UniDash
A dashboard that explains itself, instead of demanding interpretation.
01 / The Friction & Problem
Most dashboards show walls of numbers, colorful sparklines, and disconnected bar charts. Few explain them. Operators spend hours deciphering what anomalous spikes mean and debating causation before taking action. UniDash was an exploration of how a data product could narrate its own state — surfacing what changed, why it changed, and what to do about it.
02 / The Core Concept
A modular, narrative-first data workspace where every visualization is paired with a concise natural-language explainer: the change delta, the probable root cause, and a recommended tactical next step.
03 / Research & Process
- Audit of 12 existing enterprise telemetry and analytics dashboards
- Card-sorting and information hierarchy exercise with 14 data analysts and team leads
- Component library built specifically for 'self-explaining' data widgets
- High-fidelity interactive prototype with realistic time-series mutations
- Comparative usability sessions measuring decision-making speed and comprehension confidence
04 / Key Design Decisions
- Widget taxonomy divided into 4 clear cognitive types: Status, Trend, Anomaly, and Forecast.
- Each widget pairs a high-contrast micro-chart with a 1-sentence narrative verdict.
- Inline actions ('Why?', 'What changed?', 'Simulate rollback') that unfold a 3-bullet breakdown.
- A 'Daily Brief' morning view that distills twenty operational metrics down to the top 3 requiring human intervention.
Reflection & What Was Learned
Data products should optimize for time-to-decision, not time-to-insight. An insightful chart without operational direction is merely a decoration, not a tool.