Why data visualisation sucks (mostly)

Data visualisation has been a big trend in recent years. For a long time, I felt like it was something that I couldn’t do: I didn’t have the graphic design or coding skills to emulate the celebrated examples. But there was also something else: when I looked at them, I couldn’t figure out what they were supposed to mean. Was I missing something? Did everyone else understand?

Now that I’ve started to do more data visualisation myself, and I’ve grown in confidence about what I’m doing, I’m ready to say it: most examples of data visualisation suck. We’ve become excited by the beauty and power of data, but frequently failed to deliver actual value with it.

New technology and software solutions have changed the data visualisation landscape completely in the last decade. Increased computing power; the advent of languages like Python; and dedicated visualisation tools like Tableau and Datawatch have made complex analysis much more accessible.

There has also been a huge shift in the availability of data. In the past, data had to be collected by hand, making it expensive and time consuming. Automated data collection means that it is now relatively easy to capture huge quantities of data with relatively little effort and often in great detail.

But in the excitement over these new opportunities, it is all too easy to lose our way, assuming that more data is better and that any visualisation is good visualisation. We need to keep in mind the question of why we are doing it: what can we learn from the data and how best can we express that?


There isn’t space for a full critique of recent trends in data visualisation in this article, but here are some highlights:


Infographics are those illustrations that present lots of data facts laid out against a background image, or enlivened with pictograms. Although the best of them do a good job of telling a story with numbers, many are just a whole load of data facts spewed onto a page and illustrated with the most basic charts. See Why we hate infographics for a more in-depth discussion.


Lots of examples of data visualisation present an amazing looking graphic, but with no commentary at all – no story, no interpretation. What am I supposed to learn from a million interconnected lines showing the social network of everyone in Belgium? Let’s call it ‘data art’ and be done with it.


Edward Tufte has made a career out of explaining how to visualise data well. One of his key insights is that charts should minimise ‘junk’ – the decoration that does nothing to express the data, but is just there to make it ‘pretty’. A lot of recent data visualisation shows an abundance of chart junk: wrapping charts into a circle; connecting concepts together with wavy lines; and clouds of bubbles are all good examples of things gone awry.


Embedded within the hype around data visualisation software is the idea that new tools mean that anyone can do it. In reality, working with data requires a good knowledge of statistics; and a good understanding of data presentation and design.

Visualisation software makes it easy for anyone to produce amazing looking charts, but it also makes it very easy to produce really dumb charts. What it won’t help you with is knowing the difference. Get someone with the relevant skills and experience to help.


New technology makes it relatively easy to provide interactive data visualisations, allowing viewers to change filters, zoom in, and follow paths through data. While this can be initially engaging, interactivity typically adds nothing to the actual insight derived by the viewer.

Before adding interactivity, we should ask ourselves: will users understand the options that we are providing; and what additional insights could they draw by using them? In most cases, it would be better to simply provide a powerful and valuable interpretation of the data and leave it at that.


With new technology, we can have data visualisations updating in real-time (or near real-time). This may seem really valuable, but it usually isn’t. Imagine if you asked a market research organisation to run the same questionnaire with your customers every day. Would the results keep being interesting?

If the numbers don’t change that fast, and any sensible reaction would take weeks to implement, then updating the visualisations in real time is just a pointless distraction.


Too often, raw data is just turned directly into charts – no analysis, no interpretation, just charts. Good use of data should improve our understanding; change our opinions; or help us to make decisions. Visualised or not, raw data is just data, and doesn’t take us anywhere.


It may be relatively easy to collect huge quantities of data, but that doesn’t mean that it is the right data. It may be very useful to know how many users you have, but if you make your money by having them buy things, then what you should really care about is sales. Getting 50% more visitors, but no more sales is a failure, however you look at it.


Dashboards deserve a special mention of their own. They seem to be everywhere these days, and everyone seems to want one. But they also manage to combine most of the problems mentioned above into a single package.

Stijn Debrouwere sums it up well: There’s nothing like a dashboard full of data and graphs and trend lines to make us feel like grown ups. Like people who know what they’re doing. So even though we’re not getting any real use out of it, it’s addictive and we can’t stop doing it.”

For an entertaining critique of all things dashboard, see: The Laws of Shitty Dashboards.



Analysing data and using it to inform and make decisions is not the same thing asvisualising data.

Analysis might take the form of just organising and manipulating figures to find an answer. In the past, this was typically the only way to approach things.

With the advent of powerful visualisation software, it has become possible to use visualisation as tool to actually analyse and explore the data. This can be hugely valuable, however there is a danger that the tools can get in the way, persuading you that you that visualisation is enough.

The reason we are working with data is in order to extract relevant insight. Focus on the analysis, not the visualisation.


A good project starts with a problem to be solved, gathers data (or makes use of available data), analyses it, and then presents findings using visualisation. In many cases, the project will also include some iteration of those steps – initial insights will open new lines of enquiry and lead to new insights.

In a ‘scientific’-style project, the problem to be solved is very clear and well understood, usually because it has been carefully formulated. This is the ‘hypothesis’. A great example of this is A/B testing, where two variants of a design exist and the question is: “which is better?” An appropriate measure of ‘better’ must be chosen, but after that, the process is relatively clear: run both and analyse the data to see which performs best.

However, not all projects can follow this model. In an ‘exploratory’ project, the exact problem may be unclear, but the general ‘problem space’ is known. Perhaps you are concerned about the rate of user migration; or you suspect that there are sections of your site that are getting much lower traffic than expected. Analysis and visualisation are used as tools to explore the data, simultaneously deriving insights and gradually establishing a clearer problem statement.

Regardless of the differences, it is the effort to deliver an answer that keeps the process meaningful.


The final step is deriving the maximum impact from your insights by presenting the outcome well. It is all too easy to get drawn into showing lots of charts just because you have them.

Most people don’t have lots of time and attention to devote to understanding what you have to say. Edit down your message to its essentials and choose a small number of well-designed visualisations.

Really powerful visualisations are often not immediately easy to understand – they require a bit of effort from the viewer. What sets them apart from mediocre visualisations is that they deliver a disproportionate amount of value once they are understood. For this reason, it is vital that you don’t just present the charts, you explain how to read them and what they mean.

Finally, put it all together into a compelling narrative. One that explains why this work is important and relevant. Your audience will typically forget the details of your presentation. Think about the two or three points that are most important and make sure that these come through clearly.


Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s