When it comes to editorial analytics, it’s not enough to just report data. Collaborating effectively with the newsroom means thinking like a journalist and finding the story the metrics tell.

On module four of five of INMA’s Product and Data for Media Summit, data leaders from Dow Jones, Axel Springer, and MuckRock detailed how they simplified data for newsroom teams, encouraging journalists understand and make use of important data. The summit continues today.

Dow Jones creates the Data Academy

When Helen Hewitt, head of design operations at Dow Jones, was asked to be part of a small internal team addressing the way the organisation used and interacted with data, she was initially sceptical.

“I thought, ‘I don’t know anything about data, it’s not really my thing,’” she told summit attendees.

Hewitt soon found she was so passionate about data that she lobbied for a new role within the organisation, complete with a team and funding. She shared with INMA members the process Dow Jones went through to build a data training programme, from identifying the problem to launching a multi-part solution and then tracking learnings.

The challenge for Dow Jones, Hewitt explained, was that although they had invested heavily in data and made great headway, there was no “holistic view of how data was used across the organisation.”

It’s an enormous, global entity with more than 80 offices scattered around the world, she said, that had a “complicated and siloed structure” with both B2B and B2C products. It wasn’t particularly surprising that there were lots of teams using data in lots of different ways — but that also left a great deal of room for improvement.

The initiative was driven by the CEO, Hewitt said, who was extremely passionate about it.

“His belief was to really use data as an enabler to help us tell better stories, realise new commercial opportunities, and make it easier to do our jobs.”

A Data Task Force was created, a cross-functional team of six people representing different sectors of the company, which worked offsite on this project for eight months. The team was deliberately removed from their day jobs, offices, and even cut off from e-mail so they wouldn’t get distracted.

For the first two months, they interviewed 100 people (80 internally and 20 from outside the company). Hewitt said it was “critical to hear people and really understand what they’re trying to achieve, what data means for them,” and that a human-centred approach was key throughout the process.

At the end of the eight months, they delivered a set of recommendations to the executive team, focusing on what they called the “three Cs”— capability, co-creation, and culture.

Focusing on the “culture” aspect, Hewitt detailed four huge opportunities the team found as a result of the interviews they did: 

Dow Jones then launched three new programmes, which Hewitt outlined for INMA members: 

 
“Data literacy significantly increased,” she said, “and conversations around data changed — it was in more conversations. People were talking about data in a much more confident way.”

Axel Springer team is a bridge between data and the newsroom

Janis Kitzhofer, head of editorial analytics for Axel Springer, likens his work to starting a bonfire. Although it would be possible to make do with only matches and logs, having some fuel is going to be much more effective and quicker. 

He thinks of the work of his six-member team as the fuel that gets the company’s journalists inspired by and engaged with data. He started out as a journalist himself and believes that, like journalists, data specialists can use the information they gather to tell stories about how editorial content performs. 

Axel Springer is a global media company with its strongest presence in the German-language market, but its 30+ brands, including the recently acquired Politico and Business Insider, capture some of the English-speaking market, too.

In Germany, the company publishes the top two daily newspapers Bild and Die Welt. These two flagship publications have very different models which require different approaches to digital publishing.

Bild, printed as a tabloid, is very visual, and 80% of its Web traffic is direct, so its Web site appearance is key. Die Welt stories are more political and and analytical, and most of its traffic comes via search. 

Across Axel Springer, data analysts are on the business side along with the tech team rather than in the editorial departments. The two sides had difficulty communicating and so Kitzhofer’s team was recently established to be a bridge to the newsroom. 

The relative small team has a broad skill set. There are data scientists to focus on visuals and tracking as well as dashboard specialists, and a tracking “healer.” Kitzhoker thinks of himself as the “story slasher” who takes everything they learn and carries it to the newsroom. 

The team’s job is not to just generate a lot of data but to use it to inspire the editorial team and to work in a collaborative way. To make that work, Kitzhofer thinks it’s important to be approachable, to try to really understand what journalists need to do in their work, and use the data to educate and empower them, all while evolving in a constantly changing landscape.

