Media Trendlines — August 3, 2026
📰 Key Themes
- People Inc. grew its digital business by double digits in the quarter its sessions fell 22% and its Google traffic dropped 40% — the clearest sign yet that the pageview is no longer the unit publishers get paid for.
- Time started selling ads written for machines, dropping sponsored FAQs into stripped-down markdown pages built for AI crawlers that no human will ever read.
- The New York Times is treating games as a retention moat, turning Wordle into an NBC show while its Crossplay app posts what it calls world-class return rates.
- Google is folding publisher Top Stories into its AI Overviews, and qualifying may mean accepting AI summaries of your own reporting — visibility and control on the same lever, both held by Google.
- UK publisher digital revenue fell 4.55% in the first quarter, snapping a year of growth, as one industry body called it the “first tremors” of an AI-driven earthquake.
- A Manhattan judge kept Reddit’s case against Perplexity alive on a theory — circumventing access controls — that could hand publishers a second line of defense beyond copyright.
Jump to: 💡 The Business Beyond the Click · 🔍 The Search Squeeze · ⚖️ Rights & Rules · 🎙️ From the Pods · 📎 Also Noted · 🧭 Takeaways
💡 The Business Beyond the Click
People Inc. Is Growing Because It Finally Stopped Depending on Traffic
Source: A Media Operator, Jacob Cohen Donnelly.
The numbers People Inc. put up for the second quarter should be read twice, because they describe a business succeeding by every measure except the one the entire web was built on. Digital revenue rose 6% year over year and digital adjusted EBITDA jumped 18%. In the same quarter, the publisher’s core sessions fell 22%, driven by a 40% collapse in Google Search traffic, and its session-based revenue actually shrank 1%. The arithmetic underneath is the story: revenue per session climbed roughly 27%, and every dollar of digital growth came from somewhere other than a page a reader visited.
That somewhere is now 43% of digital revenue and growing 16% a year. It is advertising sold across a network of vetted domains through the company’s D/Cipher+ product, performance marketing through email and other platforms, and licensing — Apple News, syndication partners, and AI deals. The parts of the business tied to a human landing on a People Inc. page are flat to shrinking. The parts that have nothing to do with a session are carrying the whole thing.
The company is not hiding what this means for its future. Its “Inversion” projects — 19 of them — include a Southern Living branded tea and a possible Southern Living housing community, alongside events, membership clubs for super-fans of Southern Living and Food & Wine, and social-commerce distribution plays like InStyle’s The Intern. Total revenue still slipped 2% to $416.7 million as print fell 16%, so this is a company managing decline in one hand while inventing a post-traffic business with the other. The uncomfortable lesson for everyone else: a 40% drop in search traffic did not dent the growth line, which means the traffic was already worth far less than the industry pretended. The publishers who survive the next two years will be the ones who, like People Inc., stopped building the P&L on the click before the click disappeared.
Time Is Now Selling Ads to Readers That Aren’t Human
Source: Digiday, via The Media Copilot.
Ally Bank and the Project Management Institute just bought some of the first ads designed to be read by a machine rather than a person. Time began serving ads to AI agents this month, formatting them as sponsored FAQs stuffed with brand messaging and dropping them into stripped-down, markdown versions of its pages — the same markdown conversion Time applied to its entire site last month, a format built for crawlers rather than eyes.
The logic is grimly rational. If AI agents are already reading publisher pages on a user’s behalf and returning the answer without the visit, the page view is gone either way — so a publisher may as well monetize what the agent sees on its way through. The buyer pool is tiny and the format is unproven, but this is the first real attempt to build an ad unit for a reader who will never look at it. It is also the clearest admission yet of where things are heading: the audience Time is now willing to sell against is not a person at all, but the software standing between the person and the page.
The New York Times Is Betting Its Subscriptions on Games, Not Reach
Source: A Media Operator, Bron Maher.
Wordle was played 4.2 billion times last year, down from 5.3 billion the year before, and the New York Times is turning it into a half-hour NBC game show produced by Jimmy Fallon’s company to pull lapsed players back. But the show is a sideshow to the real strategy, which is retention. Games sit at the center of the Times’ All Access bundle — news, Games, Cooking, Wirecutter, Audio, The Athletic — and games are the stickiest thing in it, because a years-long Wordle streak is painful to abandon in a way a news subscription never is.
