dashboard-redesign

Dashboard Redesign

Design a decision instrument, not a wall of metrics. Preserve useful density, expose the comparisons that matter, and make data quality and scope legible.

1. Establish the decision model

Identify:

• users: primary role, expertise, frequency, environment, devices, display duration, accessibility, and export needs;
• decisions: required actions, monitoring cadence, acceptable time to notice, and the workflow from signal through investigation to action;
• data: metric definitions, owners, sources, freshness, and latency;
• context: required comparisons, targets, thresholds, and segments.

If these are unknown, state assumptions and mark metric semantics for owner validation. Never infer that a visible number is a KPI.

Checkpoint: State assumptions, have the data owner validate metric meaning and thresholds, revise the decision matrix and affected representations, then re-check until every unresolved semantic question is explicitly marked for decision.

Classify the surface using references/dashboard-models.md. A monitoring board, executive scorecard, operational queue, and analytical workspace need different density and interaction.

2. Inventory decisions before widgets

Build a compact matrix:

Decision | Signal | Comparison | Grain | Freshness | Action | Failure cost

Synthetic example: Which route needs intervention? | Risk state | Operations threshold and shift target | Route | Per vehicle | Open incident | Missed service

Remove or demote data that does not support a scoped decision, diagnosis, or required reporting obligation. Identify missing context before selecting chart types.

Checkpoint: Every row ends in a real decision or reporting obligation; remove the rest.

3. Create the information architecture

Organize the page in this order when it matches the task:

1. scope and data state;
2. conditions requiring attention;
3. core outcome and comparison;
4. causes or contributing dimensions;
5. records, evidence, and actions.