Case study: Rewriting a SaaS analytics dashboard to reduce cognitive load

Tech · 5 min read

Background: MetricsLab is a 120-person SaaS that provides marketing analytics to mid-market teams. Users struggled with the legacy dashboard: too many widgets, inconsistent color semantics, and unclear data provenance. Product analytics showed 42% of new users never created a dashboard within the first week, and weekly active users plateaued.

Design process: Over 10 weeks the cross-functional team ran contextual interviews (n=18), card sorts, and tree tests to rationalize the IA. The team reduced widgets per dashboard from an average of 14 to a recommended 5 and introduced a “dashboard recipe” library with templates mapped to common jobs-to-be-done. They created a lightweight visual language for KPI semantics (positive/negative/neutral) and added hover states that revealed data provenance and last-updated timestamps.

Outcomes: A controlled rollout to 30% of accounts resulted in a 24% increase in dashboard creation, 12% lift in weekly active users, and a 14% reduction in support tickets about “confusing metrics.” The team emphasized that small motion and microcopy changes — contextual help, explainer tooltips, and consistent color semantics — delivered disproportionate improvements to comprehension and retention.