Last month, we talked about the nature of data storytelling, why analysts should learn it, and the difference between it from the technologies that it might use.

This post deals with capturing your audience’s attention and building your first data story.

In Medias Res

If you talk to professional editors, they’ll easily identify the “infodump” as among the worst story openings: page upon page of backstory and context that you have to slog through, and by the time you’re done with it, you’ve lost all interest in what happens next.

They tell authors to start in medias res – literally, in the midst of things.

We should do the same thing with data stories. Start with the situation that will grab your audience’s attention. Create a stark visualization, if appropriate. Give the minimum needed context: If you need to communicate a more critical context, do it after they’ve got the first image firmly planted in their minds.

Star Wars starts with a quick scroll, which gives a fraction of the story’s essential and immediate context. It’s just enough to set up that awful vision of Darth Vader’s Star Destroyer seizing Princess Leia’s much smaller ship. There’s no "infodump" about the political situation on Alderaan, the background of the Empire, the nature of the rebellion, or the nature of the Force. We immediately know whose side we’re on, what the immediate threat is, how powerful the enemy is, and how much trouble we’re in.

News stories are the same way, and they show that we don’t always need cool visualizations to tell a story. Here’s the opening of a Wall Street Journal article from February 1: “Stocks in China are primed for a steep fall Monday when markets in Shanghai and Shenzhen reopen after a week-long closure, even as Chinese authorities try to calm frayed nerves over the fast-spreading Wuhan coronavirus.”

Notice several things about this sentence. First, it has a general context (stocks and markets). Second, it provides its most important piece of information upfront, which in this case is a prediction (“primed for a steep fall”). Third, it contains just enough information about the current situation and recent past (week-long closure, rapidly spreading disease) to give that prediction credibility.

Well done, WSJ. That’s a lot of past, present, and future information about a market that’s unfamiliar to many people clearly conveyed in just 35 words.

This is a news story more than a data story – but they do convey more with data farther down the page.

Now check out this story from February 26, Europe Girds Against Italian Outbreak. In the print edition – old school, I know, but the online version has changed a lot and isn’t as good as I’m about to describe – the headline is supported by a large photo of some Carabinieri at a checkpoint.

But the thing that leaped out to me was right below that: the giant red dot in an infographic that shows the number of confirmed coronavirus cases in Italy as of that morning. And that infographic was supported by a photo of people confined in a hotel in Tenerife, Spain.

The combination of headline, infographic, and photos is arresting, and immediately conveys the main points: Italy is the center of contagion for Europe, and (it implies) we’re about to tell you more.

Here are some other things it does well.

It doesn’t bog us down in unnecessary context. For instance, the article mentions China only once, in passing, at the end of paragraph four.

It immediately engages our imaginations, partly through personal impact. (Note that it does so with the addition of a non-data-oriented picture. That’s completely legitimate, even with data analysis.)

It’s a story, not an argument.

It has a limited scope.

And now, understanding the need to start in medias res, we see that it also starts us off with a bang: No infodump, but an instantaneous understanding of the immediate situation. Everything else jumps off from there.

That’s great visual data storytelling.

At this point, the Journal dives into news analysis with strong data support. You’ll notice that there’s nothing driving readers to make a specific decision; that’s not what newspapers do. Data analysts might instead provide more data analysis with a strong newsy feel; they’d drive their audience toward some sort of decision point. The most important thing is knowing your audience and doing the right thing for them.

Getting Started

If we consider the things we discussed last time and in this post, we have enough to get started with our own data stories. Think about how you might tell a short data story this way.

Audience. First, identify your audience and the idea or situation you want to convey. (We’ll talk more about knowing your audience in the future.) Your idea should be more than the status of something your audience already knows about, and it should make them want to take action. If your idea won’t move your audience to action, choose a different story or a different audience.

Key point. Next, consider “is.” (Remember “telling it like it is, and was, and will be” from last time?) As a data storyteller, you have the advantage of being able to choose a visualization that communicates a lot about the current situation. That said, the situation you’re describing might be summed up in a sentence. Maybe it’s as simple as, “Sales in our restaurants are down in the southwestern US this quarter.” Remember also that, as in the Wall Street Journal article, the current state could be a prediction: a current change in your belief about a future state. “It is now our belief that sales will drop next quarter.” It’s a little meta, but it works.

Context. Next, carefully select the “was.” This is just enough context to let your audience know why the situation is happening, and perhaps why it’s important. It might include news about a new competitor in the southwestern US, alongside a chart that shows dips in sales that have occurred alongside competitor store openings. Tear out any chart, graph, or sentence that doesn’t contribute to the story.

Consequences. Finally, choose your “will be.” This might include possible scenarios that will occur if your audience doesn’t address the situation or scenarios based on possible actions they take. It might address things that you know, don’t know, and don’t know that you don’t know, which implies the need for getting additional data or pursuing further analyses. Regardless, it should give your audience a sense of how important their own actions are based on where they are in the story now.

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