AI search and zero-click: what changes for publishers, and what still works

Between May 2024 and May 2025, something fundamental broke in the relationship between search engines and publishers. Zero-click searches rose from 56% to 69%, meaning more than two-thirds of Google queries now end without anyone visiting a website. Organic traffic to US news sites fell from 2.3 billion monthly visits to under 1.7 billion, a loss exceeding 600 million visits in less than a year.

The culprit isn't subtle. Google's AI Overviews, which appeared in just 6.49% of searches in January 2025, now dominate over 15% of queries. When these AI summaries appear, click-through rates plummet by 47%, with only 8% of users clicking any link compared to 15% without the AI summary. Just 1% click the citation links within the overview itself.

For publishers, this represents an existential threat disguised as a product feature. The traffic that sustained digital journalism, funded investigative reporting, and supported independent media is being systematically redirected into Google's own ecosystem. And it's not coming back.

The advertising model is collapsing

The mathematics are brutal. Publishers report that 78% of their digital revenue still comes from advertising. Every percentage point of search traffic lost squeezes the budgets that fund actual journalism. When CNN loses 28% of its traffic year-over-year, when Forbes drops 50%, when Business Insider falls 55%, these aren't just statistics. They're newsrooms shrinking, investigations cancelled, and beats abandoned.

The traffic decline hits different publishers differently, but almost everyone is hit. Major publishers experienced losses ranging from 26% to 55% year-over-year. The Washington Post is down 40%. NBC News is down 42%. The Huffington Post is down 42%. Even relative winners like CNN still lost more than a quarter of their search traffic.

Educational publishers face similar devastation. Chegg reported a 49% decline in non-subscriber traffic as AI Overviews began answering homework questions directly. Health information sites warn of 20-35% session losses as medical queries get answered without users ever visiting the source.

The cruel irony: publishers whose content trains these AI systems are the same ones watching their traffic evaporate. AI companies crawl publisher content at ratios estimated between 1,200:1 and 1,700:1, meaning they extract vastly more than they return in referral traffic.

What's actually changing

The shift from search engine to answer engine fundamentally alters the value exchange that sustained digital publishing for two decades.

The old model: Publishers allowed search engines to index their content. Search engines sent traffic back. Publishers monetised that traffic through advertising or converted visitors to subscribers. Everyone benefited.

The new model: AI systems still crawl and index publisher content. They still use it to construct answers. But they no longer send meaningful traffic back. The value flows in one direction.

Similarweb data shows this isn't a temporary adjustment. Zero-click searches have grown steadily, and AI Overviews continue expanding into more query types. What began with informational searches now increasingly covers commercial and transactional queries. The share of navigational searches triggering AI Overviews jumped from 0.74% in January to 10.33% by October, meaning even branded traffic faces AI competition.

Mobile users experience this shift more acutely, with zero-click rates reaching 77% compared to 46.5% on desktop. As mobile continues to dominate web usage, the problem intensifies.

The licensing gold rush

Some publishers are negotiating their way out of the crisis through direct licensing deals with AI companies.

OpenAI has pursued agreements with major publishers. News Corp secured a multi-year deal reportedly worth hundreds of millions. Dotdash Meredith signed a reported £16 million agreement. The Financial Times, The Atlantic, Vox Media, and Associated Press have all struck deals. Meta confirmed multi-year AI licensing agreements with seven major publishers, including CNN, People Inc., and USA Today.

These arrangements typically bundle three rights: access to archives for training AI models, real-time content display with attribution in AI products, and technology access allowing publishers to use AI tools themselves.

The problem: licensing creates tiers. Publishers with vast archives and unique content can negotiate meaningful deals. Smaller publishers lack leverage. The result is a "licensed web" of premium content behind APIs and agreements, and an "open web" of everyone else whose content gets used without compensation beyond minimal referral traffic.

Industry observers note that even publishers accepting deals acknowledge the money doesn't fully replace lost advertising revenue. Roger Lynch of Condé Nast said their OpenAI partnership "begins to make up for some of that revenue" lost from traditional search changes, a phrasing that admits it doesn't fully compensate.

Building direct distribution

Publishers who survive this transition will do so by reducing dependence on search traffic entirely.

Newsletters have become critical infrastructure. Publishers including Axios, Eater, The Guardian, and theSkimm are expanding their email offerings, creating more personality-driven content designed to build direct relationships with readers. Newsletter open rates for these publishers range from 45-65%, far exceeding typical engagement metrics for search-driven traffic.

The Guardian launched multiple US newsletter products specifically to convert one-off visits from platforms into longer-term relationships. Total global subscribers to Guardian US-produced newsletters grew by 680,000 in the past year. Axios is building paid membership tiers from its free newsletters, charging £1,000 annually for premium access.

First-party data becomes the critical advantage. Every newsletter subscription, every site login, every direct visit represents a reader relationship that doesn't depend on algorithmic intermediaries. Publishers are prioritising these authenticated visits because they provide the data needed for personalisation, the foundation for subscriptions, and the targeting capabilities that make advertising valuable.

