AI Overviews Jump 3.5x, Paywalls Fracture, Condé Nast & Hearst Agree Amazon AI Deal...
Welcome to this month’s sponsor, Decoded Labs, founded by two former publishing CTOs who work closely with media brands to improve platform performance and resilience. If your site is fragile, slow, or struggling under increasing AI complexity, Decoded identifies the underlying issues and helps implement long-term tech stability—whether it's related to integrations, AI tooling, or scaling challenges. Read more here.
Today’s Long Read comes from Rob Toth, CEO at Songbird.Group, and a veteran advisor in publisher M&As. Drawing on data from dozens of deals, he makes a blunt case: the old SEO-playbook is finished, and media brands must implement new strategies now. Or die.
Let’s crack on…
Publishers’ Keywords 3.5x More Likely to Trigger AI Overview vs March
A new report by Enders Analysis shows the speed at which AI Overviews are infiltrating search results. The estimated drop in total traffic is still small (2%), and the commercial impact minimal (for now), but its trajectory looks set to become an existential threat to discoverability. Entertainment & info queries currently dominate AI Overviews.
How AI Bots Quietly Dismantle Paywalls via Web Search
ChatGPT and AI chatbots have figured out how to get around paywalls through ‘live’ web search, piecing together fragments from social media posts, archived sites, and secondary coverage to rebuild complete articles they've never seen—and they're doing it systematically and quietly across major publications. The findings come from publishing veteran, Henk van Ess [ex-Axel Springer, Google, Poynter, etc].
Appeals Court Strikes Down ‘Click-to-Cancel’ Rule
The U.S. appeals court has thrown out a regulation requiring subscription services to offer an easier way to cancel. The FTC's click-to-cancel rule was due to come into force this week but 'fatal deficiencies' sank it. P.S. NY and California already have a similar rule, and some publishers implementing it haven’t found it to be the catastrophe they originally thought. Case in Point: Minnesota Star Tribune.
6 Tips for Protecting Your Content in the Age of GenAI
As a leading authority on content licensing, CLI's Paul Gerbino has skin in the game—but the legal picture still remains messy with Thomson Reuters v. Ross Intelligence in one corner and Meta/Anthropic in the other. Gerbino lays out six concrete steps publishers must take now. It's a solid list of 'dos' although robots.txt is as effective as a chocolate teapot.
WhatsApp Channels: The Promise Is There, But Problems Remain
Many publishers (list here) are seeing huge engagement with WhatsApp Channels, offering as it does algorithm-free, direct reach. But the same issues plaguing it since its 2023 launch persist: manual posting, no scheduling, no real data (beyond views), no monetisation, and crap discoverability. Apart from that, it’s great. Key quote: “The manual workload is significant.” It’s also owned by Meta, the pivot kings.
4,000 Fake News Websites Created to Game Google Algorithms
In early 2024, French journalist Jean‑Marc Manach identified 70 GenAI news sites designed to game Google Discover and Search. The number is now north of 4,000, with the sites mimicking real media outlets to scrape money via Google AdSense or via SEO backlinking. Bottom line: It's a clear threat to legitimate publishers relying on search traffic. Part of the crumbling SEO model Rob Toth outlines below.
Condé Nast and Hearst Strike Amazon AI Licensing Deals for Rufus
Amazon’s LLM-powered shopping assistant—Rufus—will use media lifestyle content to help answer customer questions and make product recommendations e.g. “What’s the best moisturiser for dry skin?” Brian Wieser warns that if the LLMs are successful, and license enough content, they could become a clear threat, but adds, “It’s better to lean into this space and learn.”
Zero to 10,000 in One Year: How NPB Used WIZA to Build Its Audience
For those growing a U.S. newsletter audience, this Alisa Cromer deep dive outlines how B2B publishers can quickly, safely, and accurately build their newsletter list. It’s a paywalled webinar, but a free month’s trial gives access. Outstanding takeaways. P.S. Non-GDPR compliant, but for publishers growing B2B U.S. audiences, it’s gold.
Alliance for Audited Media: Ethical AI Resources for Media
The AAM has put together a collection of resources to help media organisations apply its Eight Pillars of Ethical AI with practical guidance and clear strategies for using AI responsibly. If you're drafting an AI policy, training staff or reviewing risk & governance, etc, the list offers a starting point for building trust and editorial integrity. The resource is being continually updated.
Superintelligence? Self-Adapting Language Models Are Here
MIT researchers have just developed Self-Adapting LLMs (SEAL), a framework that enables LLMs to improve on their own via reinforcement learning loops—these reward AI for self-edits that lead to better performance. The MIT AI learned more effectively from its own notes than from GPT-4.1. A heads-up for future product teams…
AI TOOL OF THE WEEK: Genuin
Genuin is a video-first platform that helps publishers build curated communities without public metrics or YouTube algorithms. The appeal lies in full control over data, context, and community—it’s also useful for cutting, tagging and distributing 30-second clips from archives, serving up content by theme and audience interest. It’s used by TED.
REPORT: Webinar Benchmark Report 2025
According to Communique’s 2025 Webinar Benchmark Report, delivery matters as much as content. Webinars with custom branding, cinematic video, and dynamic layouts saw engagement jump by up to 450%. Even changing the title to a ‘series’ or ‘masterclass’ lifted registrations by 22%. Some decent hacks.
EVENT FOR YOUR DIARY: Future of Magazine (FoM25) Conference
The biennial Future of Magazine (FoM25) Conference returns to Portugal this October. The theme is ‘(Re)Thinking Magazines: Challenges & Opportunities’ with the event bringing together practitioners and academics for in-depth discussions on magazines. 29-31 October | Lisbon P.S. See you there—please reach out if you’re going!
