Chart Design Rules That Make Spreadsheet Charts Readable

Most spreadsheet charts fail not technically but editorially: wrong type, sorted alphabetically instead of by value, decorated until unreadable. A handful of rules covers nearly every case you will meet in business reporting.

What you need

• A summarized table of data, ideally from a pivot
• Excel, Google Sheets, or LibreOffice Calc
• One question each chart must answer

Step by step

1. Pick the type by relationship, not novelty. Time trend: line. Comparison across categories: horizontal bar. Part-of-whole with few slices: pie, otherwise stacked bar. Correlation: scatter. Everything else defaults back to bar.
2. Sort categorical bars by value, descending, unless categories have intrinsic order like age bands. Alphabetical order buries the story mid-chart.
3. Truncate the value axis honestly. Start bars at zero; lines may start above zero when showing change, and say so in the subtitle if so.
4. Delete non-data ink: gridlines beyond three or four, chart borders, 3D effects, shadows, and legends when there is only one series. Label series directly at their endpoints instead of a detached legend where possible.
5. Order colors by meaning: highlight the one series that matters in a saturated color, mute all others in gray. Rainbow palettes imply distinctions that do not exist.
6. Add data labels only where precision matters, and then only on the points being discussed. Labels on every point of a 24-month line chart become noise.
7. Write the takeaway as the title: "Churn fell 12% after the pricing change" beats "Churn by month". The chart proves the sentence; readers should never reverse-engineer it.
8. Resize to actual placement. Charts squeezed beside text need bigger fonts and fewer series than full-width dashboard tiles; check legibility at final size, not edit size.
9. Keep source pivots or tables adjacent and clearly named so numbers can be traced when someone challenges a figure.

Common mistakes

• Pie charts with more than five slices or near-equal values, where angle differences become unreadable
• Dual axes with unrelated scales, inviting misinterpretation of correlation
• Default theme colors that give every series equal visual weight regardless of importance

Related guides

Summarize raw data first with [[build-pivot-table-from-scratch]], then chart the pivot output rather than the transactions. For dense numeric context around a chart, a [[conditional-formatting-heatmap]] often communicates better than adding another series.