Design a dashboard people will actually use

Role: Act as a BI analyst designing a dashboard. Context: Audience and the decisions they make: [AUDIENCE_DECISIONS]. The questions the dashboard must answer, in priority order: [QUESTIONS]. Metrics and their definitions: [METRICS]. Data grain, history and refresh: [DATA]. Comparison needs (vs target, prior period, segment): [COMPARISONS]. Tool and any design system: [TOOL]. Constraints (screen size, load time, accessibility): [CONSTRAINTS]. Task: Produce: the audience and top tasks; a question-to-metric-to-visual map (for each question: the metric, the chart type and why it suits the question, the comparison shown, and the grain); a layout (what is above the fold, the reading order, grouping); the interactions (filters, drill paths) and their defaults; and the accessibility considerations (colour, labels, alternatives to colour-coding). Format: Sections: Audience & tasks | Question -> metric -> visual (table) | Layout | Interactions & defaults | Accessibility. ~500 words. Verification: Flag any chart choice that could mislead (dual axes, truncated axis, pie with many slices, 3D) and any question with no clear metric. Every tile must map to a listed question. Validation: Test with the top decision the audience makes: does the layout put the answer in front of them fast? Reorder if the key tile is buried.