Thousands of years ago, when we lived on the savannah, we were all data analysts. A lion passed by, we analysed the situation and killed the lion or tried to run very fast. We found berries, scrutinised them before eating (and died or survived).
The need to be analytical is not new. It’s always been important, even long before the digital big data era.
Being able to analyse data is an important skill set because we create our models of the world based on the results we see when we experiment. If we can adequately examine our data, our models will better represent the real world.
But data is not enough.
To be an excellent berry investigator, data analyst, or astronomist, you need two other skills: 1. You need to communicate your findings to others (“Last night I found a planet I’ve never seen before, you should look at it too”), and 2. You need to convince them that your conclusions are correct (“Finding a new planet means that our solar system consists of at least twelve planets”).
These two steps might be familiar, maybe you know them as storytelling…
Storytelling is a tool so powerful that we sometimes get moved by something entirely fictive or fake. Because storytellers have no duty to base its stories on facts. Instead, it often becomes the consumers’ role to find and analyse the data, to see the errors in the original analysis and adjust their worlds accordingly.
But, the relationship between storytelling and data is most often a fruitful one, one where they make each other better. The result is usually best when both get to shine.
The balancing act between data analysis and storytelling is continuous. It impacts every slideshow ever made, every ad ever cut. Too much data makes the story halt. To little data makes the story lose its power.
Very few data analysts masters the art of presenting data. Also, storytellers and analysts are not known for their vivid collaborations. But when we start to see our data reports as storytelling, our perspective shifts.
As someone with a somewhat rational logical brain, I sometimes have a hard time with things that removes focus from the fact I want to communicate. Hence, I present the point I want to make and most often, nothing more.
Data is boring
But turning my data analysis into storytelling is not for me — it’s for you, the recipient. Because data is boring ( — and beautiful if you’re anything like me, but most people aren’t). So, I sometimes lose the people who’ve got a hard time putting the data I display into context. And if I want to make sure people listen to what I have to say, I need to move it into a narrative or framework that will help them understand what I’m trying to say.
So, what’s the secret to moving from boring data presentations into something more engaging? Creating great stories based on data is a little bit like building a house. You start with the foundation, add on the framework, continue with the roof, and keep going until you have some final steps of painting and decorating.
Here’s how I’m trying to go about it:
Step 1 — Find you why
I write down my own “why”. Sometimes this is: “I believe this client is better than it’s competitors”. Or, “I believe I can learn something interesting by looking at data from racist Facebook groups”. Most often, it’s more in line with “I need to figure out how we’re doing so that I can make the necessary adjustments.”
Step 2 — Collect your data
I collect my data points in a very unstructured form. A lot of my data analysis is exploratory, and I believe that’s fine. Most of the time, I have a topic I’m exploring, but I don’t want to already have a fixed storyline in mind when I start this process. I just collect what I find, often as bullet points and with visualisations from Tableau in a Dropbox folder.
Step 3 — Organise your data
I organise my data points in a straight line. The findings that support or are close to each other should be presented either together or following each other. I also want to find a natural way to introduce the story and a natural way to end it. I often start with broader strokes, move further into detail later, and move back out again to summarise.
Step 4 — Map out a storyline
Starting to get a storyline in place, I translate my findings into a narrative. Here I’m making sure to use “everyday language” setting the context to the real world.
Finding a story and making a coherent argument is as relevant for an audience with no domain or data knowledge as it is when presenting for experts. However, in a domain-specific context, they will want to know more about particular tests, definitions of variables and other things that I leave out for most audiences.
This step is often where I make that differentiation. Putting an analysis into context means two very different things for a group of domain experts and a group of school children. Here’s a great example of how to change your story based on your audience. So, knowing your audience is always crucial.
Step 5 — Start to think about visualisations
I decide what visualisations that are needed to communicate my argument well. Sometimes it’s with a chart, but more often, it’s not. Most people I work with are not data literate, and if I show them a data visualisation, they tend to get very stressed.
If I decide to use a chart, there are some things I have in mind:
- I. The visualisation should be able to stand on its own entirely. So, it needs a title, and both the axis and the included data need to have names.
- II. Categories should be simple to understand and make sense to anyone
- III. The colours used should make sense and be coherent. Not in a “pink for women, blue for men” type of way, but keep your colours throughout your story and make sure they are easy to separate and remember. Remember that a lot of people are colour blind, so green and red is a bad combo.
- IV. Make sure the chart is possible to decode in the format you will present it. Make sure text is big enough to read for instance in a video or on a screen
Step 6 — Make your story come to life
When I have a storyline in its raw form, I try to find ways to make it come to life. This is, by far, the hardest step for me. How can I invite the audience to feel, or to use their imagination so that they are well primed for the argument I’m going to make? What examples would make people care?
This is also the part where I most often bounce ideas around and ask for help. I’m slowly starting to realise that using a story to introduce and end an argument help people care. It’s my responsibility to make them care.
Using cases is also an effective way to make a story come to life. It’s so much easier to understand the problems of a small business owner when they are presented as the daily struggles for a 55-year-old widow owning a petrol station instead of 15 statistics. Building a narrative around one or more data points are not tampering with the results, it’s a smart way to make the result stick.
Be careful with metaphors. Most people are misusing them, forcing a metaphor to work when it’s not even close to telling the same story as you’re trying to communicate. When they work, it’s a fantastic storytelling tool. But you really need to think twice before you use them.
Step 7 — Connect back to where you started
End by connecting back to where you started. Either by tying it to the initial story or by summarising it all neatly. Zoom out and put your argument into a broader context.
Step 8 — Test!
Test it on both your audience and on people who know the underlying data. Data analysis and storytelling is not an either-or relationship, they need to coexist. If your storytelling is making your data analysis incorrect, (for instance, by using metaphors that don’t capture your argument well), you need to go back and change it. So you’ll need to double-check that the human-friendly version is perfectly aligned with the data analysis you started out with.
This was a very long guide to how I combine data and storytelling. Sometimes I start with the data, sometimes I begin with the message. You can attack from either end.
And while it often sounds like data and storytelling are far apart — and we usually give people responsibility for one but not the other — we should try to push them together. Sometimes, we just need to educate ourselves a bit before it feels comfortable to work closely with tools that you’ve not used before.
We need both. Most stories need data, and most data need storytelling to get comprehensible by a larger mass. And even if you decide not to learn both skills yourself, you’ll at least need to learn how to collaborate with someone who knows what you don’t know.