 50 to 100 questions for testing: 

• Leverage reporting data: Identify the most frequently asked questions and common user interactions. This data-driven approach ensures that the questions are representative of real-world usage. 
• Reflect user expectations: Formulate questions that align with how users interact with the content, and their expectations. This helps assess the ability to meet user needs. 
• Include future state questions: Incorporate questions that anticipate future needs and potential new subject matter areas. This forward-thinking approach ensures the system can grow and adapt over time. 
• Direct audience engagement: Engage directly with your audience by working with colleagues or conducting surveys. This can provide valuable insights into the most frequently asked questions and common inbounds they handle. 
• Review onboarding materials: Examine the materials provided to new onboarding agents, including the questions they are expected to answer. This helps identify foundational questions that are crucial for new users.