How Public Speaking Can Help You Design Better Data Visualisations
Nine foundational principles from speechmaking that will help you see data visualisation in a new light
What do good speeches and good data visualisation have in common? More than you may think.
Aristotle — the founding father of all things public speaking — believed that the job of an orator was to discover the best available means of persuasion. That includes, first, defining all the arguments that can be made for and against a given proposition, then selecting those that will hold most sway with the audience and communicating them in the best possible manner.
Does this sound familiar? To me, an orator’s work seems very similar to that of data visualisation designer’s. We explore data to find patterns and insights that will be useful for the audience, and then communicate them in a visual way.
How can this parallel improve our thinking about data design? In this article, I’ll explore nine foundational principles from speaking in public and explain how they can help you improve your data work.
1/ Build Trust
At the centre of Aristotle’s approach to speechmaking is the concept called ethos. Ethos is essentially answering the question, why should I trust you?
If I were to give a talk on data visualisation, I’d start by introducing myself. I’d talk about my experience in the field and probably show some of my work. My posture, tone of voice and gestures would reflect confidence. All this would demonstrate that I have expertise on the matter. You’ll be more compelled to listen to me after I show you I know what I’m talking about.
See what I’m getting at? Ethos is the foundation on which your relationship with the audience is built.
In data visualisation, trust is also crucial, although it’s created in a different manner. Andy Kirk dedicated an important part of his book to trustworthiness. As he discusses, one way to show the reader they can trust you is to link to the data sources and put forward any assumptions they should know about. A good example of this is Datawrapper’s coronavirus charts where all the caveats in the data are clearly stated.
Another way to gain your audience’s trust is to pay (a lot of) attention to detail. That means you should consider every single word and dot in your chart: align all elements, remove unnecessary clutter, format the tooltips and much more. Take the graph from FiveThirtyEight below — isn’t it flawless? That’s what you should aim for, too.
If you design with ethos in mind, the audience will trust you. Do it consistently, and you will build a longstanding professional reputation.
2/ Teach, Move, Delight
In public speaking, the presenter can have one of three objectives — to teach, to move or to delight. Sal Kahn taught about online education in his 2011 TED talk. Martin Luther King Jr. inspired the country to take action against racial injustice in his 1963 I Have a Dream speech. President Obama entertained in his 2016 White House Correspondents Dinner address, most famous for its mic drop moment.
In data visualisation, the same three objectives apply. You may be teaching the audience about a topic or a chart type that’s unfamiliar to them (see this chord diagram breakdown by Nadieh Bremer). You may be trying to move them to take action (see this visualisation by David Borczuk). Or you may purely want to entertain with a playful visual (see My Fallen Kingdom by Judit Bekker).
Why is this important? If you define the objective at the start of the design process, it can guide you in choosing content. For example, if your aim is to move the audience to action, how will you present the problem? How will you show it’s relevant to them? How will you frame the conclusion?
3/ Connect the Dots
Once your objective is clear, you can define a through-line. If it’s the first time you’re hearing this term, you’re not the only one. The concept of a through-line is common in theatre plays, films and novels; I believe it was first introduced in public speaking by the team at TED.
So what’s a through-line? It’s your core message, the take-away you’d like the audience to go home with. It’s the connecting theme that ties together each narrative element of your work. Chris Anderson — the head of TED — cites two examples in his book. The first one is the start of a talk without a through-line:
I want to share with you some experiences I had during my recent trip to Cape Town, and then make a few observations about life on the road…
And the second one is the same opening, rephrased:
On my recent trip to Cape Town, I learned something new about strangers–when you can trust them, and when you definitely can’t. Let me share with you two very different experiences I had. . .
The message is much clearer in the second opening, isn’t it?
What would a through-line look like in data visualisation? As an example, let’s look at Ludovic Tavernier’s visual called Two Years Late. What’s his through-line? I’d phrase it something like this:
The human stories behind the US immigration policies.
How can you craft a through-line for your own work? Define an objective first, as discussed in the previous section. Then, go deeper. What exactly are you trying to convey? What do you want your audience to remember after they’ve explored your work? Be as specific as possible, but keep the message to a single sentence.
Now, when you add content to your visualisation, choose only those bits that relate to this core message. If you do so, it will be much easier for the audience to know where you’re headed. Your visual will be more focused and more impactful.
4/ Put the Parts in the Right Place
I love this quote from Quintilian on the importance of structure:
<…> though all the limbs of a statue be cast, it is not a statue until they are united, and if, in our own bodies or those of any other animals, we were to displace or alter the position of any part, they would be but monsters, though they had the same number of parts.
He paints quite an image, doesn’t he!
It’s customary to divide a speech into three parts: introduction, body and conclusion. I’m sure you already know this from school. What you may not have thought of is that you can (and often should!) apply this structure to data visualisation, too.
To understand how, let’s look at Ludovic Tavernier’s visualisation I introduced in the previous section.
Introduction
The objective of an introduction is to hook the audience, to incite curiosity, and to show them what the story will be about.
An introduction in data visualisation can comprise the title and the subtitle, the explanatory paragraph, and perhaps the first chart that provides context in a dense visualisation or an infographic.
In Ludovic’s visualisation (excerpt on the left), the reader’s attention is immediately captured with the two brief stories in the map and the strong title. Then, the reader is drawn further into the topic through the touching human story in the first paragraph.
Body
This is the meat of your visualisation, where you explain your message. In each part of the body, reveal something the audience doesn’t know yet, and then build on it, brick by brick.
