Data metrics are transforming journalism in deep and rapid ways. Journalists today are inundated with data about which stories attract the most clicks, likes, comments, and shares. These metrics influence what stories are written, how news is promoted, and even which journalists get hired and fired.

A new book by Caitlin Petre, All the News That’s Fit to Click, gives a behind-the-scenes look at how performance analytics are transforming journalism today—and how they might remake other professions tomorrow.

Petre joined INMA for an exclusive INMA Meet-Up as part of the Smart Data Initiative, to discuss this topic with INMA Smart Data Initiative Lead Greg Piechota. Joining them was Alan Hunter, co-founder of HBM Advisory and the former head of digital at The Times in the U.K.

The Meet-Up couldn’t have come at a better time. In spite of the obvious impact data is having on newsrooms around the world, as Piechota pointed out, many journalists and editors are still skeptical about its use, and some media companies are not adequately transforming their company cultures.

“There are many challenges that make some of the newsrooms struggle a lot,” Piechota said about implementing the use of data. “This is a very insightful book.”

The goal of the INMA member Meet-Up was to address and answer a number of important questions with Petre and Hunter.

How data metrics are changing journalism

A key question to this discussion is: Should newsrooms be data-informed or data-driven?

In Petre’s experience, it’s rare to hear anyone say that data should drive the newsroom.

“It seems that there is something like a consensus emerging that newsroom decisions should be data-informed,” she told INMA members. “But I think it’s important to be crystal-clear about what that means and what it doesn’t mean.”

Hunter concurred. In his experience, it would be very difficult to make newsrooms driven by data first and foremost, given the nature of journalism.

Petre wrote the book because she wanted to study the role of data metrics in news organisations. A sociologist by training, she is driven by comparisons and wanted to look at case studies of two very different newsrooms when it came to their approaches to data.

“Crucially, I wanted to make sure they were both newsrooms that were using the same analytics tool,” she said. “I looked specifically at Chartbeat.”

The two news media companies she compared as case studies were Gawker and The New York Times. One thing she found was that people had a lot of anxiety about metrics resulting in “clickbait” journalism.

“It was almost [a fear that] this would be an automatic process: The dashboard comes into the newsroom, and suddenly all the news becomes about Kim Kardashian or Megan Markle.”

However, Petre found the use of the same data metrics tool varies greatly between newsrooms, depending on their different cultures. For example, even though Gawker and The New York Times both use Chartbeat, the data was used in vastly different ways.

At Gawker, a digital news start-up that was always transparent about using metrics behind everything, journalists were immersed in real-time data almost 24/7. As a result, Petre said there was a speed-up of work productivity.

“Journalists still wanted to do the stories that they cared about, but they also felt this pressure to boost their traffic numbers and compete with their colleagues,” she said.

The New York Times, on the other hand, used Chartbeat and analysed the data gleaned from it, but the company was reticent about rolling out dashboards in its newsroom, which was very traditional, hierarchical, and editorial.

“Editors almost hoarded the data,” Petre explained. “But reporters at The Times did not when I was there. Even if a reporter wanted to know how many pageviews a story got, they would not readily be able to get that information.”

Her take was that metrics can sometimes pose a threat to editors because of the decisions made from them — decisions that have traditionally been editorial ones to make, not data-driven choices. At the time of her study, editors wanted to keep control of the story that metrics were telling.

Piechota asked Hunter how The Times of London responded to metrics in the newsroom when he was there.

“I think data in the newsroom is a good democratising force because too much journalism is ‘handed down from the mountain’ by people who very often look and sound a lot like me,” he said — meaning white males, showing a lack of diversification and representation.

The resulting news coverage, therefore, isn’t necessarily representative of what greatly varied readership groups want to consume.

“If you follow the data, in my experience you  would not be doing as much on climate,” Hunter said. “Because those stories, unless they’re about what individual readers can do to improve the environment, they don’t do particularly well. What data can do is help point you in a better direction and better serve readers.”

That is what his team at The Times of London was trying to do during his tenure. In practice, data was not driving the editorial decisions, but it was certainly informing them. As a result, they produced less content, but that content was engaged with far more.

Hunter added that Times’ editors weren’t giving reporters the data, but rather insights based on the data — a clear distinction. The editorial content produced as a result, in fact, was the exact opposite of clickbait. It was content that resonated with the readers.

How can news media organisations respond?

As Petre explained, news media organisations tend to focus greatly on the impact of metrics on news content — which is important — but they also need to look at the impact they make on journalistic working conditions and on the sustainability of journalistic work at that level.

When it comes to reader interest, this is a very complex thing, she said. And metrics reduce it to a very small set of circumscribed behaviours. For example, readers may say they want more content surrounding climate change or international news, but the metrics show a different story: that the readers aren’t consuming those stories in great numbers.

The reality is a much more complex picture than metrics alone can provide. Using herself as an example, Petre said climate change is an important topic that she wants covered — even though she might not read every story about it on a regular basis.

“I don’t click on every story about climate change that I see. But do I want it to be a top-level news item for months and months, until our world leaders are pushed to really take action? Absolutely. That kind of feeling is not captured by the metrics that we have now.”

When metrics are extremely prominent in the newsroom, these other aspects can get drowned out, Petre warned.

“One of the ways to avoid that is to try and make the metrics a little bit less immersive,” she said, referencing Hunter’s experience at The Times of London as a good example: using data as insights to inform the newsroom, without driving it.

Pageviews are overrated as a metric for which stories are doing well, Hunter said.

“They are the least interesting metric as far as I’m concerned. We never talked about pageviews or traffic. Say you want to cover climate change — what data can do is tell you how to cover it in the most effective way, to make it most attractive to readers.”

Today’s metrics give more opportunities to discover what readers really want and help newsrooms incorporate that into the journalism. Engagement metrics such as time spent reading an article, commenting, and sharing are more important indicators, Hunter said.

What about automation?

Piechota asked Petre and Hunter how newsrooms can automate their work and what work shouldn’t be automated.

Petre pointed out that much automated content includes financial news, real estate, or sports. The argument is often that it frees up journalists’ time to do the type of reporting that can only be done by a human.

However, too often, those extra resources aren’t actually kept in the newsroom, and staff that previously covered those automated topics may be considered unnecessary. Therefore, she is skeptical of that line of reasoning.

When it comes to tools that can help with good automation, she pointed to A/B testing as a most effective one.

“But at the same time, I’m skeptical of the notion that [automation] is going to be this great thing so that journalists don’t have to do the boring stuff, because the robot can do it, so they’re freed up to do more civically important reporting,” she said.

Hunter echoed this, saying most content that can be successfully automated is not the kind of coverage that tends to get readers excited.

“It’s stories based on data that is publicly available,” he said. “From my point of view, automation is most likely to be used on the back end of the story process — in fact-checking and stylebook-checking and so on. The kind of journalism that journalists really want to do is the kind of stuff that can never be automated.”

Princeton University Press, publisher of All the News That’s Fit to Click, has provided a discount code for INMA members to receive a 30% discount off the book. Visit the Web site and enter the code PETRE at check-out.

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