Guide

Choosing the Right Chart: A Decision Guide for Analysts and PMs

Most dashboards are chart soup — pie charts where bars belong, line charts on categorical data, heatmaps with no scale legend. This guide gives you a fast decision tree.

The Decision Tree

• Comparing categories? → Bar chart (vertical for <7 items, horizontal for long labels)
• Trend over time? → Line chart
• Distribution of one variable? → Histogram (use our Histogram Generator)
• Relationship between two numeric variables? → Scatter plot
• Many-to-many relationships? → Heatmap (use our Correlation Heatmap Generator)
• Parts of a whole, ≤4 slices? → Pie or donut. Otherwise → bar.

The Universal Generator

For bar, line, and pie charts from any dataset, use our Chart Generator.

Common Anti-Patterns

1. 3D pie charts. Just don’t.
2. Truncated y-axes on bar charts. Exaggerates differences misleadingly.
3. Dual y-axes. Almost always create false correlations.
4. Rainbow color scales. Use sequential or diverging palettes.
5. Pie charts with >5 slices. Becomes unreadable — switch to bar.

Annotation > Decoration

The single biggest improvement to any chart is a one-sentence title that states the insight (“Mobile checkout drops 32% on iOS Safari”) instead of describing the data (“Conversion rate by browser”).

FAQs

What about stacked bars? Use only when total matters and segments are few (≤4).

When is a table better than a chart? When exact values matter or there are fewer than 5 data points.