How often Los Angeles County police officers use pepper spray might not strike you as having much to do with how organizations should navigate the Fourth Industrial Revolution. But Matt Hu-Stiles, a writer for the Los Angeles Times, shows exactly why businesses need to prioritize storytelling in their data strategies.

Matt used to think of himself as a data journalist when he helped launch the Texas Tribune, a digital non-profit news service, a decade ago. “Now I think of myself as a data scientist,” he says, explaining that he uses Python to automatically provide data analysis that he uses to inform his next step, which is actually shoe-leather reporting. “This is where the AI and the stuff doesn’t get you that far,” he says. “I still have to go out and interview people.”

Frankly, it doesn’t matter what he calls himself. Matt’s facility with machine learning, big data and analytics would make him a unicorn in the business world. Demand for workers with Matt’s unique set of skills is increasing while the supply is falling, which is why 46% of U.S. chief information officers say “big data and analytics” is the segment of the country’s job market with the worst labor shortage. (Note to his editors at the Los Angeles Times: Matt’s due for a big raise.)

With so few “full-stack” data scientists available, organizations are seeking specialists. If you can’t get a Matt who can design a Python routine to mine data and tell a story, can you get someone who can do one of those things? The danger in this, writes Google’s Cassie Kozyrkov, is that AI and data scientists enjoy primacy in the job market, relegating analytics to second-class status. “Far from being a lesser version of other data science breeds,” writes Kozyrkov, “good analysts are a prerequisite for effectiveness in your data endeavors” because “analysts are data storytellers.”

Storytellers such as Matt. He knows that “nobody wants to read an academic report,” he says. “They want to read something that’s compelling and interesting. You have to tell a story. That’s the hard part.”

He’s done it the other way. One Sunday he had a front-page story on firefighter pensions that was just a data dump. In other words, he didn’t interview people to find the human stories the data represents. The story flopped. Because data measures everything these days, he could tell how many people clicked on the story, how many seconds they spend on the page, and exactly where they stop scrolling. “The first thing I’m looking at is how many people are looking at it, how many people are staying on it,” he says. “They got a few graphs in and stopped.”

“I should have made it sexier, and it was just detailing the data findings. It still ran on Sunday page one, so my editors think it was a good story,” he says now. Still, it wasn’t a complete loss for Matt, the father of three. “With me, I like it because we get the Sunday paper delivered and I can show my girls that I’m still writing.”

What this taught him is that while data can improve all aspects of reporting – imagine having the crime data on hand before getting the chief of police on the phone? – nobody can replace the storyteller’s essential role.

Data journalism used to be much more rudimentary even than a boring story about dry statistics. At first, there wasn’t even the boring story. The Tribune became initially notorious in Austin, the Texas capital, for uploading the salaries of public employees to searchable databases. But that was about it. “It was all about learning how to put the data online. It was less about learning what the story is,” he says. “We were just dumping data on people and hoping they’d figure it out. I’m much more interested now in figuring out what the story is and explaining that.”

Now he tackles stories with two guiding questions: Who is the user, and what do they want? “How can I make this story relevant to someone who may not be directly impacted by it?” is the way he puts it. “If we’re not doing this work with the users in mind, we’re just spinning our wheels.”

It’s gotten easier. “Now you’re just awash in data and free tools to use it. The barrier now is learning how to use the tools,” says Matt, who embodies the mindset of continuous learning. He’s constantly challenging himself to tackle more ambitious problems.

Granted, what might be technologically ambitious might not appear to the reader as the next coming of Woodward and Bernstein. There was a hubbub about growing acrimony at Los Angeles County Board of Supervisors meetings. Specifically, there were complaints about profanity, so Matt imposed a data architecture on council transcripts and found that, yes, people were swearing a lot more often at council meetings. “It’s not gonna save the world or win a Pulitzer or whatever, but when a reader sees that, they are going to know we’re trying to tell stories in different ways,” he says.

The most impactful example of Matt’s work is also the most applicable for organizations undergoing a data transformation. Ideally, an analyst finds a story in the data relevant to creating value for the organization. And that story spurs further research that informs strategic decisions by leadership.

This is what brings us back to pepper spray. Kids in juvenile detention centers were complaining that officers were using pepper spray more often. No, said the guards, the problem is that the juvenile offenders are assaulting us more often. So Matt and his colleague Ellis Simani used the California Public Records Act to get the data and found that assaults against jailers were up sharply and – after doing further reporting – that “found cases in which guards used pepper spray when it wasn’t necessary — or not as a last resort.”

And, because we are talking about Matt, he didn’t just rely on the stories to tell the story. He designed data visuals you don’t need a doctorate to understand, uploaded all the public records used in reporting the stories to the cloud and designed an easy user interface for any reader who wanted to dig deeper.

Matt checked the boxes for data analysis. He used data to tell a story relevant to his users, in this case Los Angeles County taxpayers. The original story begat follow-up reporting, and in February 2019 his reporting led to a decision by the Board of Supervisors to discontinue the use of pepper spray in juvenile facilities.

It is vanishingly unlikely that a simple data dump would have ever led to this decision. In fact, all data, personnel and offenders were under the supervision of Los Angeles County all along, and they did nothing about it until someone analyzed the data. To tell data’s story, you’re always going to need a storyteller.