Praxis is a big data service for data scientists, analysts, and engineers to explore and analyze SAP ABAP application log data. We developed the service to dramatically increase the availability and access to the data; improving time to insights by as much as 90%.
I conducted research with representative end users, and documented the results in insights boards. The insights boards helped communicate the key needs and perspectives of our internal customers. The biggest need was to make the service understandable for developers with an interest in data science.
High-Level User Flow
I developed a high-level user flow to articulate key design principles that needed particular attention, and to break down the key spatial and temporal experience considerations. This helped us focus on what was most important for our primary audience; being relevant but not overwhelming for non-data scientists.
Logo, Color, & Typography Exploration
We learned that our users would not try the big data service or take it seriously, if it wasn't both technically and aesthetically credible. Using the SAP Fiori iOS design system as a starting point, I created a visual identity to inspire confidence and encourage exploration.
Working closely with the engineering team and the product manager, I documented all user flows in detailed UI mockups using Sketch. I shared UI specs using Zeplin. It was important for our users to see and understand the status of a notebook and stop it quickly when necessary so I created simple, consistent semantic standards, and surfaced critical functionality to the home screen.
I developed a concept model to help articulate the high-level purpose and value of the initiative. It helped the team create a common language and coordinate our efforts. It was used for team and stakeholder communication. I iterated and refined it a number of times to highlight the core value and key components.
I created high level descriptions and scenarios so that the team could prioritize our roadmap, and focus on supporting the most critical functionality for users. One key research finding was how hard it was--for all but the most specialist technical user--to interpret the log data; help was needed.
From research we discovered that our users' first priority was a simple service for starting and managing data science notebook jobs. It was very important that users with limited data science experience could confidently begin to explore application log data.
The big data service needed administrative functionality to allow users with specific permissions the ability to easily stop services and check performance metrics. The admin view is only available to a smaller group of users, but for those managing multiple jobs and instances it's critical to see (and understand) statuses at a glance. Critical data is intentionally contrasted for preattentive processing.