Structure

Knowing what you’re building and who builds it is the easy part. The harder question is where the team sits and how it works with everyone else. Different newsrooms solve it differently.

Four survey respondents, four completely different setups:

I report on a data team of six (two player-coach editors and four reporters) that reports to the investigative team. One of the reporters is in DC, another in SF; two of us are based in NY, along with the editors. There are also other data reporters throughout the newsroom embedded in financial investigations and the corporate team.

I’m not sure we really have a data team. Just me! I make infographics and data visualisations to accompany articles we publish. The data I work with is mostly provided by academic authors who publish with us.

We are a group of three data visual developers with one data editor and one data fellow. Usually, our team is about five people depending on the time of year.

There are many reporters qualified to do data reporting; maybe only a handful have that in their official title.

Part of the variance stems from the diversity of organizational needs across newsrooms. Some of it stems from the fact that data journalism, in its contemporary form, is still a relatively new field, and many newsrooms are developing their approach as they build. Based on our research and experience, we are fairly indifferent to the literal org chart of your data team. More important is how the team interacts with other teams. It is also fine to start small: a data team of one or two can produce excellent work.

The most important call you’ll make is how much your data team works with editorial versus product. Some data units are independent news desks that cover the news on their own; others function more like in-house visual and app designers. Either way, data reporters should report to data editors, not to IT managers.

Editorial interactions

At some organizations, data teams sit squarely within editorial, more often than not under the umbrella of investigative journalism. For these teams, data editors look like other editors: they attend editorial meetings, pitch ideas, and receive pitches. As one respondent described it: “Most projects where my team collaborates with the newsroom start with a conversation between beat reporters and data reporters. Others begin with another editor reaching out to me. This year, especially, we have to say no to great ideas more often than I would like.”

Other data units are slightly more detached, and collaborations arise when beat reporters come to the data team with questions or visualization requests. Whether the unit falls explicitly under editorial or not, we believe the most productive collaborations stem from early involvement. Well-integrated organizations bring data reporters into projects at all phases, from early questions about data integrity and methodology to full co-reporting on feature-length stories.

Product interactions

The contemporary digital newsroom is more versatile than the early online publication. It’s easier to store and access data than it was 20 years ago, and easier to build immersive digital experiences. Data teams are often responsible for this kind of work, and because of the technical complexity involved, publishing modern data journalism implies closer relationships with in-house product and engineering teams.

The split usually falls like this: data editors, researcher-reporters, and even visual journalists sit on the editorial side of the house. Full-stack devs are best placed between editorial and product. At large enough organizations, news-app developers constitute their own team; for smaller teams, hybrid roles that report to both editorial and product leads make more sense. As one respondent put it, “We work with our engineering and product team to get our graphics working seamlessly in our CMS and matching the latest style of our site. They also help set up our servers.” Good collaboration requires good communication, and most of our advice for pairing data work with product work boils down to erecting clear communication channels between the teams.

The realistic playbook

For a newsroom under a million dollars, the move that works is smaller and cheaper than people expect: hire one multiskilled person, embed them as a full member of a desk, and borrow everything else.

Embed is the operative word. A data reporter who sits in edit meetings, calls sources, and shares a byline is worth several of one who takes tickets from a service window. Run data as a desk, not a help desk. Lean nonprofits like Grist, Mississippi Today, and MLK50 do exactly this with small teams.

Then stretch that one hire with things other people already paid for:

• Clean data you didn’t have to assemble, from Big Local News at Stanford, which standardizes hard-to-get government datasets and shares them free.
• Free training, like ProPublica’s Data Institute and Local Reporting Network, for the people you already have.
• A collaborative or network (the Institute for Nonprofit News, a regional partnership) that shares infrastructure and reporting so you don’t carry the fixed cost alone.
• Report for America or a similar fellowship to help pay for that first seat while you build the case for a permanent one.

One hire is where you start, not where you stop. As the work proves itself, add the role you’re missing most; the four archetypes in People double as a hiring order.

Bigger newsrooms can buy their way to capacity instead of building it. In 2024, CalMatters acquired The Markup and folded a whole data-investigations team into a roughly hundred-person nonprofit. It’s a great picture of what scale can look like later! It’s probably not a plan for a shop of five, though.

Takeaways

Generally, these three points are the key to productive collaboration with the rest of your newsroom:

• Limit concurrent projects. Again and again, we’ve heard that data journalists feel pulled in too many directions. Understand how many people your technologists report to, and don’t ask more of one person than is reasonable.
• Try to avoid surprises. Early involvement produces the best results. Bring your data team in early and often.
• Understand the talent you have. There’s no such thing as a “data person,” because data people aren’t monolithic. Some love scraping; others love statistical modeling; still others are itching to build the next scrollytelling library. Understand your team’s capacity, and position it accordingly.

Get those three right, and the exact org chart matters far less.