Collaborating and Being Informative (in-house) and AI (It’s Coming) Should Help Guide Your Metrics

The next leap in metrics may be AI-based automation. The Globe and Mail has built a prediction engine they call Sophi that analyzes data on article performance to key the homepage and landing pages. It also uses AI to determine which stories users will pay to access and which can deliver more ad revenue. But even with that day coming, metrics should remain a collaborative affair.

I mentioned last week that at our Editorial Council meeting at the Neal Awards, one editorial director recounted a story about a new reporter who kept telling her about his articles’ analytics, who was reading and other metrics. She literally reveled in this feedback, especially since in this remote age, it can be a big challenge to get new people on the same page.

This reminded me of the Lessons From a Leader my colleague Amanda McMaster conducted with Lucy Swedberg, executive editor and senior editorial director at Harvard Business Publishing. “One of the best practices that we’ve embedded is an analytics meeting for our editors, so that they can really see their work and how it ends up performing out in the world,” Swedberg said. “You start to hear them thinking and observing, ‘Oh, this thing did really well—let’s do more of those.’ I love when I hear they’re getting insights from the data. Because, at the end of the day, that’s what will keep us going and [allow us to] make an impact.”

In a Reuters Institute report—Overcoming Metrics Anxiety: New Guidelines for Content Data in Newsrooms—Elisabeth Gamperl, managing editor, digital storytelling unit, Süddeutsche Zeitung in Munich, looked at narratives around data analytics. Ultimately, she writes, analysts should be seen as vital members of the newsroom.

That train—if not already here—is pulling into the station. Here are more ideas on metrics and analytics from the report and other sources.

Don’t overwhelm – find your key metrics.
In article recently on The Fix, David Tvrdon argues that putting too many metrics on your journalists’ plates could be risky. “With every added metric the chance of more people not getting it simply rises exponentially. I would rather use a simplified metric and tweak it in time than risking colleagues in the newsroom having different goals.” The Financial Times used RFV (recency, frequency, volume) to help hit one million digital subscribers. Later, they pledged allegiance to a more consumption-based Quality Reads.

Adjust to what machine learning can do.
“Even with some of the more effective paywalls or data walls, most of the time they’re ‘a one size fits all’ or ‘one size fits a segment’ thing. I think what we’ll see is much more automated, AI-driven single user journeys,” said Gabe Karp, EMEA director at digital agency 10up, in the PressGazette. “So if the machine can figure out that I only read one article a day from The New York Times but their subscription is valued based on me reading five articles a day, can they give me a different offer at a different price point?”

Be positive and concise.
One analytics team developed a list of questions they work through before submitting data to the newsroom. Leaders also advise to be careful in sending around individual rankings or standings. Instead, promote information on screens that is helpful to the newsroom. For example, “Did you know that most people read us between 6 and 8 am?” And be concise. “If you provide too much, it has a counterintuitive effect of making people less engaged with it because people don’t know where to focus. It becomes a little bit overwhelming and disengaging to just see reams and reams of data,” said Jörn Rose, head of strategic growth & insight, BuzzFeed.

Focus on measures that support your editorial and revenue model goals.
“If your goal is audience growth, you must start measuring new users. If your objective is to generate more subscriptions, perhaps you should consider measuring conversion journeys in more detail, from anonymous to registered readers,” writes Gamperl. “Media outlets with subscription models pay attention to engagement metrics: time on site, pages per session and bounce rate (or the percentage of readers who visit a single page and look at nothing else on your website before leaving).

Don’t look at metrics as static and immovable.
It should be an ongoing process—and include a positive feedback loop. The question should not be: What is the number? But rather: What can you do in response to this number? A poor-performing story might be repackaged in another context. “If a story should work and it doesn’t, we try to look at the presentation, change the headline, change the picture and publish it again at another time,” one editor said. Another editor added: “We have as many open conversations about when things haven’t worked as possible without everyone getting really upset. That is not easy because people work incredibly hard in the newsroom.” What lessons can be learned?

Look at opens vs. “dwell times.”
If an article has a high open rate but a low dwell time, then it might be a “one-fact” story, or the headline “missold” it. On the other hand, if opens are low but the engagement with the story is high, then maybe people aren’t finding it. Try plugging it on social media or giving it a better position on your website. “Working with metrics is all about trial and error, adjustment and retrial,” Gamperl writes. “Every failure is a step closer to success.”

You can download the report here.

Comments are closed.