Last year’s webinar exploring the potential of Power BI demonstrated that data visualisation tools are increasingly accessible. Most business users have access to Power BI within their organisations through Microsoft 365 licenses. And it doesn’t just have to be dedicated visualisation software, as the recent webinar for the Charity Community on bringing finance board packs alive in Excel demonstrated. All of this means that more finance professionals than ever are engaging with new ways to present data to stakeholders, and while there are some excellent examples of well-presented data, there are also some that leave a lot to be desired! 

In this article, we’ll explore three fundamental principles of data visualisation that all charts and dashboards should abide by, which should help anyone who is starting to dabble in the field.

Through a combination of nature and nurture, humans’ minds mostly all work in similar ways. Our eyes perceive, and our cognitive abilities are trained to interpret these visual cues according to predictable patterns, as a way of essentially minimising the time it takes for the brain to interpret the information being provided to it.

In Western society, there are certain innate rules that the we all follow:

• We read information from left to right, top to bottom
• We know that red means stop or bad, and green means go or good
• We know that when something is highlighted it is worthy of greater attention

So when it comes to data visualisation, one of the most fundamental principles is to harness these intuitive behaviours. Putting the most important information in the top left portion of the dashboard means that users will naturally identify it first. If you want to show good and bad, use green and red. Use visual cues to draw attention to particular features of the data that warrant the most attention, such as contrasting sizes, colours or typographies.

Another key element of this is that dashboards should be inherently intuitive – if you need to explain it, you’re probably doing it wrong. Over-labelling might feel necessary to assist the user with understanding the key messages, but it adds clutter and can cause more confusion than it solves. Make the dashboard as self-explanatory as possible, and provide supplementary notes separately only if they are absolutely necessary.

The old adage that ‘data is truth’ doesn’t really ring true when it comes to data visualisation, and unfortunately as we use visualisation techniques to tell a story with the data, this can give rise to dashboards and charts that more closely resemble fiction than fact. But as accountants, we must, as far as possible, remain truthful to the underlying information we are looking to present. 

Yes, inevitably, when we take raw tables of data and move towards presenting this in a meaningful way, we cannot avoid some sort of narrative, so we need to ensure we always take that step back to check that what we are presenting is as free from bias as we can make it, and that the data is allowed to speak for itself. 

A large part of this is ensuring you have a good knowledge of the audience that you are presenting to. What is their level of understanding of the data? What will they be expecting to see? Are you seeking to nudge them towards a particular decision, and if so, is that influencing the way you seek to present the data?

An example of this principle can be found in the following three charts, all of which show the exact same data but in three different ways:

Now, none of these charts are lying, per se, but they each paint a very different picture of the trajectory of the business. In particular, the first two charts are identical in every way except for the scale; by failing to include zero on the y-axis, the sales and budget journey looks substantially more volatile than it actually was. Equally, a chart showing purely variance between sales and budget fails to note that the 2021 performance was still measurably below the 2019 peak. 

This is a common trick for certain less reputable news and media outlets who are looking to accentuate differences that in reality may be negligible. And another trick is the total lack of context for the data being presented. In this example, it doesn’t take long to realise exactly what happened in 2020 that had such an adverse and unexpected impact on sales, but in the real world it is rarely quite so obvious.

The last principle is sadly an often-overlooked one. Designing dashboards inclusively is about more than just making them friendly for people with disabilities (although this is also important and links again to knowing your audience). How are people going to consume the dashboard? Does it need to be easy to interact with on a tablet or mobile device? Does it need to be interactive at all? Not all dashboards have to be crammed full of bells and whistles, so a clean, simple, uncluttered set of static visualisations may be all that is required. But even if interactivity is desired, this should be done in a way which remains easy to engage with and avoids clutter.

As for accessibility, the temptation is often to align with an organisation’s brand colours, which might be fine, but might not be… it’s important to understand that colours are a powerful tool in visualisations, but can easily exclude those with visual impairments including the roughly 4.5% of the population who are colourblind, so ensuring that there is sufficient contrast between the colours used matters more than sticking rigidly to brand guidelines. As many as 1 in 10 of the UK population have some degree of dyslexia, and 1 in 25 have ADHD, both of which can make a complex, text-heavy dashboard tricky to immediately engage with. Again, simplicity is key.

As the title of this article suggests, data visualisation is a discipline in its own right, distinct from data science and requiring a unique set of skills. The design process of a dashboard requires an eye for aesthetics, for how users will perceive and interpret information, and how to get key messages across – artistic flare is of some value. Meanwhile, the build process requires meticulous detail in the creation of visualisations, working with the data to deliver the desired outputs – a much more scientific approach.

It is the combination of both art and science that gives us the discipline of data visualisation – a creative process that follows clear principles and structure, or a technical capability that requires visual craft. To master it fully is no mean feat, but to be aware of its guiding principles is something that is of importance to all analysts and consumers of data. 

If you’d like to learn more about data visualisation techniques, sign up to our upcoming virtual classroom on 26-27 March.