What is Data Storytelling?

How do you tell a GREAT story with data? 

Everyone likes hearing a good story. However, being asked to “tell a story” using data and visualization is often a big source of anxiety for analysts of all backgrounds.

An informal Twitter poll returned the following responses to the question “When I’m asked to show the data, I feel….”

Frustrated, because I don’t think I’ll tell the story effectively, and might miss important parts.

I feel pressure, pressured to make it clear for everyone, and what if people don’t like my story?

Inadequate, because I’m sure there are questions people will have that I haven’t anticipated.

Being able to tell stories with data is a skill that’s becoming ever more important in our world of increasing data and the desire for data‐driven decision-making. As more and more data visualizations are produced, they start to become a commodity and their quality suffers. This turns off viewers and people begin to rethink their investments. However, great Visual Storytelling can send the effectiveness and reach of your analysis through the roof and produce significant business influence, value, and career rewards.

Have a look at the graphic below. What story do you get from it? 

Here’s what most people see:

1. The average surface temperature is trending higher.
2. Multiple independent data sources all show the same trend, which lends further credibility.
3. Temps were trending lower at one point, but that reversed and has started growing in lockstep with the rise of industrialization.

Even though it’s little more than a few words, numbers, and some colored lines, it tells a very compelling story, with strong supporting evidence, and makes its key points very persuasive.

Storytelling with data is no different than regular storytelling. Storytelling is by far the longest-running and most effective method of human-to-human knowledge transfer. The reason storytelling is so effective is that it engages emotions along with cognition (ability to learn). Emotion keeps you interested while you learn. 

• Build Characters: First and foremost, stories involve characters. Without humanized characters, there’s nothing for the viewer to relate to; It’s not a story. Think about it. Every story you’ve ever read has human-like characters for you to relate to. Even movies and stories about animals and inanimate objects impart human characteristics to those characters: they talk, they react, they have expressions and emotions, they act human. In data, the character(s) can be you, or the reader, or named people, or people in a certain role, customers, employees. e.g. “Our Sales managers wanted to know how to ….” “I sought to uncover why ….” “Our CEO, Mike, asked me to investigate why ….” or even, “My daughter asked me about our company, and she wondered why X product was so successful….” A great way to introduce characters visually is with earlier Survey Results. The survey results let you introduce and describe characters, “Male employees, under 40, working in our US Offices had the following to say in a recent survey:” and also give them a voice.
• “Use the Force, Luke”. The next thing that is required is some kind of goal, challenge, or objective. This line from Star Wars was Obi-Wan’s way of challenging Luke to learn the Jedi way and set out his hero’s journey. The characters must want or need to do something. That’s the hook. That’s what gets the reader to go along with the character, to put themselves in their shoes. In data, a great one here is answering a difficult question or solving a business problem. Even better would be showing how your insights and answers resulted in a measurable improvement. So give your characters a challenge to overcome. The bigger the challenge, the more interesting the story becomes because….. 
• ….“Never Tell Me The Odds”: Give the character some stakes, some consequences of complete failure. Optional: Let the character fail, but use it to illustrate what they learned. Failure makes the stakes seem real and pulls the reader in even further. This is called “rising action”. Just don’t lay it on too thick.
• “Have a Point!“: Every story needs a climax, a point where the action peaks. This should involve being the main message you’re trying to communicate. It should always involve your character(s) achieving or exceeding their goal. For example, if your “challenge” was to use analysis in order to create a plan to change something, you can include a Gantt chart or Change Roadmap here as the deliverable of that analytics journey. You could also phase shift this slightly to show a visualization of what the Outcome of that Change was.
• “Falling Action.” Now that you’ve made your main point, you can use this part to tie up loose ends, resolve other challenges or conflicts besides the main one, describe what happened to the characters after the story, or even use it to tease a sequel. You can also use this part as a “Call to Action” for the viewer if you want them to do something, such as give feedback or share your dashboard with others. 

Now that you have the main foundations of Data Storytelling under your belt, give it a try. The best way to master it is to practice it often. Learn what works best for your data and your audience. A-B test different approaches and get feedback on what worked. Look back over earlier work from yourself and others and list out the things you might have done differently now that you have this new knowledge. Hopefully having a structure like this to start with will give you confidence in choosing and arranging your next visual exercise to maximize its message and persuasiveness. How will you know you’ve succeeded? People will tell you. People know good storytelling when they experience it. Good luck!