A good chart makes one point obvious. A poor one hides it or misleads.
Start With the Message
Decide what the reader should take away — "sales fell in Q3 in the north only" — and design the chart around that.
Choose the Right Chart
- Comparison across categories: bar chart.
- Change over time: line chart.
- Distribution: histogram or box plot.
- Relationship between two numbers: scatter plot.
- Part of a whole: stacked bar; pie charts only for a few parts.
- Exact values: a table.
Be Honest With Scales
- Start bar chart axes at zero.
- Avoid dual axes that imply relationships.
- Use consistent scales when comparing charts.
- Don't use 3D effects that distort size.
Reduce Clutter
Remove unnecessary gridlines, borders and legends; label lines directly; round numbers sensibly.
Use Colour Deliberately
Use grey for context and a strong colour for the point you're making. Ensure colours are distinguishable for people with colour-vision deficiencies, and don't rely on colour alone.
Titles That Say Something
"Northern sales fell 18% in Q3" is more useful than "Sales by region".
Show Uncertainty
Error bars, ranges or shaded intervals prevent readers over-interpreting small differences.
Test It
Show the chart to someone unfamiliar with the data and ask what it says. If they don't get the point, simplify.