Chart Pattern · Reference
Scatter with Outliers
Two-variable relationship · regression + outlier highlight
Customer lifetime value vs onboarding time
Each dot = one customer · regression line shows central tendency · red dots = statistical outliers (residual > 2σ)
✓ When to use
- Showing relationship between two continuous variables
- Spotting outliers that need investigation
- Communicating correlation strength visually
✗ Avoid when
- Lots of overlap (>500 points) — switch to density heatmap
- Categorical x-axis — use box plot or strip plot
- Showing time — use a line, not a scatter
★ Best practice
- Add a regression line to show central trend
- Highlight outliers in red; everyone else in muted blue
- Label only the most extreme outliers — the chart is a guide, not a directory
- Always start the y-axis where the data lives, not at zero