The Beauty of Data Visualisation

One moment you’re listening to one of your favourite podcasts, the next you’re interviewing one of the guests yourself. That’s basically how Rick and I ended up at Nadieh Bremer’s place. An astronomer turned data scientist, turned freelance data artist, who has created captivating one-of-a-kind data visualisations for clients such as Google News Lab and The Guardian. Not surprisingly, Nadieh has been a much requested interviewee and conference speaker. Nonetheless, we set out a challenge to come up with creative questions and talked about the importance of design, the magic of pen and paper, and why art is priceless.

Design matters

Back in 2012 Nadieh, a graduate student at Leiden University at the time attended Deloitte’s Business Course where she discovered that although strategy consulting may not be for her, the analytics team within the Business Intelligence department could be a great match. As such, she applied for a position and eventually received a job offer. Being a layman in the field turned out to be helpful she explains: “At the start of each case I was far from a subject-matter expert. Therefore, it was quite easy to take a step back and remind myself of how I would have explained it to a younger me.” By doing this she learned about the importance of design: “People trust things that look good. Once our slides were well-designed the client presentations usually went smoother.”

Chord diagram: Switching between mobile phone brands (Deloitte Global Mobile Consumer Survey)

Although aesthetics has always had her utmost attention, Nadieh only later found out there’s actually such a thing as a data visualizer. At a Strata conference in 2013, she was introduced to the principles of the JavaScript library D3 which turned out to be an important turning point in her career. “I fell in love with all the possibilities. For example, I saw these amazing interactive charts on the New York Times website and thought: hopefully I’ll be the one creating them someday.” However, she still needed to acquire the necessary skillset, so she started reading books about design and playing around with D3 in her spare time. Initially, she could barely put her newly acquired knowledge to use at Deloitte, though slowly but surely this changed as she became more proficient in the programming language: “Over time I became more and more efficient so that I could eventually use the tools in my everyday work for clients.”

From then on she increasingly specialized in dataviz and even changed career tracks to Adyen as a full-time data visualisation designer in 2015. After a year of primarily (re)creating dashboards, she eventually decided to continue as a freelancer because of the freedom and variety of projects.


Building community

As you can see above, Nadieh’s decision was received with great enthusiasm by the community which she had built up through her blog: Visual Cinnamon. I asked her about the story behind it: “The reason I started making these advanced data visualisation tutorials was because I couldn’t find anything like it at the time. After I finally managed to put together all pieces from various Stackoverflow threads, I thought: let’s share my learnings with the rest of the world so that others can hopefully benefit from it. Also, I was a bit driven to connect with the community: if I published my work, I may get under the attention of a few people and I’m no longer a complete outsider, I thought.”

Her first conference presentation in which she explains how using default shapes in an unconventional manner can lead to interesting results.

Over time she slowly transitioned from blogs to conferences. In 2016 she gave her first talk at OpenVis called “SVG Beyond Mere Shapes”. Nadieh elaborates: “Here I discovered my true passion for presenting about data visualisation. Every single time I truly enjoy the conversations I have with the community members afterwards, especially if you see that people have suddenly become enthusiastic; that they realise there’s more to it than just bar and line charts.”

Always start sketching with pen and paper, don’t limit yourself to just the default charts.

She can’t stress the latter enough: “I would highly recommend anyone — regardless of their technical proficiency — to explore what types of data visualisations exist. Search for it online or pick up a good data visualisation book and look what’s out there.” Her second advice builds upon this idea: “Having this scala of visualisations at the back of your mind, always start sketching with pen and paper; don’t limit yourself to just the default charts. The fundamental question is: how can you best present the data? A sankey diagram, bubble chart or, network diagram? Don’t worry too much about the technical implementation, that will come later. Oftentimes, you can come a long way with your go-to tool even though you might not think so at the start.”


Tech vs Art

At first glance, you may not immediately associate programming with the data visualisation designer’s daily job. Hence, I asked her how much time she dedicates to the technical nitty-gritty as opposed to the actual design process. “The preparatory data cleaning can be long and intensive but most of the time I leave that up to the client. After some preliminary exploratory data analysis, I start brainstorming and sketching on paper which usually doesn’t take longer than 2 to 3 hours. Turning these drawings into pixels doesn’t take too much time either, in fact 90% of the time is spent on design, interactivity, browser compatibility, and responsiveness. Simply scaling back visualisations for mobile devices doesn’t do the trick, more often than not I really need to adjust the whole lay-out to make it work. So you’re basically designing while programming. Although it’s more of a technical job, it’s your personal creative style that makes your work stand out and enables you to earn a living as a freelancer.”

3m x 1.5m data art piece for Transavia (Dutch airline) that nicely illustrates how technology and creativity can complement one another.

So then the question arises: how does she describe her own style? After a long pause — seeking for patterns in her work — Nadieh continues: “I love working with large datasets that tell not one but multiple stories. In general, I try to present the data in several dimensions so that the audience can come to the same conclusion from various angles. Although my visuals are intuitive, they are not always straightforward. There is usually a deeper level of knowledge and understanding to be revealed. I like to show the underlying data, so for example not only the average but also the variation around it.”

Preliminary sketches of a “Baby Spike” visual that illustrates how variation can be incorporated into visualisations.

Art is priceless

Now that she’s working as a freelancer it’s no longer a matter of only data and art, but also business plays its part. What challenges does she face in that regard? “Pricing my services remains tough; it’s always difficult to estimate upfront how long I spend on a given project. Moreover, the return on investment for data visualisation is somewhat hard to determine for clients; unlike e-commerce, you can’t speak about higher conversions, it’s about positive brand associations.” One of her pricing tips is, therefore, offering multiple packages ranging from simple to complex: “That way you change the direction of the discussion from yes/no to which one.”

On a final note, she likes to share another word of advice with all aspiring data visualisers: “Just like data cleaning is an integral part of any data pipeline, data visualisation is an indispensable skill: knowing how you can effectively communicate your insights should be part of any data scientist’s toolkit. And don’t forget to experiment; not each and every visualisation you create has to be absolutely beautiful, it’s only the last one that counts.”


Data is beautiful once it’s the best possible version of itself: on the inside and outside. Although you should be careful to mess around with its core truth, it’s your duty to make it shine. It goes beyond communicating insights, it’s about exciting people and reaching an audience that has never heard of matplotlib, seaborn, and ggplot. After all, data science is an art and art is for everybody. Are you the next Picasso?