The redesign they loved, and the problem it hid
Users loved the rebuilt telemetry explorer. The same study found a harder problem underneath: people weren’t confident they were asking their data the right question.
TL;DR
Customers had been asking for more room to see their data, and the rebuilt telemetry explorer gave it to them. Before launch, the team wanted to know one thing: is it better? A strong majority said yes. But when I read the interviews and the survey side by side, the same struggle kept showing up, at every company and every skill level. People liked the new layout, but they weren’t confident they were building the right query to answer their question, especially once it got complicated. So I built the readout around that gap, and a simple thumbs-up became the plan for the next phase: helping people get to the right question, not just a nicer view of the answer. The research was part of what led to today’s Gemini Cloud Assist; it gave early support to the idea of an AI assistant that helps bring different kinds of telemetry together during an incident.
The big picture
How a thumbs-up became a roadmap
The study was set up as a yes-or-no question, and the answer was yes. The redesign was a real improvement, and people called it cleaner, more responsive, and easy to learn. Stopping there would have made a tidy readout, but it wouldn’t have given the team a next step.
What the ratings were measuring
The satisfaction scores were mostly about the layout. What would make or break the product sat one layer down, in building the query itself. Most people also learned the tool by trial and error rather than documentation, so whatever help they needed had to live inside the tool.
A trick question
One survey item was a deliberate extreme: the tool is already perfect and can’t be improved. It’s a quality check borrowed from psychology, there to catch people who are just being nice. They disagreed (thankfully), which told me the positive ratings were real.
From finding to strategy
How a launch check became a roadmap
What I almost missed
The easy readout
The easy readout was one slide that said “Success, ship it,” and it would have been accurate. It just wouldn’t have told the team what to work on next. What changed my mind was reading the interviews and the survey together instead of separately. The same struggle showed up in both, at every skill level, and once I saw it, I couldn’t unsee it. People were rating the layout, while the real risk to the product was in the query.
“The hardest task is understanding the different metrics and analyzing the data to identify patterns and trends.”Cloud engineer, interview participant
Impact