Star journalists are flocking to Substack for editorial freedom and income, but readers face growing subscription overload – raising questions about how long the newsletter boom can last.
Referrals from ChatGPT are surging, but they’re no silver bullet for news publishers watching their search traffic collapse. In the past year, generative AI answers have led nearly 69% of news-related web searches to end without any click to an article, up from 56% before Google rolled out AI “Overview” snippets. Not surprisingly, organic search traffic to news sites has plummeted, dropping from about 2.3 billion monthly visits at its mid-2024 peak to under 1.7 billion as of this spring. Readers are increasingly getting headlines and answers directly from AI summaries or chatbots, cutting publishers out of the equation.
At the same time, news consumption via ChatGPT is skyrocketing. From January 2024 to May 2025, news-related prompts to ChatGPT jumped by 212%. And crucially, ChatGPT referral clicks to news sites have 25× increased: from under 1 million in early 2024 to over 25 million in 2025. It’s a dramatic rise, indicating that some users do ask ChatGPT for news and occasionally click through. But 25 million is a drop in the bucket compared to the hundreds of millions of clicks publishers have lost from traditional search engines. As one analysis notes, this AI-driven traffic bump is “hardly enough to make up for publishers’ losses” amid an overall 27% year-over-year decline in news search visits.
The AI Impact on Search Traffic
From an SEO perspective, the game is changing. Google rank and headlines alone no longer guarantee traffic if users get their answers from an AI box on the results page. Similarweb’s data underscores that even top-ranked news stories often see no click-through now. Publishers have spent years optimizing for Google, only to find the traffic payoff diminishing as AI summaries siphon off casual information-seekers. One stark example: after Google introduced its AI “Overview” feature in Search, zero-click news queries became the norm for over two-thirds of users.
Meanwhile, OpenAI’s ChatGPT has quietly become a notable news aggregator in its own right, though one with a fraction of Google’s reach. ChatGPT’s referrals to news sites spiked by 25× year-over-year. Certain publishers are benefiting more than others: Reuters, the New York Post, and Business Insider each saw ~7–9% lifts in traffic from ChatGPT, according to Similarweb. By contrast, The New York Times, which has taken a more adversarial stance, even suing OpenAI for allegedly scraping its content without permission, saw only a 3% uptick and remains wary. This suggests that outlets more open to AI might reap slightly better referral rewards, for instance, Reuters +8.9% from ChatGPT, whereas those limiting AI’s use see little benefit. Still, even for the “winners,” these gains are modest in the bigger picture of lost search traffic.
Another noteworthy trend: which topics users consult AI for news. Thus far, ChatGPT news prompts have skewed toward quick-update beats like stocks, finance, and sports, areas where timely data or scores likely drive people to ask AI. However, Similarweb observed growth in prompts about politics, the economy, weather, and other issues. This could hint at a shift from using ChatGPT just for instant facts toward more in-depth, issue-driven queries that might previously have led readers to analysis pieces. If users begin trusting AI for deeper explanations, the risk to publishers isn’t just losing traffic on breaking news, but on explainer and feature content as well.
Publishers Scramble for Solutions and Silver Linings
For news publishers, these patterns are alarming – but not hopeless. The bright spot is that new referral channels are emerging: tens of millions of visits from ChatGPT, while small relative to Google, are something publishers couldn’t tap into a year ago. The surge suggests an audience segment actively using chatbots to find or consume news. This could encourage publishers to ensure their content is accessible to AI services. In other words, those who adapt to the AI landscape, by partnering with aggregator apps, optimizing content for voice and chat, or creating their own AI assistants, may capture some of this emerging traffic stream.
That said, no amount of ChatGPT traffic will single-handedly fill the void left by the collapse in search referrals. Publishers are confronting a future where chasing clicks via SEO is less viable. Many have already been forced to restructure operations and cut costs. Insider, for example, recently laid off 21% of its staff, citing the double hit of “shrinking search traffic” and users relying on AI summaries instead of clicking articles. Its CEO bluntly noted the company must withstand “extreme traffic drops outside our control” by reshaping the business model. Across the industry, similar pressures are leading to drastic measures, from paywall experiments to outright shutdowns, as the old advertising-driven model falters.
