Pick the chart that fits your question, build it from a summary table, and label it so nobody can misread it.
Why it matters
A good chart lets a manager, client or colleague see the answer in five seconds. A bad one hides it or, worse, suggests something false. Choosing the chart type and labelling it properly matter more than colours or 3D effects.
Start from the question
Decide what you want the reader to compare, then pick the chart:
| Question | Chart |
|---|---|
| Compare amounts across categories | Bar or column chart |
| Show change over time | Line chart (or columns for a few periods) |
| Show the relationship between two measurements | Scatter chart |
| Show how values are spread out | Histogram |
| Show parts of a whole (2 to 4 parts, adding to 100%) | Pie or, usually better, a stacked bar or plain bar |
When in doubt, use a bar chart. People compare lengths along a common baseline far more accurately than angles or areas, which is why pie charts with many slices are hard to read.
Build from a summary, not raw rows
Chart a small summary table (from a pivot table or SUMIF), not 500 raw rows. For spend by category:
| Category | Total |
|---|---|
| Consumables | 349.80 |
| Safety | 271.00 |
| Office | 161.00 |
Sort it from largest to smallest first so the chart reads in order.
Then select the summary (including headers) and:
- Google Sheets: Insert > Chart; change the type and labels in the Chart editor (Setup and Customize tabs).
- LibreOffice Calc: Insert > Chart; the wizard walks through type, data range, series and titles. Double-click the chart later to edit.
- Excel: Insert > Recommended Charts or pick a type; use the + button next to the chart for titles and labels.
Label it properly
A chart should make sense if someone sees it on its own:
- A title that states the point, not just the topic: "Consumables were the largest cost in September 2026" beats "Chart 1".
- Axis titles with units: "Spend (USD)", "Temperature (°C)", "Week".
- Start bar and column axes at zero. A bar's length is the value; starting at 150 makes 161 look a tenth of 349. Line charts may use a narrower range, but say so.
- Data labels on a few bars can replace gridlines; do not label everything on a busy chart.
- Remove clutter: no 3D, no shadows, no unnecessary legend (one series needs no legend), no more decimals than the data justifies.
- Colour with purpose: one colour for all bars, plus one highlight colour for the bar you are talking about. Do not rely on red versus green alone, because many people cannot tell them apart.
- Name the source in a note: "Source: supply orders sheet, 10 orders, Sept 2026."
Worked example: same data, two charts
A 3D pie of the three categories with a legend makes the reader match colours to labels and judge tilted slices; Safety and Consumables look nearly equal. A sorted horizontal bar chart with the title "Consumables were the largest cost in September 2026", an x-axis titled "Spend (USD)" starting at 0, and a data label on each bar shows immediately that Consumables (349.80) is about twice Office (161.00).
For a time question such as "Is spending rising?", you would add a Month column, summarise by month, and draw a line chart with months in order on the horizontal axis.
Common mistakes
- Choosing a chart because it looks impressive rather than because it answers the question.
- Charting raw rows, producing hundreds of unlabelled bars.
- Bar axes that do not start at zero.
- Unsorted categories and default titles like "Total vs Category".
- Line charts for categories that have no order (a line implies a trend between points).
Quick recap
- Pick the chart from the question: bars for comparing, lines for time, scatter for relationships, histograms for spread.
- Chart a sorted summary table, not raw data.
- Title the point, label axes with units, start bars at zero, cut clutter, colour with purpose, cite the source.
Try it
Open the Do tab to build and label two charts from your data.