Publishers are figuring out how to generate revenue with data, machine learning
The covid-19 crisis has accelerated trends away from advertising and toward reader revenue for news publishers. And now everyone is talking about how to use data from online reader behavior and tastes to drive the business models for quality journalism.
At an academic conference in Spain that I attended virtually last week, SEPSevilla 2021, a parade of studies by academics and industry representatives emphasized the importance of putting the readers or consumers at the center of the business model. Lots of slides showed some version of a reader icon surrounded by news, platforms, personal networks, and brands, all vying for their attention.
Keynote speaker at the conference was Pepe Cerezo, managing director of Evoca Media, who set the tone with his presentation that showed signifcant increases in reader revenue while advertising revenue has plummeted.
“There is no magic formula” for replacing the broken media model that depended on advertising, he said. However, he noted that the freemium subscription model, which keeps some high-value content behind a paywall, is the one that is dominating in Europe.
The #metoo and #BlackLivesMatter movements are driving publishers to increase diversity on their staffs and in their contents. And, by the way, publications with this diversity are also more profitable, as they attract new audiences, Cerezo said.
Loyal users, not masses of users
Noemí Ramírez, chief product and customer officer for Prisa Group, which publishes El País, said that of the 82 million monthly users of their website, at least 90% were casual users, not loyal, and thus unlikely to ever subscribe.
She emphasized the importance of building a relationship by steps–first getting users to register, then making sure that they are being served content that interests them enough to return frequently, and finally to ask them to subscribe. All of this is driven by user data that continually learns how to predict the best strategies for monetizing reader attention.
A similar story came from Javier Martínez Gómez, director of subscription development for La Vanguardia, another leading newspaper in Spain. They use geo-location data to serve each reader news about their particular town or barrio. Local news is a big driver of subscriptions, he said.
If a subscriber fails to visit the website for six or seven days, data shows that this person is more likely to cancel their subscription, he said. So the data-driven system sends something personal to maintain the relationship.
A reason to take a risk
Maybe publishers just needed an excuse like the covid-19 crisis to take a risk. The Reuters Institutes’s survey of 234 media executives from 43 countries–Journalism, Media, and Technology Trends and Predictions 2021--found that three-fourths viewed driving digital subscriptions as important or very important this year. This was the first time subscriptions ranked higher in priority than advertising in this annual survey.
Three-fourths of those surveyed also placed importance on audience and data insights (74%) as the best way to generate new ideas, which fits with what we heard at the conference.
Also interesting was that almost half (48%) selected “learning from other media companies” as an important way to innovate. This fits with some of my own research presented at the conference on a growing trend toward collaboration rather than competition among news media organizations to develop viable business models.
Data, data everywhere
Once you begin looking for articles about data, machine learning, and artificial intelligence applied to journalism, you discover an explosion of innovation.
One example: the Crosstown Neighborhood Newsletter project creates and sends news items customized for residents in each of 110 Los Angeles neighborhoods. University of Southern California Professor Gabriel Kahn is the editor and publisher, but he partnered with the school’s engineering department to develop the newsletter.
Here is how Kahn describes it: “Our formula starts with data. We collect data about everything we can in Los Angeles, from traffic and crime to COVID-19 cases and building permits. Much of this data is hiding in plain sight, housed on local government dashboards that are hard to navigate. We divvy up the data by neighborhood. One citywide dataset about parking fines becomes 110 stories about how many more or fewer tickets were issued in each neighborhood during the COVID lockdown.”
“We automatically parse the relevant data into each one of the neighborhood newsletters and we place it in context. Boyle Heights, it turns out, is currently the epicenter of car theft in Los Angeles, while last year Chatsworth put up more new buildings than anywhere else.
“This is not just the work of journalists. It’s a collaboration between software engineers, designers and journalists. I doubt it could have been pulled off anywhere other than a university, where there are so many different areas of expertise to pull from and so many eager and talented young people willing to work on thorny problems. We were fortunate to get funding from the Google News Initiative.”
The team has also created a dashboard where audiences can search for statistics specific to their neighbourhood on crime, air quality but also coronavirus cases and deaths, and how these compare with other neighbourhoods. (This slidedeck explains the service better visually.)
More reading on AI and audiences
The latest is from the UK’s Journalism site, “How to identify new audience growth opportunities.” By identifying data of interest to particular audiences, a local news site can increase reader engagement, loyalty, and propensity to subscribe of pay for other related services.
Also: Automated Journalism: AI Applications at New York Times, Reuters, and Other Media Giants.
Here is a synopsis of a new book recommended to me by an esteemed colleague, All the News That’s Fit to Click: How Metrics are Transforming the Work of Journalists.
Put readers at the center of everything