SaaS AI Dashboard
Lead designer contributing to several observability projects for an enterprise cloud data services dashboard used by thousands of businesses globally.
This is a reproduction of a 0-1 dashboard I worked to prioritize and streamline with 6 PMs, 1 UXD, and 9 Devs to scope out the MVP and deliverables within 3 weeks to present to leadership.
Client
Google
Launched
2024
Role
Senior UXD
Partners
1 UXD + 6 PM + 3 Eng
Define
Problem Statement:
How might we streamline a troubleshooting dashboard for data professionals to find and solve observability problems quickly?
Dataproc is a complex Google Cloud platform that data professionals use to monitor when systems are down.
As Dataproc grows, dashboards become more complex. Dashboards may include multiple filtering and hard to understand data viz.
As workloads increase, Dataproc customers’ overall costs may spike.
Feature Prioritization
In a brainstorming session, I scoped initial MVP features of Dr. Proc with 1 UXD and 1PM to focus on a dashboard showing a central summary of troubleshooting. This solution would help our users minimize information access cost.
We focused on our user: a data professional’s troubleshooting journey. Pillars I identified were: What happened, Why this happened, and How can I fix this?
Ideation & Iteration
Here are initial low fidelity wireframes. These were created quickly to visualize our user journey.
I created 2 versions to A/B test with 5 real users (data professionals) from a large retail client.
Insights from our user tests directed us to iterate towards a simpler, numeric data scorecard dashboard designl
Iterating for Gemini
5 PMs were added to my pod after MVP and had multiple ideas to integrate Gemini AI. I drove UX strategy towards:
-Focusing on 1 clear way to integrate AI in the batches
-Using icons to distinguish the AI feature
-Breaking out AI troubleshooting into multiple steps and requesting feedback