The tactics all point the same way. The Times moved the Mini crossword behind the paywall last summer and added a subscriber-only “Midi” that is pulling Mini players deeper into the habit. Its new two-player Crossplay app has passed 3 million downloads and 30 million games, with what head of games Jonathan Knight calls “world-class” return rates. Family plans — priced at about 1.5 times a normal subscription and counted as two subscribers each — drove the company’s biggest quarterly subscriber jump since 2020.
None of this depends on referral traffic, and that is the point. With 13 million total subscribers, the Times has built the one asset an AI answer engine cannot disintermediate: a daily habit a reader chooses to return to, tied to a bundle that gets more expensive to quit the longer you stay. While most publishers fight to survive the collapse of search referrals, the Times is quietly proving that the durable business was never reach at all — it was routine.
🔍 The Search Squeeze
Google Puts Publishers’ Top Stories Inside Its AI Answers — on Google’s Terms
Source: Newzdash and Press Gazette, via The Media Copilot.
Google has begun embedding the publisher Top Stories carousel directly inside AI Overviews, and new data from Newzdash puts the format in 15.5% of trending-news searches in the US that trigger Top Stories, drawn from more than 17 million appearances across 40-plus countries. For a publisher, that sounds like a lifeline — a way back into the answer box that has been eating referral traffic.
The catch is the whole story. Qualifying for the embedded placement may require accepting Google’s AI-generated summaries of your own reporting, and the embedded and standalone formats never appear together, so opting out of Google’s AI features risks losing the placement entirely with no guarantee of getting the old box back. Newzdash’s John Shehata called it a “double-edged sword,” which is generous. Google is offering publishers visibility in exchange for control, and it is holding both levers at once. The choice on the table is not whether to participate in AI search — it is how much of your own content strategy you are willing to hand over for the privilege.
UK Publishers Just Posted Their First Down Quarter in a Year
Source: AOP/Deloitte Digital Publishers’ Revenue Index and Press Gazette, via The Media Copilot.
Digital revenue across UK publishers fell 4.55% in the first quarter of 2026, ending four straight quarters of growth, according to the Association of Online Publishers and Deloitte. AOP managing director Richard Reeves called it the “first tremors in an earthquake being felt across the industry” as AI-generated answers and shifting habits cut into referral traffic. The survey is small — 13 publishers — and not uniformly grim, with 62% still reporting growth. But a reversal after a full year of gains is the signal worth watching: whatever cushion publishers built against AI search is starting to wear thin for a meaningful slice of the market, and People Inc.’s numbers suggest the ones absorbing it best are the ones who already moved their revenue off the session.
⚖️ Rights & Rules
A Judge Just Gave Publishers a Weapon That Isn’t Copyright
Source: reported by Ars Technica and others, via The Media Copilot.
A federal judge in Manhattan, Paul Engelmayer, refused to dismiss the core of Reddit’s lawsuit against Perplexity and three data-scraping firms, letting Reddit pursue claims that the companies illegally bypassed technical protections to harvest its content. The ruling decides nothing about whether Perplexity broke the law, but it keeps alive a theory that could end up mattering more than copyright: that circumventing a platform’s access controls is itself grounds for liability.
That distinction is the one to watch. Copyright cases force publishers to prove infringement, a slow and uncertain fight the AI labs are well funded to wage. A circumvention claim asks a simpler question — did you climb the wall the site put up? — and for any publisher building paywalls, bot-blocking, and licensing terms around its content, an answer of “yes equals liability” is a far sharper deterrent than a distant copyright verdict. If courts keep this theory alive, the leverage in every unlicensed-scraping standoff shifts a little toward the people who own the wall.
The EU Now Requires AI Content to Say So
Source: the Guardian, via The Media Copilot.
As of August 2, any company operating in the European Union must disclose when images, audio, or text have been artificially generated or manipulated to look real, using both a visible label and a digital watermark — and must tell people outright when they are talking to a chatbot. The rule, part of the EU’s AI Act, hits new systems immediately and gives existing ones a longer runway. The United States has nothing close to it, which means global brands and publishers will likely end up running two content pipelines: one labeled for Europe, one unmarked for everywhere else. The era of quietly shipping synthetic content into a European feed is over; the era of two-track disclosure has begun.
