What makes a good data visualization?

Reading Time: 6 minutes

Common pitfalls in charts and how to avoid them

Aims and Goals

Most people think charts are boring. In many ways they are but that is because they are so familiar. Hence, many data visualisation experts spend a lot of time creating “new” and never-seen-before visuals. Unfortunately such visuals require lots of mental work to understand and in many cases they are a rather ambiguous representation of the underlying data.

The evolution of a chart

Most people actually prefer working with

Scatter Plots

Line Charts

and Column Charts

  • added horizontal tick marks from left to right on the y-axis. I love this trick because anyone looking at the chart can immediately see how the values oscillate between 40% and 60% across the entire length of the chart.
  • tilt the x-axis labels at 45 degrees as opposed to the default vertical orientation. That’s again much easier to read!
  • Another important thing to remember especially when displaying percentages is to show the zero on the y-axis. It is a convention that such charts are from 0 to 100 as opposed to zooming in, e.g. 20%-60% in this case. This approach makes charts comparable because the scale is always the same.
  • Never forget to put a title on your chart. It is scientifically proven that readers do look at it for information about the visualisation.

There are still problems

The main defect in the plots above is called overplotting. The display area is always limited to a certain size, e.g. page or screen dimensions, which means that if there are too many data points they start to overlap .

Fixing overplotting in scatter plots

a. Coalesce points



Tidy up


You can find an interactive version of the chart here.



One thought on “What makes a good data visualization?

Comments are closed.