It’s Not A Google Algorithm Problem—It’s a You and Me Problem
Rob Toth, Founder of Songbird.Group M&A, Shares Insights From 10 Years of Selling Publishing Businesses
For more than a decade, our team has advised, sold, and evaluated digital publishing and media businesses. Before that, I spent 10+ years in targeted advertising and lead generation. That’s a twenty-year timeline of seeing the Internet’s many shifts.
WordPress, which revolutionised blogging and turned the casual web blogs into a real business model, launched in 2003.
Google’s AdSense, arguably the first mainstream programmatic ad network launched the same year.
Roughly 2005 is when the practice of SEO started becoming mainstream.
SEO, WordPress, programmatic ad networks are effectively 20 years old. That’s just long enough for us flawed-logic humans to regard a norm in our lives as having been how things always were – and falsely believe that’s how they always will be.
In approximately that same window of time, we swapped landlines for smartphones (2007), TV viewing time for YouTube (2005) and print magazines for niche digital media. We all stopped driving to DVD rental stores once Netflix streaming launched in 2007.
This isn’t a history lesson. It’s an alarm.
If all you observed in the previous timeline is the new tech, you’re missing the key point. Each new solution we moved to swapped out businesses that were once established norms, and often 20+ year old industries themselves, yet now are obsolete or inconsequential.
If your publishing business still runs on the model of “organic Google traffic + programmatic ads = revenue”, then you own a DVD rental store.
This is not a Google algorithm or an AI Observations issue. It’s a user convenience-seeking problem.
Answer Engines and Voice Prompting
Recently, I was traveling and met one evening for dinner with industry friends. And I observed that three of the four of us, that is 75%, were using some of the downtime “speaking” to our respective AI bots. We were voice prompting queries that would have previously been Google searches and webpage clicks.
I do hobby-level investing and the most core mandate in my investment criteria is the Convenience Multiplier. How much does the new product or service multiply the overall convenience for the user. Smartphones vs landlines, Spotify vs physical CDs, Uber vs manually calling taxis.
I use that as my first, second and third consideration in an investment because if the Convenience Multiplier is significant enough, it trumps just about everything else.
Consider: would you be keen to go back to manually picking up DVDs at Blockbuster vs having a Netflix streaming account?
That is what the industry is already battling. Whether you are letting yourself see it or not.
Nobody who engages with voice prompting, speaking to an answer engine, will be going through the already archaic steps of manually searching Google and clicking through 10 blue links.
And the number 1 thing AI chats replace is content; the product of your media business.
Google knows this. Which is why they continue to soft-boil publishers with updates that shrink their search traffic as Google search transitions into an Answer Engine, AI Observations being just a sample of that.
We wrote about this in an industry insights report in June 2023, three months before the tsunami of Google’s Helpful Content Update (HCU) that rolled out in September.
GA Data
Given our business model, we have the privileged advantage of seeing the data from behind the curtains. A month after HCU hit, we had view of 42 media brands’ Google Analytics accounts. I spent days digging into the stats to see the aftermath.
3 sites of the 42 were untouched.
28 saw traffic drop by 25–40%.
7 dropped 50–60%.
4 collapsed by 75–80% — functionally out of business.
Then, after the March 2024 Core Update rolled out, we dug into the then 48 GA accounts we had access to and compared those stats to pre-HCU data.
100% had a decline.
22 dropped 25–40%.
20 dropped 50–60%.
6 lost 75–80%.
Those six were effectively put out of business, and all were handed a pay cut.
In our June 2023 report, we forecasted this shift, though at the time it felt a bit “Chicken Little” to many. Those predictions are now mainstream headlines.
Media M&A: From Favourable to High-Risk
From our launch in 2015 to 2021, we regularly sold SEO-driven publishing businesses at 5x–8x earnings — even as market comps hovered at 2.5x–3.5x. Some of our premium deals saw 11x, 18x, even 21x EBITDA.
Premium media was an exciting, favourable, market advantage providing acquisition.
Today, if it’s Google traffic driven, it’s high-risk.
Case study: A personal friend owns a once-thriving tech media brand. At its peak in 2022, that site would’ve sold for $20 million USD with competing offers. Today, the same asset would struggle to close at $4 million. Why? Because they remained focused on an already old model: creating more, SEO-optimised content and continuing to play the Google game.
Passive, Build or Sell
There are three ways forward.
First, if your publication is already small, it likely won’t land you a meaningful exit. You can look at cutting back costs, simplifying and letting it run passive. It will depreciate in revenue and value but for smaller sites, it might be a best path forward.
If you have a stronger media brand, then ensure you’re diversifying away from Google traffic by way of content distribution partnerships, newsletters, video-first channels, events and private communities. Or add on relevant transaction businesses such as e-courses.
The most successful clients we’re working with now have a full mix of these: product offerings, private communities, paid ads to supplement their organic traffic, thriving email newsletters.
Alternatively: sell. Whether on a marketplace, a generalist broker or a specialist to media M&A like our team at Songbird.Group.
Some in the industry pursue a fourth option: hope that this is just a Google algorithm blip like the many they’ve experienced over the past twenty years.
Spoiler: it’s not.
About Songbird.Group (SBG): SBG, formerly operating as OODIENCE M&A, is a ten-year veteran and specialist of media and publisher business sales, serving clients on the sell-side to land a maximised exit by selling to hard-to-reach Strategic Buyers || Rob Toth’s LinkedIn.