Big tech companies are belatedly acknowledging the problem they helped create. Google, under pressure from publishers, launched a tool called Offerwall in June to help news sites monetize without relying purely on traffic. Integrated into Google Ad Manager, Offerwall lets a visitor choose alternative ways to access content, for example, pay a small fee per article, watch a video ad, take a survey, or even sign up for a newsletter, instead of hitting a dead-end paywall. The idea is to give publishers a chance to convert some of those fly-by readers who arrive from search or social. It’s an intriguing concept and one publishers have experimented with before in various forms, but its effectiveness is uncertain. Micropayments for news have historically struggled to gain traction, and readers often balk at any extra friction online. Still, with Google’s backing and AI helping to target when to show the Offerwall, some outlets hope this could recover a bit of revenue from the vast number of one-and-done visitors who would otherwise bounce away.
Meanwhile, AI itself is a double-edged sword for publishers. Many newsrooms are adopting generative AI tools to cut costs or speed up production, even as those same technologies threaten their traffic base. In a recent interview, OpenAI CEO Sam Altman acknowledged the disruption to jobs and industries, noting that even changes ultimately “good for society and the economy” will come with “real pain... in that moment” for those affected. We’re seeing that pain now in journalism. But we’re also seeing publishers attempt to innovate out of the crisis: doubling down on subscription programs, events, and direct reader relationships to reduce dependence on platforms, and lobbying for regulatory or legal remedies such as copyright lawsuits against AI firms scraping content.
Our take
The decline in search traffic is a watershed moment for digital media. It forces the question of how journalism survives when the referral pipelines it relied on dry up. In the short term, no single fix will replace that lost volume. But publishers can seize this chance to rethink their strategies. Building loyal audiences via newsletters, podcasts, and communities is more critical than ever if drive-by traffic is evaporating. Embracing new platforms and formats is a must, whether that’s ensuring content surfaces in AI assistants, or leveraging short-form video and other channels where younger audiences now get their news. Crucially, news organizations and tech companies will need a new alignment of incentives: if AI and search platforms increasingly answer users’ questions without clicks, they may have to directly compensate or integrate publishers in new ways. Otherwise, the “free traffic” era is ending, and with it the sustainability of many ad-supported newsrooms.
Change is never easy, as newsroom layoffs and closures attest, but those outlets that adapt, diversify revenue, and offer unique value will have the best shot at weathering this transition. The rise of ChatGPT referrals, while not a cure-all, is at least a sign that quality original reporting still has a place in the AI-centric ecosystem if publishers can figure out how to capture value from it. The media industry’s next challenge is turning these fewer but more intentional clicks into a viable business model in the age of AI-driven information.
What We're Reading
📉 Google’s AI-generated summaries are causing a surge in “zero-click” searches, leaving publishers with less traffic even when they rank highly, reports The Register. As users get answers directly from AI, traditional SEO strategies are becoming less effective. It’s forcing a major rethink across digital media.
💰 Google has launched a tool called Offerwall that lets readers pay per article, watch an ad, or take a survey to access paywalled content, writes TechCrunch. It’s designed to help publishers recover revenue as search traffic declines. Whether readers will actually engage with it remains to be seen.
🚫 Publishers are stepping up efforts to block AI tools from scraping their content, reports The Wall Street Journal. Some are deploying technical blocks while others pursue legal action, hoping to preserve both traffic and copyright. The battle could define the next era of journalism online.
📝 Substack co-founder Hamish McKenzie has launched a new media colum, on rival platform Beehiiv, not Substack, reports Feed Me. The column, titled Hamish’s Hot Sauce, is part of the Beehiiv-based newsletter Breaker and will run biweekly. The move has raised eyebrows given McKenzie’s role in building Substack’s writer ecosystem.
