Good data visualisation starts with one clear message per chart, uses the simplest chart that fits the data (bars for comparisons, lines for trends), removes clutter, uses colour only to highlight what matters, labels directly and clearly, starts bar axes at zero, and checks the chart cannot mislead. The title should state the insight.
Start with the message
Before choosing a chart, write the sentence you want the reader to take away: "Enquiries from WhatsApp doubled after we added a booking link." Then design the chart to make that sentence obvious.
Choose the right chart
| Purpose | Best chart |
|---|---|
| Compare categories | Bar chart (horizontal for long labels) |
| Show change over time | Line chart |
| Show parts of a whole (few parts) | Stacked bar, or a simple pie with 2 to 4 slices |
| Show relationship between two variables | Scatter plot |
| Show distribution | Histogram or box plot |
| Show a single key number | Large number with context |
Reduce clutter
Remove unnecessary gridlines, borders, 3D effects, shadows and decorative images. Every element should help understanding.
Use colour with purpose
- Use neutral colours for most data and one strong colour to highlight the key point.
- Keep colours consistent across charts.
- Ensure enough contrast and do not rely on colour alone, for readers with colour vision deficiency.
Label clearly
- Write titles that state the insight, not just the topic: "Revenue grew 18% in Q3" rather than "Revenue by quarter".
- Label lines and bars directly where possible instead of using distant legends.
- Include units and time periods.
Avoid misleading charts
- Start bar chart axes at zero.
- Use consistent scales when comparing charts.
- Show uncertainty where relevant.
- Do not cherry-pick time periods to exaggerate trends.
Tell a story
In reports and dashboards, order charts logically: overview, then detail, then recommendation. Add a short sentence of interpretation under important charts.
These principles are part of our data analytics course and apply in Excel, Power BI and Python. See also Power BI for beginners.
Frequently asked questions
Are pie charts bad?
Not always, but they are hard to read with many slices. Bar charts are usually clearer.
How many charts should a dashboard have?
As few as needed to answer the main questions, often four to eight.
Which tool is best for visualisation?
Excel, Power BI, Tableau and Python libraries can all make excellent charts. Principles matter more than tools.