Diversifying beyond advertising. Reuters found that 77% of publishers now focus on subscriptions and memberships as their biggest revenue priority. Events (48%), affiliate revenue (29%), donations (19%), and related businesses (15%) provide additional streams that don't depend on search traffic.

The New York Times' Games app exemplifies this diversification. By creating products people want directly, rather than relying on search to deliver readers to articles, publishers build sustainable businesses that don't depend on Google's goodwill.

What still works

Not everything has collapsed. Some strategies retain value even as the landscape shifts.

Branded search remains relatively stable. Research shows that branded queries with AI Overviews actually see an 18% increase in click-through rate. When people search for your publication by name, they're more likely to click through even when an AI summary appears. This creates a performance gap: generic terms see 34-46% CTR decreases, while branded terms increase 18%.

The implication: brand-building matters more than ever. Publishers with strong brands have a defensive moat against AI disruption that smaller players lack.

Proprietary journalism provides protection. AI systems excel at summarising consensus information but struggle with original reporting, investigative work, and local news. Industry analysts recommend prioritising opinion pieces, analysis, and stories that can't be found elsewhere. This content gives audiences information they can't get from a bot.

Community and engagement trump traffic volume. A smaller audience that actively engages, subscribes, and participates in community features provides more value than large volumes of anonymous search traffic. Publishers are shifting metrics from pageviews to measures of loyalty, conversion rates, and lifetime value.

AI referral traffic is growing, albeit slowly. Similarweb data shows AI referral traffic to news and media sites grew from 35.3 million global visits in May 2025 to 35.9 million in June. ChatGPT drives the bulk of this traffic (78.5% for Yahoo, 95.4% for Reuters, 99.4% for the Guardian). While these numbers don't offset search losses, they represent a growing channel that publishers need to optimise for.

The metrics that matter now

Pageviews no longer tell the story. Publishers need new frameworks for measuring success.

Citation frequency in AI systems. Even if citations don't drive clicks, they indicate brand visibility and authority. Publishers are beginning to track how often their content appears in ChatGPT, Perplexity, and Google AI Overviews. This provides insight into brand mentions even when referrals don't materialise.

Conversion rates over traffic volume. A thousand engaged visitors who might subscribe matter more than ten thousand anonymous search visitors who bounce immediately. Publishers are optimising for quality over quantity.

Direct traffic and newsletter growth. These metrics indicate audience relationships that don't depend on algorithmic intermediaries. Growth in direct channels suggests a sustainable business model.

Subscriber lifetime value. For publishers pursuing subscription models, the revenue generated by each subscriber over their entire relationship matters more than any single traffic metric.

Engagement depth. Time on site, pages per session, return visit frequency, and comment participation indicate genuine audience connection rather than drive-by traffic.

A practical plan

For publishers navigating this transition, several concrete steps provide a path forward:

Audit your traffic sources. Understand how much of your audience comes from search, how that's changed over the past year, and which content types are most affected. This baseline reveals your vulnerability.

Invest in newsletter infrastructure. Build email products that provide value beyond what's available on your website. Personality-driven content, exclusive analysis, and curated recommendations justify subscription.

Develop first-party data capabilities. Implement registration walls, loyalty programmes, and identity frameworks that let you know your audience. This data enables personalisation and provides value to advertisers even as traffic declines.

Diversify revenue streams. Don't depend entirely on advertising. Explore subscriptions, memberships, events, affiliate partnerships, and related businesses that align with your editorial mission.

Protect high-value content. Consider placing investigative work, proprietary analysis, and unique reporting behind registration or paywalls. This content justifies direct relationships and can't be easily replicated by AI.

Optimise for AI citations. While citations don't guarantee traffic, they provide brand visibility. Ensure your content is structured, authoritative, and citable. Use schema markup, clear sourcing, and expert attribution.

Build community features. Comments, forums, newsletters, and events create reasons for audiences to engage directly rather than arriving via search and immediately leaving.

Track the right metrics. Move beyond pageviews to measure engagement, loyalty, conversion, and lifetime value. These indicators reveal business health better than traffic volume.

What comes next

The transition from search-driven to direct-audience models will take years and won't be smooth. Many publishers won't survive it. Those that do will look fundamentally different from their predecessors.

The International News Media Association makes the stark prediction that "organic search traffic is expected to shrink to potentially near zero over the next two to three years." While this may prove overly pessimistic, the direction is clear.

Publishers face a choice: adapt to a world where search traffic continues declining, or watch their businesses slowly suffocate as the oxygen of referral traffic gets cut off.

The good news, such as it is: publishers who move quickly to build direct audience relationships, diversify revenue, and reduce search dependence can emerge stronger. The bad news: there's no going back to the old model. The era of abundant, free search traffic is ending. What replaces it will require more work, more investment, and more direct value creation.

The publishers who thrive will be those who stop optimising for Google's algorithm and start optimising for their actual readers. That's always been the right strategy. Now it's the only one.

The writer covers technology and media industry trends.