Ludovic created three charts in the body of his visualisation (see below). He introduced three bits of information: the overall decrease in the number of accepted refugees, the situation by state and then its evolution over a longer period of time. Each of the graphs builds on the previous one, and once we’re done reading them, we understand the depth of the topic better.
Conclusion
This is where you summarise, provide a takeaway, add a call to action or open a bigger debate. In data design, the conclusion can take the form of a simple sentence, an action button, or even a visual.
Ludovic chose to talk about the political situation in the US as a potential cause for the issues he described, leaving us with some food for thought.
Structuring your visual well will make it easier for the reader to navigate it and will keep them engaged. Even if you’re not making a long infographic like Ludovic, you can still work with this structure. In a smaller visual, the introduction will be the title and the subtitle, the body will be the main chart area with its annotations, and the conclusion will be the biggest call-out, highlight or note at the end.
As an example, see Alli Torban’s visualisation on real-estate below.
5/ Strike the Right Tone
Imagine a scientist defending their PhD dissertation and a mother telling a bedtime story to her toddler. How will their speeches be different? The scientist will probably use complex terms and speak in a serious voice. The mother will be more creative and playful. The ancient Greek philosophers called this decorum — adapting the style of a speech to its audience and theme.
What’s decorum in data design? It covers your stylistic choices: colour, typefaces and illustrations. When reporting on a sensitive topic — the coronavirus pandemic, for example — it’s best to avoid rainbow colour palettes and fun fonts. You have much more freedom, though, when you work on a lighter topic with lower stakes. This is what Sarah Bartlett did with her European cities on a budget visualisation, shown at left. She used a creative typeface for the titles, many colours and icons.
Decorum is how your audience will decide if they can take you seriously, and if they want to engage with the visual. Don’t take it lightly!
6/ Less is More
As speech needs to be digested by the listener right away, short and simple words and sentences work best. You need to get your message across in as few words as possible, or, in other terms — less is more. A brilliant example is one of the most famous speeches in history, President Lincoln’s Gettysburg Address. With only 271 words, Lincoln inspired the audience and made them forget the hours of speech that came before him.
In data visualisation, the less is more principle also applies. While orators remove unnecessary words from their speech, you should remove any design elements that do not add value. Antoine de Saint Exupéry said, “Perfection is achieved not when there is nothing more to add, but when there is nothing left to take away.”
Keep that in mind for your next project. When you think you’re ready to publish, go through the visual again and ask yourself: is there anything else I could remove?
7/ Let It Breathe
Think of the last time you were listening to someone who was speaking very, very fast. How did it feel? Was it difficult to follow, difficult to understand? Or was it confusing? Tiring? Now think about the opposite — the way President Obama speaks. It’s pleasant to the ear because he takes his time and pauses often.
Pauses are a crucial part of any spoken expression. A speaker needs to pause so that the audience has time to digest what they’ve just heard. A pause can also give emphasis to an important part of the message.
In a data visualisation, white space is a pause. Leaving enough white — or negative — space will allow the audience to go through the visual in a calm and pleasant fashion.
Let’s illustrate this with an example. Below are two versions of the same visualisation. The version on the left is the equivalent of a fast-spoken-difficult-to-digest speech. The version on the right side, on the other hand, includes negative space that allows the visual to breathe.
It’s not easy to decide how much white space to add and where. If you need tips, graphic design techniques are your best shot. I suggest you start with this article from the Interaction Design Foundation.
8/ Be Bold, but Don’t Scream
Think of the last time you were telling a story to a friend. Can you remember what your voice sounded like and how your body moved? You probably told most of the story in your natural voice. At a few occasions, you may have gotten excited, shouted something out, or added a swing of an arm. Those are the bold moments in the story: by changing your voice and body language, you show the listener that this is important. You can only do it sparingly, though, or you’ll overwhelm them.
In a data story, you also create bold moments. One of the most common ways to do so is with colour. You use colour to make something important stand out, to draw the reader’s attention to key data points. However, the same way you don’t want to listen to a friend scream through an entire ten minute story, your data readers don’t want to be screamed at either.
So how can you moderate colour usage? I like the 60–30–10 rule from interior design: 60 percent of the space should be a dominant colour, 30 percent a secondary colour and only 10 percent the accent colour. Just like in the example from FiveThirtyEight on the left.
9/ Test It
If you’ve ever made a speech or a presentation, you were probably told to practice in front of others. But what for? After working on the script on your own for a while, you get too used to it. Complicated sentences start to sound normal, and gaps in reasoning go unnoticed. Rehearsing in front of people helps fix that. You collect feedback, and then improve your speech.
The same applies to data visualisations. You should collect at least one person’s feedback on how they perceive the visual. That person doesn’t need to be an expert — they can be your friend, your spouse or even your mom. Federica Fragapane, a famous Italian information designer, shares her creations with her mom over WhatsApp.
The questions you ask to receive constructive criticism are important. I wrote a quick blog post that can help you get the most from that experience.
What’s your favourite speech of all time? One of mine is Martin Luther King Jr.’s I Have a Dream. This speech still resonates and inspires us to fight racism, even though it was delivered nearly sixty years ago. Data visualisation can have the same effect. We still talk about Florence Nightingale’s revolutionary work, for instance.
This year proved — once again—that data visuals can have a profound impact on our society. Washington Post’s coronavirus simulator was their most-read story of all time. It changed the way we understand the virus and the way we act.
To create such impactful and long-lasting data visualisation work, imagine it as a speech. What will you teach the audience? How will you make sure your message remains memorable years from now?
