AI agents, Bloomberg automating 1000's of stories, Podcast downloads dead as a metric, and more...
Some cracking stories this week, including a special WNIP feature: 'Everything You Need to Know About AI Agents.'
A shout out to today’s sponsor, Mocono, the all-in-one subscription platform that helps small to medium-sized publishers monetise content with customisable paywalls, subscriber management, and powerful analytics. You can read our interview with Michael Northcott, Mocono’s Founder, here.
OK, let’s dive in…
This week’s top publishing stories
Bloomberg automating over 1,666 stories a day
TL;DR: Out of the 5,000 stories Bloomberg posts each day there is AI involved in more than a third. Key quote: “Breaking news will be more perishable than ever and ingested into machines like ChatGPT to become immediate general knowledge.” Could a single hallucination disrupt an entire financial market?
Mirror journalists given individual online page-view targets
TL;DR: Reach's editorial leaders say that page views are the most valuable metric, as they drive the online ad revenue critical to the publisher's future success. Journos fear, amongst other things, that these will be used as a metric in any future rounds of job cuts. Tsunami of clickbait inbound.
Dow Jones new onboarding strategy: App downloads
TL;DR: Research found subscribers who downloaded the WSJ app in the first two months had a 28% higher retention rate. By adding a QR code to the app download card, app downloads jumped 181%. Brilliant two-pronged strategy.
Product strategies: what publishers should be prioritising in 2025
TL;DR: Wessenden’s Jim Bilton sits down with Media Voices’ Esther Kezia Thorpe. Key quote: “The enabler of everything is tech stacks, and the people who are delivering some really interesting stuff at the moment have been investing in tech for a few years to come up to where they are now.” Two of our favourite analysts.
Reuters Institute: Journalism, media, technology trends & predictions 2025
Nic Newman and Frederica Cherubini’s yearly magnum opus lays bare the challenges facing publishers. But opportunities for news publishers do exist. Tip: don’t read it before bedtime if you want to sleep well.
China weighs sale of TikTok US to Musk as a possible option
TL;DR: In one scenario discussed, Musk's X would take over TikTok US and manage both platforms together. With over 170M US users, TikTok would strengthen X's push to attract advertisers. US ban comes into effect Jan 19th, developing story...
Community Notes are flawed, but that’s their strength
TL;DR: Fact Checkers Vs Community Notes is a divisive issue, yet we can all agree that Zuckerberg’s pivot last week was a masterclass in compromised ethics and political expediency. This piece from Glide’s Rob Corbidge is a balanced take.
New Research Shows AI Strategically Lying
ICYMI Slipping under the radar over Christmas > Previously the stuff of science fiction, a new paper reveals that a version of Anthropic's AI model, Claude, strategically misled its creators during training to avoid being modified. We can’t say we weren’t warned.
AI TOOL OF THE WEEK: Polygraf.ai
Polygraf AI’s Content Detector will spot if text has been written by humans or whether it has been generated/modified by AI. It offers AI content identification, source detection, humanised text analysis, and writing style assessment. Free and premium plans. Using AI? You can’t hide.
EVENT OF THE WEEK
Join MACMA's ‘Interactive Keynote Leadership Forum’ TODAY at 5 PM UK / 12 PM ET, featuring Almar Latour, Publisher of The Wall Street Journal & CEO of Dow Jones. Almar believes 2025 is a year of optimism for publishers. Open to all.
Step Aside 007: Meet 008, the AI Agent
Last week I attended a talk by a senior Google AI Dev (in a small provincial bar in Portugal of all places) who stated that AI Agents would be a primary focus for the tech giant in 2025.
Why does this matter? The implications for publishers are vast.
Unlike traditional large language models (LLMs) that only process queries, AI agents can gather information in real time, analyse it, and take autonomous actions to achieve specific goals.
Imagine an agent detecting signs of subscriber disengagement, such as fewer logins. The AI Agent could trigger targeted campaigns—personalised emails, special discounts, or feedback forms—without requiring any human input.
As Ethan Mollick notes, "Practical agents, at least for narrow but important areas, are now viable. We’re looking at capabilities that could transform many knowledge-based tasks….”
Google has released a comprehensive whitepaper on AI agents, and I’ve penned a TL;DR breakdown of the key insights, along with resources to help you take it forward.
P.S. I took the Google whitepaper and (with the help of ChatGPT) ran it through the HeyGen AI video editor. You can view the result here.
Podcast Downloads Aren’t Enough Anymore
Chris Stone, Head of Podcasts at the New Statesman, recently reviewed his 2024 metrics and believes he has cracked the ‘podcast data code’. Writing in Podcast Strategy Weekly, Chris argues that downloads stats are flawed – they don’t reveal who’s really listening, for how long, or even whether they’ll come back.
Chris suggests ranking metrics by importance (e.g. total listeners > listen time > average listen-through > downloads), and then weighting these metrics to evaluate episodes more accurately.
It’s a better measure: Chris found that some of his podcasts with mediocre downloads vastly outperformed in overall engagement. It turned out the New Statesman’s top performer by downloads didn’t even make the top 10 by weighted score.
This has huge implications for any podcast publisher’s content strategy - what you think are winning episodes might be nothing of the sort and vice versa. P.S. If you publish podcasts, Chris’ Substack is a must read.
Got this far?! Then, as one of our core readers, you are invited to join our WNIP LinkedIn Group. QR Code below:
Until next week….