Data Storytelling, Part I: Telling It Like It Is, And Was, And Will Be

How to do data storytelling - Part I

We’re told that data storytelling is a new way of communicating using data.

I submit that it’s among the oldest ways of communicating.

In fact, data storytelling is effective precisely because it’s not really new. Storytelling is primordial, tapping into something deep within us. Data storytelling connects to that same something using numbers, narrative, and visuals. It forces our lizard brains to give data the respect it deserves.

Regardless, data storytelling had better be something more than dashboards or visualizations, or it’s Just More Hype. 

Why Storytelling Has to be Learned

I believe data storytelling can be something more. In fact, it gets at the heart of a data analyst’s value: It communicates insights that would otherwise be hard to absorb.

But many data people aren’t natural storytellers. They’re arguers.

Rather than tell a good story, data people create logical arguments. They start with hypotheses, gather relevant evidence, and then doggedly pursue the implications of their research until they come up with the answer to the problem they’re supposed to address. In fact, they’re often drawn to data specifically because it helps them create that logical structure.

There’s nothing wrong with that. I love a good argument, and I want my decisions to be backed up with impeccable logic. It’s essential in fields like philosophy and law.

But philosophers and lawyers aren’t known for capturing our attention with vivid tales that show their points clearly.

All of this implies that analysts should regularly follow two distinct processes: One to attain the insights the business needs, and another for communicating those insights. Data storytelling addresses the second process. It doesn’t take away anything from what analysts currently do. It adds a technique to their tool kit that helps them communicate what they know more effectively. And mastering that technique will open up a new mentality and appreciation for framing data arguments into compelling narratives.

The Nature of Storytelling

In fiction, good stories are easy to spot. They thrust interesting characters into challenging situations, forcing them to make character-defining choices that have significant implications for themselves, their friends and family, even the world.

Stories are active. We don’t passively receive stories: We experience them. We engage our imaginations to understand the flow of the story. In doing so, we let the story stimulate our desire to see what happens next.

A great story makes us yearn to turn the page.

Cause and effect. Implicit in these statements is the flow of cause and effect. Some people talk about stories having beginnings, middles, and endings, which is generally true, but for our purposes, I’d like to be a little more flexible.

After all, an “ending” might resolve some issues and open up new ones, and a story that has reached resolution may turn out to be the opening of a larger story. Think The Hobbit and The Lord of the Rings, or Patriot Games and the rest of the Jack Ryan stories. So instead, let’s think of the flow of cause and effect in terms of what was (context), what is (the current situation), and what will be (predictions, possible outcomes, and so on).

This flow doesn’t need to be all-encompassing, by the way. Trying too hard to capture every historical detail can lead to paralysis by analysis. It’s okay to read The Hunt for Red October without reading Patriot Games first.

As a data storyteller, then, your first job is to determine the scope of the story you want to tell. What is the compelling situation that needs to be understood and addressed? What is the essential background needed to understand its causes? What are the effects of actions the business might take, including the act of doing nothing?

Personal impact. Stories are engaging and entertaining, but their value comes from the window they provide into our world. Many great stories tell us things we already know about ourselves, but in a way that allows us to understand them more deeply than we otherwise would. And some stories communicate something so deep that they lead us to change our minds in profound ways.

Finally, stories provide us with new information. Nobody checks status by reading a story. Even when we re-read stories, it’s to get more from the same story, not to get new data.

That’s what we’re looking for with data storytelling: Interesting characters, choices, and implications, demonstrating the flow of cause and effect.

Data Storytelling vs. Specific Technologies

This is the point where many articles start talking about combining data, narratives, and visualization. I think that’s premature: Just like it’s premature to talk about technology before you talk about the business results you want, we need to understand the value of data storytelling before we decide how we’ll implement it.

Dashboarding

One good way to do this is by contrasting data storytelling and dashboards.

Dashboards are so common that it’s easy to forget that they’re a metaphor. Cars, aircraft, and other conveyances have dashboards that provide an up-to-the-moment status of its important components: speed, fuel remaining, and so on.

These things are all well understood. In fact, the whole point of the dashboard is to let the driver or pilot know the status of things she already understands. A dashboard should only call attention to one of its elements when something’s out of whack; in normal circumstances, the standard reaction to a quick dashboard scan should be “Good, nothing to see here.”

Business dashboards, whether strategic or operational, are the same way. Key Performance Indicators (KPIs) or Objectives and Key Results (OKRs) show up on dashboards because they’ve already been defined and chosen by the business as good representations of status.

That’s the opposite of a story.

A story tells us something new – something we’re trying to understand. No part of it is assumed to be “normal” or “fine” and therefore ignorable: Everything is part of a larger narrative. Stories help people absorb new information by putting it in context and showing the cause and effect among different characters and events.

Dashboards don’t change anyone’s worldview. At most, dashboards lead to questions that lead to more analysis, and it’s the follow-on analysis – often and ideally in the form of data storytelling – that leads to changes in perspective, worldview, and decisions.

To sum up: Stories make you turn pages to find out more; dashboards make you turn away, satisfied that you know what you need to know.

Visualization

Many people connect data storytelling with visualization technologies. There’s good reason for that, too: Visualization technologies support data storytelling because a picture is worth a thousand words. (Please don’t hold me to precise picture-word equivalencies!) Like storytelling, visualization is primordial. “I can’t unsee that” is a common phrase on the Internet precisely because visuals grab onto a ton of brainpower and don’t let go.

Visualizations excel at conveying important context or event information, especially with respect to change over time. Since data storytelling involves the flow of time, visualizations are critical for compact, easily understandable data stories.

Visualizations support storytelling, then, but data storytelling requires more than just using visualizations. We’ve all seen dashboards that use great visualizations to show status – and we know that status-oriented dashboards aren’t storytelling.

Natural Language Generation

Natural language generation, or NLG, is a technology that takes a data set as input and uses AI to create a human-language output. For example, one use case takes highly volatile data as an input and returns a sentence that describes things that would be hard to see on a chart: “The average sales price fluctuated 9.2%, with a peak at $2.19, decreasing approximately .5% over the reporting period.”

It’s a great technology – but NLG shouldn’t be mistaken for data storytelling, either. Whether you represent a few static data points with a line graph or a few sentences, you still haven’t told a story. Narrative is crucial, as we’ll see, but narration – merely using words – isn’t enough to qualify as data storytelling.

Conclusion

In posts to follow, we’ll go into more detail about knowing your audience, knowing how to start your data story, understanding who your characters really are, and much more. See you then.

Jake Freivald

Jake Freivald spends his time at the intersection of data management, analytics, and business communication. His twenty-five year career in software includes developing and marketing products for data warehousing, data...

More About Jake Freivald