Nimbus — analytics that answer questions, not just show data

Analytics tools drown teams in widgets. Users open the dashboard, can't quickly find the metric that matters, and end up exporting everything to a spreadsheet to actually make sense of it — which defeats the product's purpose.

• ◎Hierarchy over densitySurface the 4–5 KPIs that matter; demote the rest.
• 📈Trend at a glanceEvery KPI shows change vs. last period, not just a number.
• ▤Scannable tablesStatus you can read in one pass — no interpretation required.

Target metrics: time-to-find KPI < 5s · table-scan task success > 90% · SUS > 80 · reduce “export to Excel” reliance.

Key findings

• 1A handful of metrics carry the weightRevenue, active users, orders and conversion — everything else is situational noise.
• 2Numbers need context“Is that good?” — users want comparison and trend, not an isolated figure.
• 3Tables are where ops livesAnalysts need status (paid / pending / refunded) legible instantly.
• 4Mobile check-ins are realLeaders glance at numbers on their phone between meetings.

Structure-first wireframes established a clear reading order: sidebar → KPI row → charts → table.

A calm, hierarchy-driven dashboard: four KPI cards with delta + sparkline, a focused revenue trend with period toggle, a traffic donut, and an orders table where status reads at a glance — plus a mobile companion for on-the-go checks.

Key design decisions

• ◆KPI cards with contextValue + delta chip + sparkline answers “what & is it good?” together.
• ◆Period toggle on the chartWeek / Month / Year keeps the view focused instead of five charts at once.
• ◆Status badges (colour + dot + label)Scannable and accessible without relying on colour alone.

6 participants (product & ops roles), moderated task-based sessions.

Tasks: (1) “Is revenue up or down this week, and by how much?” (2) “Find an order that's pending.”

What testing changed