A designer knows he has achieved perfection not when there is nothing left to add, but when there is nothing left to take away.
The use of charts to communicate quantitative business information is now so commonplace that we assume that the knowledge of their effective use is equally as common. Just because they are relatively easy to create with the help of widely available software, it does not mean that their design will automatically make the message hidden in the numbers clear and easy to understand.
There are many good examples of data visualisation from which we can learn, but there are also many that fail to communicate the message clearly and can be confusing and even misleading. The reason for it is the lack of knowledge, training and experience in creating well designed charts.
This section, based on the experience and advice provided by data visualisation experts, will highlight examples of good practice and where appropriate tell you which design mistakes to avoid. It draws upon expertise provided by many well respected writers, especially Edward Tufte, Stephen Few and Andy Kirk.
Keep your charts simple
One of the most fundamental rules in creating charts is to keep them simple. Tufte (2013), the graphics guru who in the last 30 years has shown us how to visualise data with simplicity, clarity and elegance, has campaigned against, what he calls 'chartjunk'. He describes this term as “what happens when your data design focuses too much on the design aspect. Distracting patterns, overbearing colours and even unnecessary grids and outlines are just some of the elements of chartjunk”.
Tufte’s advice is to minimise non-data ink. By this he means the amount of ink used in a chart which isn’t used to display data itself. Any chart elements like for example frame lines around the charts, background colours, or excessive grid lines, which don’t enhance the understanding, are considered as non-data ink and should be treated as superfluous.
Consider Tufte’s advice and use a balanced approach to your chart design. You don’t have to cut out all visual elements – things that emphasise the key message are useful if they help get your points across to the audience.
Focus on purpose
Visualisations must serve a clear purpose eg to communicate, explore, show detailed or high level view. Don’t try to do everything with one visualisation.
Choose the right chart for the job
If you know and understand the strengths and limits of each chart form, it will help you to use the best chart for the data type and required analysis. More information on this topic can be found in the type of charts section.
At its foundation, data analysis is about comparisons and the visualisation of differences within data sets. For that reason it is helpful to always ensure that the data is set in a context.
Imagine looking at data trends based on the last couple of years forgetting about the long-term perspective.
Be careful when showing changes in values expressed as percentages. If the absolute numbers are small, even a small increase can appear high when expressed as a percentage. For example increasing student admissions from 10 to 15 will show as a 50% increase, compared with much lower 20% growth if the absolute numbers were 200 and 240.
When trying to explain the scale of the area affected, as in this example by wildfires fires, it can help to superimpose the size of a well known region or a country on the map shown, to help the reader to better understand the difference in relative sizes.
Utilise micro and macro perspective
Depending on the purpose of the chart and the audience, you can choose to show your data at either a very high level, showing the pattern and shape; or if the details and numbers are important, then focus on the lowest level of granularity or consider presenting the data in the tables.
Senior management are often less interested in small detail, their focus being on trends and the big picture. Charts used at more operational level might need to show fine detail for example helping with day to day monitoring of students admissions.
Interactive charts that allow data to be viewed at several levels by the user drilling up or down can serve wider users’ needs.
Avoid information overload
It can be tempting to try to squeeze too much information into one chart. After all you can show several variables using different shapes, colours and sizes, but would this approach aid the understanding? In some cases it could be beneficial, but instead you could consider displaying several charts dealing with related data variables side-by-side, or in a trellis display, which is a layout of smaller charts in a grid.
If you go for this option, make sure that all individual charts have matching scales and the variables are in a similar order to help comparing between charts.
Overload, clutter, and confusion are not attributes of information, they are failures of design.
The Graph Selection Matrix from Stephen Few’s book ‘Show me the numbers’ offers useful guidance for when choosing graphs.