Blog

What Not To Do / Data Analysis  / What not to do when visualising data

What not to do when visualising data

The art of data visualisation

In the world of business almost nobody can escape the visualisations, the graphs, the charts, and the dashboards. As a Data Analyst, this is my bread and butter, but I have had my fair share of failures trying to bring a point across in a chart. In order to help you make the most of your data visualisations I’ve put together some of the mistakes I’ve made in the past and tips on how to avoid them.

Reporting vs Informing

My first job in data analytics was largely based on internal reporting and building big dashboards for the leadership team. When I moved onto a more external-facing role I realised quickly that in terms of data visualisation these were two totally different categories. Going from providing a large amount of data with tools that can be filtered and modified by the user, to creating PowerPoint presentations to share with clients, was a much bigger change than I expected. As I found the latter a lot more challenging, I’ll be focusing on the once-off type of data visualisation in the following failures.

Say it all at once

Trying to ‘say it all at once’ is a mistake that I still catch myself making, even after years of visualising insights. If you’ve put a lot of effort into an analysis and only have your audience’s attention for a limited amount of time it can be tempting to put everything you’ve found into one graph. This often leads to people spending more time decoding the visual than actually listening to the presentation.

As the creator, you can choose the most important takeaway for the people you’re presenting to and either drop the rest or give them the attention they deserve in their own visualisation. A rule I use for my own presentations and reports is to have only one key takeaway per graph and if necessary, use the same graph with different focal points multiple times.

Know your audience

Another issue that often comes up when visualising data is failing to recognise your audience. That doesn’t just mean to be aware of the data literacy of the group and make sure you don’t use complicated visuals, this also means being prepared to answer detailed questions about your methodology from people who are very experienced in data analytics.

I’ve run into both issues many times, most notably when I was presenting an analysis to one of my clients and they asked surprisingly detailed questions. In theory, I knew all the answers and was confident in my analysis but when I was put on the spot without having prepared detailed notes, I wasn’t able to answer many of them. This, of course, made my analysis look a lot less thorough and since then I always add detailed notes to my graphs just in case.

On the other extreme using complicated explanations of a methodology might make the graph look more trustworthy, but if the audience did not need that much detail the main point might get lost. This is why I prefer to focus on the action point of the visualisation and give space for questions to add detail if needed.

3D pie charts that rotate

Like many of us I’ve seen my fair share of incredibly confusing and plain ugly data visualisation. My favourite examples are pie charts, especially if they are in 3D and even better if they rotate. Secretly I am a big fan of the pie chart, even though it is frowned upon in data analytics, but there are good reasons why it rarely makes sense.

One thing I learned while reading Cole Nussbaumer Knaflic’s great book ‘Storytelling with Data’ is that it’s much harder for us to compare surface area than length. Think about comparing two lines, we can fairly easily say which one is longer and if one is about double the length of the other. Now try and do that with slices like in a pie chart, it is much harder to say if one is double the size of the other and very difficult to say which is bigger if there is only a small difference.

Nowadays I try to always compare line-length to save myself and my audience from mental gymnastics trying to compare two surfaces.

Top Tip

One very practical piece of advice I can give is to use colour to your advantage and ignore whatever Excel or PowerPoint is trying to suggest to you. The best way to go about it is to look at your graph and make the whole visualisation grey, including the writing, the graphs, everything.

From here on you can now pick out the one point you’d like to highlight with a colour. This could for example be, one bar in a bar chart being blue while all others are grey. This way the audience knows exactly where to look.

It’s a simple trick but it makes it so much easier for me to highlight what I want my audience to focus on.

No Comments

Post a Comment

testing placement for now