Amey-Thakur Spreadsheet charts
Default chart settings produce cluttered, hard-to-read output. A chart should make one comparison obvious, and most improvement comes from removing elements rather than adding them.
Method
- Choose by the question. Change over time as a line, category comparison as bars, composition rarely as a pie, and correlation as a scatter (see data-visualization).
- Sort bars by value. Alphabetical ordering hides the ranking that is usually the point.
- Start value axes at zero for bars. A truncated axis exaggerates differences and is misleading rather than emphatic.
- Remove decoration. Gridlines, borders, shadows, and legends that duplicate labels all compete with the data.
- Label directly where possible. Series labelled at the line beats a legend the reader must map back and forth.
- Title with the finding. Sales fell after the price change is more useful than Monthly Sales (see slide-deck-creation).
- Check it works without colour. Colour-only distinctions fail for many readers and in print (see color-contrast).
Boundaries
Charts show patterns and hide detail, which is their purpose and their risk. Spreadsheet charting is limited compared to dedicated tools for complex visualisation. A chart cannot make a weak finding strong (see data-storytelling).