How to visually design your data: A chart guide

What you’ll learn

Visualise or vanish! To let data speak to your audience, you need to show it in graphic ways. In this chapter we present six simple principles (and a secret) to make your data visible in any situation and to avoid common data visualisation pitfalls in the management context.

Almost everyone today is in the data business, whether you realise it or not. Hence, we all need to become competent in conveying data effectively to others, whether it is in a presentation, report, dashboard, as part of a website or in a simple conversation. But we live in a data-abundant world, where our data is constantly competing with other compelling facts for attention.

To improve our communication of data and ensure that it is noticed and understood, we can rely on a time-tested approach: data visualisation. In fact, visualising data offers a myriad of benefits, from improved attention, quicker communication, better retention, to deeper exploration, sustained motivation and stronger engagement with the data.

To reap these benefits, however, you need to pay attention to a few key principles that ensure that a picture really is worth a thousand words – and doesn’t require a thousand words to become clear.

Having screened a plethora of books, articles and tutorials, and having conducted more than 50 data visualisation training sessions ourselves to managers from New York to Pretoria, from Bratislava to Bangalore, we have condensed the many data visualisation guidelines out there into just six memorable principles that anyone who visualises data for greater impact should consider.

These principles can be summarised with the DESIGN acronym that stands for the following six data visualisation imperatives (Figure 8.1).

Declutter

Emphasise

Storify

Involve

Give meaning

No distortions

The logic behind this set of principles is the following:

A high-quality data visualisation is one that is void of clutter or distraction, emphasises its main insight visually, tells a compelling story (with a ‘so-what’) and considers and captivates its audience. It gives clear meaning to numbers and avoids distortions or misinterpretations.

In the following, we put more meat on the bone of this acronym through varied examples and by providing a memorable tagline for each principle (Figure 8.2).

×

×

1. Declutter: The art of visual clarity

Decluttering a chart means eliminating everything that distracts from your data. So, get rid of borders, dominant grid lines, unnecessary details (such as decimals), 3D effects, (too many) colours, shades or other decoration effects (including exaggerated animation schemes). The following example (Figure 8.3) shows what you can leave out when visualising numbers. It is also an example of our next principle, emphasising key information visually (through a different colour or grey tone).

Also, make sure that your labels, legend or callouts do not interfere or overlap with the actual data display. A shocking counter example of this happened to The Economist magazine.¹

So, remember this: When in doubt – leave it out.

2. Emphasise: Highlighting the main message of a chart

Emphasising means two things: first, that you choose the graphic format that best highlights your key insight or chart purpose (Figure 8.4). Second, that you emphasise the most important element in that chart visually, for example through a different colour or by circling it.

×

To choose the right chart format for your data content ask yourself the following questions:

• ●Do you want to enable comparisons? Then choose vertical bars.
• ●Do you want to enable a ranking from smallest to largest? Then choose horizontal bars.
• ●Do you want to show a trend over time? Then use a line chart.
• ●A last option would be to emphasise deviations from a goal or reference (such as a budget plan). In this case choose upwards and downwards vertical bars.

There are, of course, other purposes and formats beyond these four main ones. Scatter plots are a good way to emphasise correlations or distributions and maps are the right format for data with a geographical dimension. In our periodic table of visualisation methods,² you can find many chart formats with examples, but remember that your best bet is often the bar chart.

So: Don’t look too far, just use the bar.

×

3. Storify: Drama for your data

To storify data means to present a chart (or a series thereof) in a way that lets you tell a captivating tale about the numbers that you are showing. In another chapter, we argue that this requires splitting your charts into a trilogy: (1) setting the scene (an overview chart or set that clarifies the situation); (2) showing the complications (one or several charts to show more details) and (3) providing a resolution (i.e., charts that show opportunities for action). You will recognise the Shrek-style ‘S’ in our acronym above, which makes reference to our data storytelling chapter. Hopefully, you will also remember the data storytelling insights we mention there (such as creating common ground – think Shrek in his bath).

Here is an example for a storified dashboard about female representation in management that we often use in our seminars and training and that follows the three acts/trilogy approach (and adds a bit of drama with the big ‘0% change’ figure in the middle). The first row clarifies the situation by showing that our company S has a female workforce that is below average and only 15 per cent of women in management. The second row shows more complications, namely that these 15 per cent are only in the lower management and that this has not changed in the last five years. The last row points to solutions, namely that to get more women to apply to management positions there needs to be more support and more flexible working conditions (Figure 8.5). You will learn more about this in the next chapter on data storytelling by the way, where we will use a variation of this data as an illustration as well.

Storytelling is all about sequencing a set of charts or animating and enriching a single chart.³

Storifying your chart also means adding emotion to it – if appropriate – and giving it a distinctive visual style. It means connecting with the audience by making the data relevant to them (selling before telling). This brings us to our next point: (audience) involvement. Before that, here’s the tagline for principle number 3:

To give data glory, tell it in a three-part story.

×

4. Involve: Engaging data audiences interactively

To involve others in the context of data means that you take your intended audiences into account when creating and presenting a chart. This can be done by giving your users simple ways of providing feedback to the chart and by enabling them to deepen their exploration by clicking on the chart.

In an interactive chart, you can involve your data users by letting them select areas of interest, zoom-in to more detail, explore different data aspects, filter out elements, customise the display, or connect it to new data.

