How to Find a Story in Data
Tips for data storytellers who struggle to find stories in data.
- By Ted Cuzzillo
- December 15, 2015
A data analyst raised her hand in a class I taught on data storytelling and asked the question I hadn’t even thought about since journalism school: How do you “see” a story in a jumble of facts?
It’s a novel problem for data analysts, but it’s an old one for journalists. In fact, as confusing as the task seems to analysts, the confusion is a mystery to journalists. Don’t analysts know a story when they see one?
Now in the grand new confluence, journalists use data and analysts tell stories -- and each side shudders with the other’s ham-handed work. Yet, as other once-irreconcilable factions have done and others may do yet, we all might as well get used to it and learn from each other.
What advice do journalists give analysts about seeing a story? I thought I’d find an easy answer with Google, and searches came up with page upon page of advice -- just about all of which stayed on the data analysis side of the chasm. Not one looked across to journalism.
I gave an impromptu answer to the data analyst: Take off your analyst’s hat and put on your journalist’s hat. Here are a few approaches to do that.
Focus on the audience. Stop thinking about the data and think of what the audience wants to know of what it means, what’s new, or what’s different.
Think of what story you would tell a member of the audience over coffee. Forget your grand entrance, forget the brass band, forget about your boss staring at you. Just tell the story simply and plainly between sips of coffee. What aspects would you emphasize and what would you leave out? How would you structure it? You might find your story’s germ there.
Is anything significant in your data? Events become newsworthy with timeliness, proximity, novelty, or impact. The reporter covering a house fire, for example, may ponder various angles: a house that burned down within the news medium’s area is more significant than one outside of it. Yesterday’s fire is more significant than last year’s fire. The mayor’s house is more significant than those of most former mayors. On the other hand, George Washington having slept in the house even one night trumps everything.
Look for anomalies. Everyone’s heard of “man bites dog,” the anomaly that explains what becomes news. “When everything goes as you expect -- the sun comes up, spring follows winter, the airplane works flawlessly -- there’s no story,” writes Stephen Denning in The Leader’s Guide to Storytelling (2011; Jossey-Bass). “Paying attention to apparent anomalies is one of the reasons that we have survived as a species.”
Remember that the data is not necessarily the story. This is the most common discovery I’ve heard from data analysts. A vice president at AT&T once told Fern Halper, now director of TDWI Research for advanced analytics, to just tell him something that is 80 percent correct. Don’t get too get down in the weeds. “For him,” she said, “good enough was good enough.”
Max Galka, a cofounder of Revaluate, an apartment-rating service for renters in Manhattan, found that his customers wanted simple data. “You have to focus on the high level,” Max said. At first, he displayed data the way he likes it. “I wouldn’t put much credence in a building’s overall score [if] there wasn’t any detail behind it,” he said. In fact, he could offer deep, rich data on scores of apartments, “but consumers like [simplicity].”
He tried to lure people into logging into the site to get rich data in elaborate tables and hierarchies, the way he likes it. Few did. “One guy checked 10 or so buildings every week without logging in.”
Deciding what data to show, lose, or summarize has to be guided by audience and medium. What does the audience really want to know? What does it know already? What incomplete stories can you support or question?
Is the anecdote about the guy checking so many buildings without logging in meaningful as data? No, it’s about just one person -- but it’s the story that’s remembered and retold. Galka, a data analyst, had taken off his data analyst’s hat and put on his journalist’s hat.