🎙️ From the Pods
Bluesky’s New CEO Wants to Get Paid Only When Publishers Do
Source: Decoder with Nilay Patel (The Verge) — guest Tony Schneider, CEO of Bluesky.
Tony Schneider, the former Automattic chief now running Bluesky permanently after founder Jay Graber moved to a chief innovation role, laid out a business model that reads as a direct rebuke to the platforms that spent a decade extracting from publishers. Bluesky sends about 2 million clicks a day to publishers and is the number-one referrer for many of them, and Schneider’s framing is blunt: “We make money when everybody else makes money.” Rather than run an ads-only model that dictates everyone else’s economics, the company plans to monetize this year by taking a cut of the conversions and traffic it enables — an affiliate logic that only pays Bluesky if the publisher gets value first.
Underneath the app is the open ATProto protocol, now handed to the IETF as an official standard, with roughly 1,000 interoperable apps built on top and 45 million users in the flagship app. Schneider is deliberately divesting control — the test, he says, is whether the network survives if Bluesky the company disappears.
The WordPress angle: Schneider reaches for his old employer as the model, and it is apt. WordPress bootstrapped an open, GPL-licensed ecosystem it does not control, and now WordPress supports StandardSite, the long-form content format emerging on ATProto — meaning a post can live natively across a protocol rather than inside one company’s walls. In a week defined by platforms tightening their grip on publisher content, the most interesting counter-move comes from the one architecture that treats the publisher’s own site, not the platform’s feed, as the thing that should own the audience.
📎 Also Noted
🔸 Newsmax licensed its current and archived reporting to Meta’s AI search and discovery tools across Facebook, Instagram, and WhatsApp in exchange for links and summaries — another outlet taking the check and the promised referrals now rather than holding out. (The Desk, via The Media Copilot)
🔸 A Nieman Lab study found Reuters, the BBC, and the Guardian have each built entirely different AI rules — research assistant, human sign-off, editor-approved tasks — a reminder that no industry-standard AI policy is coming, and the outlets furthest along wrote their own. (Nieman Lab, via The Media Copilot)
🔸 Only 34% of North American consumers view ads inside AI chat platforms positively, per a DoubleVerify study of 22,000 consumers — a workable floor, but a warning to any publisher racing into chat-ad revenue faster than the trust data supports. (DoubleVerify, via The Media Copilot)
🔸 ChatGPT will no longer imitate a named author’s voice on request, redirecting prompts like “write in the voice of Hemingway” toward generic literary traits — a narrow concession to writers, since the guardrail applies to the prompt, not to what the model already learned. (Ars Technica, via The Media Copilot)
🔸 Easyfairs completed its roughly €87 million acquisition of Belgian trade-show operator Xpo Group (€55M revenue, 61 fairs across 17 countries), extending a run of events consolidation as publishers keep treating live events as a hedge against the collapsing page view. (A Media Operator)
🔸 Axios teased a Q2 ad-economics report arguing live events are advertising’s last mass-audience advantage and that creator marketing is becoming corporate infrastructure — both bets on scarce, un-fakeable attention. ⚠️ Paywalled — summary based on the available preview. (Axios Media Trends)
🧭 Takeaways
- The session is no longer the unit of value. People Inc. grew digital revenue and EBITDA in a quarter its sessions fell 22% and Google traffic dropped 40%. Any publisher still budgeting off page views is measuring a business that is quietly ending.
- Retention beats reach. The New York Times is proving that a daily habit inside a bundle — games, streaks, family plans — is the one asset an AI answer engine can’t intercept. Reach can be disintermediated; routine can’t.
- The new customer might be a machine. Time is selling ads to AI agents and converting its whole site to markdown for crawlers. If the agent is the one reading the page, the monetizable audience is the agent — and the CMS decisions that expose content to it are now revenue decisions.
- Google’s AI Overviews offer is a trap dressed as a lifeline. Getting back into the answer box may cost you control over how your reporting is summarized. Read the terms before celebrating the visibility.
- The legal fight is shifting from copyright to access. The Reddit-Perplexity ruling suggests the sharper weapon against unlicensed scraping is a wall and a circumvention claim, not a years-long infringement case. Publishers should be building both.