KPI Dashboard Design
Design effective dashboards that communicate key metrics clearly.
KPI Selection Framework
Good KPIs are:
- Relevant to business goals
- Measurable and quantifiable
- Influenced by the team
- Updated frequently
- Simple to understand
Common Business KPIs
| Goal | KPIs |
|---|---|
| Revenue | MRR, ARR, Revenue Growth |
| Acquisition | CAC, New Users, Conversion Rate |
| Retention | Churn Rate, NPS, DAU/MAU |
| Efficiency | LTV:CAC, Burn Rate |
| Quality | Error Rate, Response Time |
Dashboard Layout
┌─────────────────────────────────────────────────┐
│ Executive Summary │
│ [Revenue ▲12%] [Users ▲8%] [Churn ▼2%] │
├─────────────────────┬───────────────────────────┤
│ │ │
│ Revenue Trend │ User Acquisition │
│ (Line Chart) │ (Bar Chart) │
│ │ │
├─────────────────────┼───────────────────────────┤
│ │ │
│ Retention Funnel │ Top Products │
│ (Funnel Chart) │ (Table) │
│ │ │
└─────────────────────┴───────────────────────────┘
Visual Design Principles
/* Traffic light colors for status */
.metric-good { color: #22c55e; } /* Green */
.metric-warning { color: #f59e0b; } /* Amber */
.metric-bad { color: #ef4444; } /* Red */
/* Visual hierarchy */
.metric-primary {
font-size: 2.5rem;
font-weight: 700;
}
.metric-secondary {
font-size: 1.5rem;
font-weight: 500;
}
Chart Selection
| Data Type | Chart |
|---|---|
| Trend over time | Line chart |
| Comparison | Bar chart |
| Composition | Pie/Donut |
| Distribution | Histogram |
| Correlation | Scatter plot |
| Funnel stages | Funnel chart |
Interactivity Features
- Drill-down to detailed views
- Date range selection
- Filtering by segment
- Export to CSV/PDF
- Scheduled email reports
Best Practices
- Limit to 5-7 KPIs per dashboard
- Show trends, not just snapshots
- Use consistent color coding
- Include comparison periods (vs last month)
- Update data in real-time or hourly
- Review dashboard relevance quarterly
Common Mistakes
- Too many metrics (information overload)
- No clear visual hierarchy
- Missing context (no comparisons)
- Outdated or stale data
- Metrics without ownership