What Exactly is Data Storytelling?
The art of turning data into actions.
Background
When I started my weekly series of articles more than 6 months go, I wrote 2 well-received articles on Visual Analytics — the first was about the concept of visual vocabulary and the next was on how to create useful dashboards. Of late, the sub-topic of data storytelling has been on my mind. What exactly is data storytelling and how does it, if at all, defer from data presentation? Is it a unique skill? Is it a unique discipline with its own body of supporting knowledge?
I teach an adult class on Visual Analytics at one of the local universities. I’ve been working with my co-instructor (Koo Ping Shung) on extending the curriculum into a series of modules, including one on data storytelling. And so I dedicate my 28th weekly article to discussing the topic of telling data stories.
(I write a weekly series of articles where I call out bad thinking and bad practices in data analytics / data science which you can find here.)
Data Storytelling as a Competency
Let’s first define storytelling and then deepening it to data storytelling in the context of visual analytics and data science. Storytelling is about using emotions to carry a point across to a specific target audience. Consultants use this well to convey the need for their services by describing the pain-points or missed opportunities that can be solved to the benefit of the organisation. Using this analogy, data storytelling similarly is about carrying a data insight across by injecting it with emotions. The end-state of ‘crossing over’ is the call to action.
It’s been said in many places that many data scientists lack the ability to communicate their data findings appropriately to their stakeholders, so much so that their proposed solutions don’t get off the starting block. Data storytelling has become a critical gap that requires addressing in the data analytics / data science community. But I would go further and call out that it is a critical ability for ALL knowledge workers, given the rise of interactive visualisation tools such as Tableau and Power BI. Singapore’s government-led skills and competency framework and ontology has defined Data Storytelling & Visualisation as a competency with multiple levels of proficiency, further underscoring its importance. The competency is often described as an ability to use appropriate visualisation and storyboarding techniques to translate data insights into compelling narratives.
Unpacking Data Storytelling
A good data story has the following sequence:
- Background context → Insights discovered → Implications → Actions needed
There are a few things to note in this sequence:
- Use of the right visual vocabulary to convey context, insights and call-to-action.
- Adding the right visual cues to convey emotions.
- Your role as the supporting agent in the narrative.
Let’s consider point (1) — the right visual vocabulary (see table below). The following visual vocabulary is useful for setting background context: distribution, ranking, change-over-time and parts-to-whole charts. They are descriptive in nature. (Note that setting context includes information about the ‘actors’ in your story narrative.) When it comes toinsights, the following visual vocabulary is useful: deviation, correlation, and flow charts are able to encode for interesting and often counter-intuitive patterns. When it comes to implications and actions, using trend forecasts, magnitude and deviation charts can help your audience understand what they are reaching for.
Let’s consider point (2). Firstly, I need to clarify what I mean by ‘insights’. An insight is defined as information that improves decision making. ’Improve’ is the operative word; and that’s why it’s naturally action-oriented. Now, behaviour psychology has revealed that emotions play a massive role in decision-making; experiments have shown that if we were purely logic-processing, we would take forever to get to a decision because of all the pros and cons to consider. Emotions are heuristic hacks for short-cutting the decision-making process. Now, most good insights tend to challenge excepted wisdom and studies have shown that your audience will likely default to a defensive position; it is delusional to believe that the “facts speak for themselves”. What this means is that the more counter-intuitive your insight, the more emotion is needed in your data visuals.
Consider the chart below when the WHO was monitoring the outbreak of Covid across the world. It is a choropleth map, belonging to the geospatial visual vocabulary. Imagine if the colour gradients were replaced by Green-Amber-Red (i.e. ‘traffic light’ colours) where Red signifies countries with ‘unacceptable’ infection rates. Or if we replace the choropleth map with a world map showing dynamically-sized icons of coffins depicting the level of infection in each country. I hope you can see that the change of visual cues obviously raises the emotional temperature of the data narrative. The use of colour and icons (officially known as ‘isotypes’) can convey emotions quite effectively. I recall a time when I used male and female isotypes to convey gender inequities in an organisational study for a client, and the room immediately ‘got it’.
On point (3), you need to consider your role in the overall narrative. Not all messaging is encoded in the data. The data story is not meant to be self-contained without the narrator. You are an integral part of the story-telling; consider what you will say and how you will say it to support the data story. The late Hans Rosling (see his famous Ted talk) was an exceptional data storyteller. It wasn’t just about the visual, but also the narrative that went along with it. The broader agenda was his clarion call about investing in Africa.
Conclusion
Data storytelling is more than just making fanciful charts. Data storytelling is about turning insights into action by way of invoking emotions through the use of visual representation. It is a competency that can be learnt once we understand the ‘rules’. Data storytelling is a great way to force clarity of thinking from any data author; and the ability to tell good data stories is only going to be more universally critical as the knowledge economy advances.
