How to Choose the Right Chart for Your Data

Updated · 2 min read

Start with the question your audience needs answered, not with the chart type. Most questions fall into one of six families.

Your questionGood charts
Compare: which is bigger?Bar, Radar, Pictograph, Pareto
Composition: what is it made of?Pie, Doughnut, Stacked bar, Treemap, Waterfall
Trend: how did it change?Line, Area, Combination
Distribution: how are values spread?Histogram, Box plot, Population pyramid, Heatmap
Relationship: are these linked?Scatter, Bubble, Heatmap
Flow and process: where does it go?Sankey, Funnel, Waterfall
Schedule and history: when?Gantt, Timeline
Single KPI: are we on target?Gauge, Doughnut

Three follow-up questions

  1. How many categories? Pies and doughnuts handle a few; bars, treemaps and heatmaps handle many.
  2. How precise must the reading be? Position and length (bars, lines, dots) are read most accurately; angle, area and color less so.
  3. Who is the audience? Familiar charts (bar, line, pie) need no explanation. Box plots, Sankeys and radars may need a one-line guide in the subtitle.

Example: one dataset, three questions

Take monthly sales for three product lines. “Which line sold most in total?” calls for a sorted bar chart. “How did each line trend?” calls for a line graph. “How did the mix change?” calls for a 100% stacked area chart. Same data, three different charts.

How did the mix change?Area chart with 7 categories and 3 data series (Solar, Wind, Hydro). Values range from 12 to 61.How did the mix change?100% stacked areaSolarWindHydro20182019202020212022202320240%25%50%75%100%TWh
A 100% stacked area chart answers a “composition over time” question.

Tools used in this guide

More guides