Getting started

It's a good question. We think that phrase -- data unit -- means a lot of different things to a lot of different people.

So let's agree on a shared definition for now. Here, we generally understand a 🔮data unit🔮 to be a team of people in a newsroom who leverage data and contemporary data-science technology to complement the traditional tools of reporting. That's it!

Your newsroom might already have one. It might already have one and not consider it to be one. It might have one and call it something else, or it might employ the constituent employees of a data unit but scatter them throughout the organization without the anchors of camaraderie and appropriate business cards. This guide seeks to offer a set of best practices for thinking about the size, position, and organization of such a team in your newsroom.

It's probably worth getting another definition out of the way. For the purposes of this guide, we consider 🔮data journalism🔮 to be the use of data and contemporary data-science technology to complement the traditional tools of reporting through data collection, statistical modeling, visualization, and frontend design. That should sound familiar. In other words: Data journalism is what a data unit does.

By the way, you're in the guide now. You are being guided. If you've ended up here by accident or via some kind of fugue state, you can check out our About page to learn more about this project and its ideal reader.

🐣 The basics

The size and shape of a given data team will, unsurprisingly, depend on your organization's budget and goals. In our research, we've heard from data teams as small as one person working part-time and from data teams as large as 20 or more full-time staffers. Most small and/or non-profit newsrooms fall somewhere in the middle of this spectrum.

Broadly speaking, though, data units are generally home to four types of people:

1. Researcher-reporters,
2. Data artists,
3. Full-stack devs, and
4. Data editors.

We'll think about each of these archetypes -- and that's all they are -- in the next section on People. At smaller organizations, one or two journalist-engineers might end up playing more than one of these roles. (The smaller your org, the more unicorn-like your employees need to be... but that's not news to you.)

At successful data shops, these people tend to be well-integrated into their newsrooms. They tend to report to an editor or two, and they tend to be positioned under the organization's investigative journalism umbrella. They're not limited to investigative journalism, though. Data units play key roles in graphic design and product development, and they tend to respond to requests from across the organization -- whether for the purposes of data visualization, dataset cleaning, news app development, or otherwise. (Indeed, one of the pain points we've often heard expressed is that journalist-engineers can feel like they have too many concurrent projects and/or report to too many people!)

Most data teams default to a preferred stack of technology, though the stack in question varies by publication. While many tools deployed by data teams are open-source (and, by extension, readily adaptable by a broad array of organizations), it's common for data teams to regularly build and deploy custom, in-house code -- whether project-specific or for the purpose of generating newsroom templates for future use. And, inevitably, each project is different. As one survey respondent told us: "We've got standards for data integrity and bulletproofing but there's no 'one way to do it' for interactive data projects, any more than there is 'one way' to write a story."

So: There's no one way to build a story, and, honestly, the tl;dr of this guide is that there's no one way to build a data team. But just because this section is full of truisms and clichés doesn't mean we don't have anything marginally useful to say elsewhere on this site! If you're tasked with recruiting this kind of talent to your newsroom -- or you want to rethink how your own team is organized -- the rest of this guide has some concrete ideas for thinking through these organizational processes.

The next section introduces readers to the archetypal roles one might come across on a data unit -- and how to hire for that talent in the first place. The following section, called Structure, offers a survey of modes by which data teams tend to be successfully organized and positioned with a newsroom. Section IV covers the Technology and Tools that data teams most frequently deploy. The final two sections concern organizational strategy and sustainability: Value addresses the question of how to articulate the value of your team (and think about growth), and Fundraising covers what it sounds like it will.