I was going to write about how media companies need to focus on clearly identifying and then serving their audience, but Adam Tinworth has already written a brilliant post on that subject with great examples including Sarah Marshall’s Audience Canvas and Dmitry Shishkin’s User Needs framework. (Some readers thought Adam was throwing shade on analytics and revenue so he has responded to those who came away with that impression.) Adam was responding to comments from Isabelle Roughhol, who was “seething” about comments about AI at the latest News Rewired. Her comments highlight the split between media companies still playing the volume game and those shifting towards value.
Media companies addicted to the volume model continue to search for ways to maintain that scale by any means necessary to sustain an ad-sustained business. These large newspaper groups and increasingly consolidated digital businesses cobbled together from the tired remnants of faded Platform Era darlings focus on using AI primarily for the efficient generation of content. I would list them, but the restless distressed asset trading amongst them would only be a snapshot in decline. Audiences are (mostly) anonymous numbers to be aggregated. Hedge-fund-owned newspaper groups in the US discuss growth in the context of their ability to pay off the debt they accrued by buying properties to achieve a scale that has never delivered sustainability.
On the value end of the media continuum are companies focused on valuable niche audiences or broader companies that can generate sufficient income from reader revenue. They focus on distinctive content and excellent user experiences. They know the audiences they serve and leverage analytics, audience research and experimentation to understand the value they can deliver to those audiences. Value publishers can operate at a range of scales, from local to national to international, and they can operate in lucrative verticals such as the Financial Times or City AM or in general news such as the New York Times or South Africa’s Daily Maverick.
And now we come to AI. Adam referred to a tweet by Isabelle Roughhol:
Isabelle told me after the session that she was “seething” because a panel member had talked about AI non-strategically exhorting publishers to get on board.
For volume publishers, they focus on how to use AI to create more undifferentiated content more efficiently. It might be using AI to repurpose a story across their network, diluting the value of the original story to generate cheap pageviews for low-margin ads. It might be for repurposing a story for a younger audience instead of a smart strategy to include more young subjects in their reporting and create authentic content for those audiences.
This is largely a hangover of what I call the Platform Era and Brian Morrissey calls the Traffic Era. He said in The Rebooting newsletter, he says:
In retrospect, the traffic era was a lot like the zero-interest rate policy era: It led to a lot of bubbles. The easy-traffic era created incentives for publishers to push out as much content as possible to feed the algorithmic machines at Google and Facebook. Times have changed. And just as the overall economy has struggled to adjust to a higher-for-longer era of interest rates, publishers have needed to adjust their strategies.
Brian Morrissey, The Rebooting
I have worked for and with volume publishers, and I was always impressed with their agility. However, as my understanding of the media business matured, I realised that they were tactically nimble but strategically paralysed. They rolled out new initiatives frequently, but they were always in the service of the same goal, scale. Their AI initiatives are in danger of following the same pattern.
Changes in technology - the internet, social media, the shift to mobile and now AI - all shift journalism’s value chain, where media companies add value. The internet was just the latest technology to change the value chain of media distribution, especially for local news. I used to enjoy looking at the back issues of the newspapers I edited in the US, and I was amazed at how the front pages in the 1960s were dominated by regional and national wire copy. The front pages of major events like the first moon landing were incredible snapshots of history, but I am sure most readers had seen the news elsewhere before they saw in the newspaper. It might have made a nice keepsake, but it was of limited news value.
In London, people used to read newspapers on the Tube - Metro and the Evening Standard, which were sold or then handed out for free to commuters. Even before the Elizabeth Line and increasingly the Tube lines had mobile internet access, commuters spent more time on mobile phones. Commuters were no longer looking to newspapers to read on their commute. The Evening Standard’s daily circulation dropped from 850,000 to 275,000 in the last five years, and the Standard published its final daily edition in September of this year.
With respect to AI, Ezra Eeman, the strategy and innovation director at Dutch public broadcaster NPO, said AI should be used to “create value more efficiently rather than replace humans”. This is the fundamental divide between volume and value publishers. Volume publishers look at technology change through the lens of their current value chain. Value publishers consider the value they create for their audiences, and they consider how new technologies change audience behaviour and expectations, which ultimately change where they can add value.
Some of the larger value publishers, like the New York Times and the Financial Times, already employ data scientists working with editorial so that they can do incredible investigations that would never have been possible without AI. The New York Times used AI to examine 2.1 million posts from thousands of Instagram accounts of young girls managed by their parents. Nearly one in three preteen girls list becoming an influencer as a career goal. “The Times found, encouraging parents to commodify their children’s images. Some of the child influencers earn six-figure incomes, according to interviews.” The Financial Times “compared photographs of the children from an official database of missing Ukrainian children with the public profiles of children up for adoption in Russia using image recognition tools”.
AI is being used for moderation to allow smaller teams to manage comments and other community features to help deepen their relationships with audiences. AI is being used to understand the propensity to register and subscribe to drive better business outcomes. AI will have a huge impact on the business and practice of journalism. Only by thinking about how it will change value chains will we create sustainable businesses. How will AI change where and how journalism adds value for audiences? Those who are asking and answering that question will thrive in the future. Those who simply use AI to eke out efficiencies for a previous era in digital journalism will continue to fade and destroy a lot of economic and civic value. It is this ongoing destruction of value that makes me, like Isabelle, seethe.
And now onto the links for this week. Many of the conversations I am taking part in touch on the best model for publishers to strike deals with AI companies. Smaller publishers are concerned they are being left out.
I wasn’t surprised to see Canadian publishers suing OpenAI. After the disastrous delisting of their content by Facebook, I can see why they want to take a more proactive approach.
One important point made in one of the discussions I took part in is that all of the ad hoc deals are not leading to comprehensive policies to address the IP issues around LLM content scraping. And this example shows how ad hoc these deals are. Dow Jones used lawsuits to protect its own content but then decided to strike AI deals for its Factiva service.
For large-scale value publishers like the New York Times, they are constantly working on how they prove the higher value of their business. While they have one of the most successful subscription businesses in the world, they are looking for an ad measurement model that more closely aligns with their overall business. They want a better way to measure the value of their campaigns.
In the wake of the US election, the Pew Center released a report showing how many Americans were getting their news from influencers and what kind of information they were getting from them. UNESCO found that most of these influencers didn’t check the information they were circulating.
News and resources to help you navigate the rise of Bluesky
If you’ve shifted to Bluesky, there is now a way to verify your accounts, either through domains you control as well as through your employer’s domains. It’s quite powerful, and it won’t require you to pay a subscription fee every month.
Bluesky is growing rapidly, and on its current trajectory, it could overtake social networks like X. To do that, it’s going to need money.
This is a fascinating question, but I also wonder about the role that high levels of inequality play in these perceptions.
This newsletter started years ago when I used Nuzzel to aggregate the links that my network of people in media and journalism shared. It was incredible that I could quickly see the most important links my network was sharing, and it made putting together a newsletter like this much easier. I was happy when Twitter bought it and gutted when the company shut it down. Now, there is a similar service for Bluesky.