In a live data presentation, you can involve the audience by letting them guess a result before showing it to them, or by asking the audience members what they find most striking in a particular chart. You can even go as far as asking them to put little sticky notes on a projected slide where they see further discussion needs or opportunities. This also works well virtually by using, for example, the Zoom annotation function that is active during on-screen presentations. Posting a downward arrow on a particular chart or section means that you want more detail about that data, while an arrow upwards would ask for the bigger picture or context of that data. A forward arrow would designate a discussion of the (action) implications of the data. A backwards arrow would signal the need to discuss the background of the data, such as the underlying sample. We call this approach to involve the audience through simple annotations ‘navicons’ as the placed icons help to navigate the dialogue about the data (Figure 8.6).

In all such presentations, remember to give your audience an overview first, and then details-on-(their) demand.

Broadly speaking, we are currently witnessing a shift in the analytics field away from simple one-way data presentation to more interactive data facilitation sessions – from merely presenting data to an audience to actively involving them in data dialogues. There is a myriad of free tools to support you in this endeavour, especially for virtual data talks. The most common software tools for this are miro.io or mural.co.

So: Be a (data) guide on the side, not a (statistics) sage on the stage.

×

5. Give meaning: Eight ways to make data relatable

Data visualisation is all about making data meaningful for your audiences. There are at least eight ways in which you can assist this sense making process (this requires what is sometimes referred to as ‘data screening’):

1. 1. Linking data directly to possible actions or responses is one way to make it more meaningful (think bar chart on the left, recommended actions on the right).
2. 2. Giving the chart an action title that expresses its so-what is another one.
3. 3. Adding self-explanatory labels and axes descriptions to a chart helps to make it more meaningful even for hurried viewers.
4. 4. Carefully adding symbols to a chart can help in its interpretation, for example a £, $ or € symbol for line charts with currency comparisons over time.
5. 5. Explaining the reasons behind outliers or other strange data patterns (for example, through mouse-over comments).
6. 6. Making data meaningful in a dashboard can be achieved by providing a reference point that shows whether a value is actually good or bad (above/below the target value).
7. 7. You can also give meaning to numbers by showing them in their development over time.
8. 8. Last but not least, you can make any number more relatable by comparing it to a phenomenon that the audience is familiar with, for example by showing that the Amazon has lost more than ten million football fields of forest in a decade. This probably gives more meaning to that data than saying that there are 24,000 square miles of deforestation in the Amazon rainforest from 2010 to 2020. Another (more business-oriented) example is to illustrate the market capitalisation of Zoom Communications (the web-conferencing company) by showing that it is equivalent to the combined market capitalisation of seven major airlines.

Using one or several of these eight interpretation aids may require a bit of space or text, but the effort is well worth it to bring the data alive in the minds of the viewers.

So: Giving data meaning requires data screening.

6. No distortions: What to avoid in chart design

The general rule with this last principle is to steer clear of graphic formats that make data difficult to understand or easy to misinterpret (Figure 8.7). Such sub-optimal formats include pie/doughnut/arch charts – as they are perceptually inefficient and hard to compare, see the figure below – stacked bar and area charts (as they have moving baselines), or charts that mix units and have two different y axes in a single image. You also want to avoid line charts with many crossing lines, as they are especially hard to read due to the many intersections and overlap, and replace them with so-called small multiples.

×

The no distortion rule not only applies to how you visualise data, but also to which data you visualise. Make sure you do not compare apples and pears or that you do not paint an incomplete picture by leaving crucial data out of the chart.

For an exhilarating compilation of chart examples that embrace such no-gos and fail in many different ways, check out https://viz.wtf/

So: Stay perceptually alert and leave the pie for dessert.

Key take-aways

Let us recap our tour of data visualisation principles from a different angle. The guidelines discussed in this chapter focus on these crucial aspects of effective data charting, namely:

• ●the clean style of the chart (to declutter it)
• ●the fitting chart format for the data content (to emphasise its main message and guarantee that there are no distortions, detours, or misinterpretations)
• ●the creation and the delivery process (to involve others and tell a resonating story)
• ●the (decision) context of the chart through its title or caption and its labels, reference points, or symbols (to give meaning to the data).

This is all very useful, you may now say, but where is the secret that was promised at the start? Believe it or not, we can learn from a 19th-century poet, Matthew Arnold, when it comes to a good data visualisation style and his so-called ‘secret of style’. He famously wrote:

‘Have something to say and say it as clearly as you can. That is the only secret of style’.

So, make sure you visualise and communicate data that is relevant to your audience – or that you can make relevant to them. Do this as clearly as you possibly can and pre-check with a colleague if it really is clear to others. The DESIGN principles presented in this chapter will hopefully help you for that purpose.

Traps

Communication traps

To be aware of the possible risks inherent in data visualisation, go through the following checklist when you have prepared your charts:

×

Further resources

You can find a very simple step-by-step example of how to declutter a chart at:

https://www.data-to-viz.com/caveat/declutter.html

You can find online examples of good guides to data visualisation at:

https://tinyurl.com/goodguidedataviz

https://medium.com/nightingale/style-guidelines-92ebe166addc

https://coolinfographics.com/dataviz-guides

http://visualizingrights.org/resources.html

https://visme.co/blog/data-visualization-best-practices/

https://killervisualstrategies.com/blog/three-rules-of-data-visualization.html

https://www.columnfivemedia.com/25-tips-to-upgrade-your-data-visualization-design

Notes

1. 1. Have a look at the chart published in The Economist and spotted (and subsequently improved) on this site: https://www.vizsimply.com/blog/redesign-for-storytellingwithdata-part-1
2. 2. At https://tinyurl.com/allviz
3. 3. A great example of such ‘scrollitelling’ can be found at http://www.r2d3.us/visual-intro-to-machine-learning-part-1/