In practical terms, education takes the form of daily reports and live and ad hoc analysis, monthly and weekly performance presentations, and less frequent but more involved in-person workshops that are good for generating meaningful interactions that spark creativity.

To empower the newsroom, live analytics really help journalists become more data confiden, Kitzhoker said. Knowledge is power, and it’s important to really make everyone embrace this and see the advantages.

In most newsrooms, there are some people who are particularly interested in working with data. Kitzhoker finds such folks at his in-person workshops. Then he assembles “secret teams” to give extra training as well as to get feedback and suggestions about the editorial team’s needs. 

Every newsroom culture is different, Kitzhoker points out, and everyone has to find their way to work together effectively. But in his experience, a collaborative approach has been the best: “Make your team feel like a part of their team.”

MuckRock uses AI, machine learning to gather usable “boring” information

When it came time for MuckRock to carve out its niche in the automated world, it used a bold and unexpected strategy. While most companies focus on “the shiny fun stuff,” MuckRock founder Michael Morisy told INMA members it decided to look at “the really boring stuff that nobody wants to do.”

Morisy went inside his company’s journey to create technology that centred on usability. 

In 2018 the non-profit company merged with DocumentCloud, which helps users share documents and works with about 4,000 newsrooms around the world. Its core product allows users to upload PDFs, search and analyse the documents, then post them online.

One of DocumentCloud’s flagship features is the searchable and accessible nature of its documents.

“It helps newsrooms become more transparent with their audiences,” Morisy explained. “And it has helped newsrooms back up their reporting in some really exciting ways.”

Newsrooms have found the service useful for both breaking news and for longer investigative journalism pieces, he said. Its usefulness is reflected in its popularity: It boasts about 80 million visitors a month and has processed more than 168 million pages.

About a year ago, it became apparent that MuckRock’s core value proposition — which was the ability to embed a document on a Web site — was becoming commonplace.

“Just hosting documents was no longer enough to just set apart our service in terms of what we are doing for our user base,” he said.

So company started talking about what it needed to do to serve its massive user base, move its network of information forward, and solve for tomorrow’s needs. As it looked at Artificial Intelligence and automation, MuckRock talked with users about challenges they encountered — which led to another discovery:

“A lot of people tackling AI tools and building start-ups in the space have teams of dozens of people,” he said. “We were at our largest technical team ever with three developers. So we had to be very thoughtful and careful about how we deployed those resources.”

Morisy said they recognised there was a tremendous potential with machine learning to change how journalists work. “But we also knew if we tried copying everything or if we tried implementing every feature ourselves, we would quickly spread ourselves too thin.”

That’s when they decided to focus on “the really boring stuff that nobody wants to do,” he said. They discovered there was a major gap in terms of middleware. Although there were a lot of researchers doing groundbreaking work, and many users who could benefit from that work, there weren’t many companies looking at usability.

“You didn’t have a lot of people focused on how to build a marketplace for interesting technology and the people who would actually use it.”

MuckRock realised it could focus on how to provide daily utility for newsrooms.

“AI start-ups love going for the most advanced, challenging problems,” he said. “Those are intellectually interesting but oftentimes don’t meet the needs of day-to-day users.”

Features have included creating ways to monitor and scrape Web sites for information and quickly provide journalists with the information they were looking for. While Morisy said developers found this boring to create, it has been MuckRock’s most popular launch add-on and the company is now experimenting with some “really exciting partnerships” that will aid newsrooms in the future.

“By solving for the boring stuff, we were able to take this exciting technology and deploy to a wide number of users very quickly,” he said. While those core tools and features may be less exciting to AI developers, it could pay off for news media companies.

“I think there’s a real opportunity for a variety of publishers and people building tools for publishers because there’s so much interesting stuff out there in the world of AI and software development and machine learning,” Morisy said. “But finding how to make that easier to integrate into workflows is a really untapped challenge. And I think that’s going to be what helps open up the potential for so many end users.”

The Product and Data for News Media Summit continues on Tuesdays and Thursdays through November 17. Registration continues throughout the summit and on-demand recordings are available.

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .