Chart Selection: Match Purpose, Not Looks

THE MENTAL MODEL: The purpose of your visualization dictates the chart type. Don't pick a chart because it looks interesting or is the default in your tool. Instead, start by asking: "What question am I trying to answer with this data?" The answer will point you to the right family of charts.

HOW IT WORKS: Chart types can be grouped by the question they answer. To compare values, use bar or lollipop charts. To show trends over time, use line or area charts. To illustrate the composition of a whole, use pie, donut, or stacked bar charts. For finding relationships between variables, a scatter plot is ideal. For showing network connections, use a network or tree diagram. And for location-based data, use a choropleth map. Choosing from the right category first dramatically increases the clarity of your message.

WHEN TO USE IT: This framework is the first step in any data visualization task, from building a real-time monitoring dashboard to preparing a slide deck. A clear purpose ensures your audience understands the insight immediately. For example, to show quarterly sales figures for three different products, a grouped bar chart (Comparison) is a clear choice. To show how total user count has changed over the year, a line chart (Trend) is best.

WHEN NOT TO USE IT: The goal is always communication. Avoid picking a chart type that your audience is unfamiliar with unless you provide clear explanations. Also, don't follow categories so rigidly that you ignore the data's specific shape. For instance, while a line chart is for trends, too many lines create an unreadable "spaghetti plot." The most common anti-pattern is choosing a chart that looks visually appealing but distorts the data, such as using 3D effects, which makes accurate comparisons nearly impossible.

ONE CANONICAL EXAMPLE: Imagine you have data on website traffic. To show total users per month over the last year, you'd use a Line Chart (Trend). To show the breakdown of traffic sources like Organic, Direct, and Social for the most recent month, you'd use a Pie or Donut Chart (Part-to-whole). To compare the conversion rates of those different traffic sources, you'd use a Bar Chart (Comparison). Each question requires a different chart.