Opening minds with data visualisations

Mateusz Otmianowski
by: Mateusz Otmianowski | May 9, 2018

Data scientists can sometimes get carried away in their pursuit of helping their audiences get insights from their data. They frequently want to present ‘all’ the findings instead of focusing on the most important ones. One of the reasons for it is that they don’t want to take the responsibility for deciding which piece of information is important and which can be ignored. Such an approach usually fails because it results in flooding the audience with too much information, which hinders their ability to understand and internalise findings.

Focus on Who, What, How

Cole Nussbaumer Knaflic, the author of Storytelling with Data: a data visualisation guide for business professionals proposes a radically different approach. She suggests that before even attempting to create any visualisations you should think about the Who and the What.

  • The Who is your audience. It is the context in which you need to immerse yourself. You should think what do they already know, what biases they might have, how they might react to your message, and what objections they might raise to counter your points. Getting the understanding of these issues allows you to move to the What.
  • The What describes the message you want to convey to your audience and/or the action you want them to take. It is usually good to recommend some actions, because your audience will be more interested in actionable insights. Even if your recommendation is wrong, it may still start a valuable discussion about how the audience could act on your findings. At first, aim at boiling down the ‘so-what’ of your message into a 3-minute story. Your ultimate goal should be to summarise your so-what in one sentence.
  • Once we know the Who and the What, we can finally start thinking about the How. The usual consideration here is what type of plot to choose, but sometimes just one or two numbers or a short sentence are enough to deliver the message. Other times, a table could do the job. You might not even need to write a report, and one sentence email will suffice.

Become a visual minimalist

If we decide that we need a plot, there are a few general guidelines. Use scatter plots for showing relationship between two things, line plots for continuous (usually time-related) data, and bar plots for comparing the magnitude of things. Avoid pie charts and 3D plots. You should keep in mind that these are only general rules, and soon you’ll get a good intuitive grasp of when you can break them.

No matter what type of plot you choose, your main goal is to maximise the signal-to-noise ratio. Knaflic illustrates this nicely with a quote from Antoine de Saint‐Exupery’s Airman’s Odyssey: “You know you’ve achieved perfection, not when you have nothing more to add, but when you have nothing to take away”. This means that we should throw away all the visual clutter from the visualisation and leave only the essential things.

When creating any visualisation, we should be thinking like designers and guide our audience’s eyes through it to help them understand it. A few tips for achieving that are:

  • Use colour sparingly just to highlight the important parts,
  • Create a visual hierarchy of information using colour (e.g. using grey color to move some information to the background), font size and weight (e.g. the more important information, the bigger or bolder the font),
  • Eliminate any distractions.

Knaflic provides great illustrations of her recommendations on her blog. The plot on the left comes from the Pew Research Center, and the one on the right was re-created by Knaflic.


The point that initial plot was trying to make was that the 2011-2012 increase observed in total new marriages was driven primarily by an increase in those having a bachelor’s degree or more. A number of things were changed here to make this message clearer.

First, the plot was changed into a line plot instead of a bar plot. That makes it easier to notice changes across time. Second, the lines were annotated directly and the legend was removed, which eliminates the need for back and forth movement between the legend and bars to identify the colours. Third, the colour palette was reduced to just three colours, so the trend that we want to point catches readers attention almost instantly. I think that the resulting visualisation delivers the point more directly, is easier to digest and is aesthetically more appealing.


Be aware of the context, focus on the most important things, keep your message short, and visualise it in simplest and most effective way possible. Keep in mind these suggestions and it will be easier for you to influence your audience’s opinion with your findings. If you want to learn more I recommend reading Storytelling with Data: a data visualisation guide for business professionals.

Mateusz Otmianowski

Mateusz Otmianowski

Mateusz is a data scientist in the Product Data Science team within the AI Products & Solutions unit at Pearson. In his free time, Mateusz likes to stay active by riding a bike, running, and swimming (and occasionally doing triathlons), and then reading science fiction books while regenerating.
Informacja dotycząca plików cookies

Informujemy, iż w celu optymalizacji treści dostępnych w naszym serwisie, dostosowania ich do Państwa indywidualnych potrzeb korzystamy z informacji zapisanych za pomocą plików cookies na urządzeniach końcowych użytkowników. Pliki cookies użytkownik może kontrolować za pomocą ustawień swojej przeglądarki internetowej. Dalsze korzystanie z naszego serwisu internetowego, bez zmiany ustawień przeglądarki internetowej oznacza, iż użytkownik akceptuje stosowanie plików cookies.