The AI Paradox: Fewer Human Clicks, More Machine Reading

What Happened?

Over the past two years, the rise of generative AI in search engines has drastically altered user behavior. According to a Pew Research Center study cited by VentureBeat, when Google displays an AI summary, users click on a traditional result only 8% of the time, compared to 15% when no summary is present. Links within the AI responses themselves receive barely 1% of clicks. This phenomenon has had a devastating impact on publishers. Chartbeat data reported by Axios indicates that visits from Google Search fell 34% between December 2024 and December 2025 across its publisher network, and small publishers have lost approximately 60% of their search referral traffic in two years. Business Insider lost 55% of its organic search traffic in three years, and some small publishers have shut down. Referrals from chatbots, though growing more than 200% annually, still account for less than 1% of visits to publisher pages.

This decline is not an isolated phenomenon. The trend began with the introduction of Google's AI Overviews in May 2024, which significantly reduced clicks on organic links. A previous study by Raptive (2024) had already shown a 20-30% drop in search traffic for affected sites. The difference now is scale: generative AI has been integrated into multiple platforms, from ChatGPT to Perplexity and Bing, further fragmenting traffic sources.

Why It Matters

The paradox is that while humans click less, machines read more than ever. The 2026 Generative AI Landscape report from Similarweb reveals that the proportion of ChatGPT responses containing live web citations grew more than fivefold in less than a year, reaching 6.8% of all responses in May 2026. In categories like travel, it reaches 22.6%. Each of these links represents an opportunity for visibility, but not necessarily for human traffic.

As Lily Ray, VP of SEO and AI Search at Amsive, points out, the quality of AI responses directly depends on the health of the underlying content layer. If a site's organic visibility drops, so does its visibility in AI responses, because models are less likely to find that content. This creates a vicious cycle: AI responses are built on an information supply chain whose funding model (ad clicks) is collapsing precisely because of those responses. This cycle is unsustainable: if publishers cannot monetize, they will stop producing quality content, degrading AI responses and ultimately the user experience.

Historically, the open web has relied on click-based advertising. The transition to a model where machines consume content without generating direct revenue echoes the impact of social media on traditional media, but it is faster and deeper. In 2023, before the generative AI boom, search traffic accounted for roughly 40% of total publisher traffic. With the current decline, many are seeing reductions of up to 60% in two years, threatening their viability.

Consequences and the New Model

Following the ChatGPT search update on May 7, 2025, which introduced visible brand links within responses, referral traffic from ChatGPT surged 157% in one week. But the nature of that traffic changed: the proportion of referrals landing on the homepage doubled, from 25% to 60%. Traditionally, search sent users to specific articles; now AI researches and compares, and the human arrives ready to act. This means publishers must optimize their homepages for conversion, not just for capturing clicks.

Similarweb also found that brands recommended by AI receive two to four times more subsequent visits than non-recommended ones. Sponsored results appeared in 26% of ChatGPT desktop conversations in June 2026, up from 14% the previous month. This suggests that direct monetization within AI responses is growing, offering a new revenue source for platforms, but not necessarily for publishers.

This shift means publishers and brands must rethink their strategy: it is no longer enough to optimize for clicks; they must ensure content is selected by AI models and that the homepage is an effective conversion point for pre-informed visitors. Machine traffic (AI crawlers) has become as valuable an asset as human traffic, but requires different metrics and monetization models. Some publishers are experimenting with content licensing to AI platforms, such as News Corp's deal with OpenAI, or creating their own AI assistants. However, these models are still nascent and do not offset the decline in advertising revenue.

Compared to the transition from print to digital, the current shift is more abrupt. In the 2000s, newspapers had a decade to adapt; now, publishers see their traffic collapse in two years. The open question is whether the open web will find an alternative business model before the vicious cycle erodes it irreversibly. As VentureBeat aptly summarizes: "If your business depends on people clicking on web pages, the last two years have been brutal." Survival will depend on the ability to innovate in monetization and diversify traffic sources beyond traditional search.

"If your business depends on people clicking on web pages, the last two years have been brutal," VentureBeat summarizes. The open question is whether the open web will find an alternative business model before the vicious cycle erodes it irreversibly.