Auto-Diagnosis Configuration

Microsoft

Reduced setup friction and configuration time for Auto Diagnosis by designing a scalable, guided wizard that eliminates a complex manual support dependency, saving users time with reduced misconfiguration errors.

Co-designed the configuration with Product Managers and Developers and quickly created Figma prototypes to gather quantitative data that informed the final model’s success. 

Client

Microsoft Inc., Azure

Role

Senior User Experience Designer

Location

On-sight, remote

Scope

Create the first self-guided Auto-diagnosis configuration UX, to liberate Service owners from relying on Microsoft for a complex, and time consuming, manual process.

Challenge

Brain Auto-diagnosis is key to helping Service owners quickly mitigate incidents with automated Root Cause Analysis (RCA) insights. However, it relied on manual 1:1 support from the Brain team, requiring multiple Developer and Product Manager cycles (1-2 weeks) to properly set up the configuration. As customer demand grew, creating a self-serve, guided configuration experience became a top priority.

I began by working with subject matter experts in Auto-diagnosis, a Product Manager and lead Developer, along with a User Researcher, to interview Service owners and uncover their needs. Users requested several features, including:  

  • A summary of correlated auto-diagnosis insights.

  • The ability to define and validate multiple configurations. 

  • A way to provide feedback that the auto-diagnosis “insights” are of good or bad quality. 

In response to Brain leadership feedback to further streamline the experience, I reduced the setup to two main steps, using the service’s existing profile to guide diagnostic type selection. To manage remaining complexity, I added collapsible sub-steps with chevrons, allowing users to hide irrelevant sections. This version tested extremely well, but we still felt there was room for improvement.  

In the final design phase, I moved to high-fidelity mockups and streamlined the flow to a “One click to success” model, removing the “Review and Create” step. The new design auto-selected Platform insights, provided the ability to add new monitors, included easy-to-Service insights, and featured a robust preview with simple feedback tools to build user confidence without time-wasting external validation.

I also added the capability to validate a sample query directly in the “+ Add log insight” context pane, further ensuring confidence and saving even more time for users. The result of all the changes was a faster, clearer experience that users could easily navigate and enthusiastically approved. 

Action

I began by creating a series of wireframe mockups that we could validate with users. My goal was to break the complicated configuration process into a manageable, step-by-step “Wizard” pattern, intended to avoid overwhelming the user with too much information all at once.

Early usability testing confirmed that users could easily navigate the UX and successfully create a configuration with no guidance.  

Result

Usability studies confirmed: The Auto-diagnosis configuration reduced setup friction and configuration time by providing a scalable, guided experience that eliminates a complex manual support dependency while saving users time with reduced misconfiguration errors.