Meridian Product Design

Introduction

Meridian shipped a dashboard with forty-one widgets and a support queue full of people asking where to find things. Research showed most sessions were spent answering one of three recurring questions.

Watching Real Sessions

We sat with fourteen users across four account types and recorded what they did rather than what they said they did. The same three questions came up in every account type, and the path to each of them ran through at least two screens that existed for other reasons.

Rebuilding the Default View

We rebuilt the default view around those three, moved everything else behind an explicit "explore" mode, and let teams pin what they actually used.

• The default answers the three questions above the fold, with no configuration required on first login.
• Everything removed is still reachable, so no workflow was broken by the change.
• Pinning is per team rather than per user, so a shared view stays shared.

Testing Before Building

Each round went in front of users as a clickable prototype before any of it reached engineering, which is where most of the wrong turns were caught.

• Three rounds of five participants, each on a task rather than an opinion.
• Time to first answer was the measure, not preference between layouts.
• The one direction that tested badly was dropped in week two rather than after a build.

Conclusion

Time to first answer dropped from minutes to seconds in testing, and the widget count in the default view went from forty-one to six.