Data is Not Information
We have more data than ever before, but we are starved for insight. Sticking a standard Chart.js library into a dashboard and feeding it a JSON array doesn't make you data-driven; it makes you confusing.
The goal of data visualization is cognitive offloading. The user shouldn't have to do math in their head to understand a trend.
Principle 1: Kill the Pie Chart
Pie charts are almost always the wrong choice. Humans are terrible at comparing the relative area of angles. Unless you are showing exactly two variables (e.g., Yes vs. No) where one drastically outweighs the other, do not use a pie chart.
If you want to show parts of a whole across multiple categories, use a horizontal stacked bar chart. It is infinitely easier for the human eye to compare linear lengths than pie slices.
Exponential Growth Curve
Principle 2: Maximize the Data-to-Ink Ratio
Coined by Edward Tufte, the data-ink ratio states that every drop of ink (or pixel) on a graphic should present data. Anything else is chartjunk.
- Remove heavy grid lines (use faint, light gray lines or remove them entirely).
- Remove background colors from the chart area.
- Remove 3D effects, shadows, and gradients that don't represent a variable.
- If a trendline makes the point clear, you don't need to label the exact Y-axis value of every single node.
Principle 3: Color Has Meaning
Never use color purely for decoration in a chart. If a bar is blue and another is red, the user will assume the color difference signifies a data difference. If they are both just 'Revenue,' make them the same color.
Reserve stark, contrasting colors (like red or green) for highlighting the specific data point you want the user to focus on (the anomaly, the drop, the spike). Let the rest of the data fade into a neutral gray.
