Might it be that Google’s AI-enhanced search, all overhyped changes notwithstanding, hasn’t cut down the overall number of clicks to websites, after all? That’s the contention of Google’s Head of Search, Liz Reid, who recently said, “total organic click volume has remained relatively stable year-over-year.” But this statement comes without a shred of backing evidence, forcing search pros and publishers to weigh the company’s cheerleading against a mounting corpus of outside research.
The struggle between Google’s story and outside analysis is one that is felt. Pew Research Center’s recent survey, for example, revealed that users who were presented with an AI Overview clicked on links to other sites only 8 percent of the time, about half the rate of users without an AI Overview, who clicked 15 percent of the time. The research went on to observe that users were more inclined to close the browsing session following an encounter with an AI Overview, as 26 percent of sessions were closed on those pages, a figure higher than 16 percent for regular results. In spite of Google’s protestations about the methodology of the study, the findings reflect the fears of numerous publishers who have seen their traffic drop off since the launch of AI-powered abstracts at the top of search listings based on several industry analyses.
The center of controversy is the technological development of Google’s search engine. AI Summaries use large language models (LLMs) to generate answers by accessing a starburst of documents sometimes even creating surrogate queries to broaden the information base. The operation, detailed in newly filed patents, involves fetching documents not only for the initial query but for related, recent, and implied queries as well. This method constructs a “custom corpus” of strongly relevant documents from which the LLM derives an answer summary. Citations are extracted from the corpus, but importantly, AI-constructed summaries tend to fulfill the user’s intent outright, minimizing the requirement for additional clicks as explained in technical explanations.
Google’s response to the click-through controversy has been to shift the focus from raw click numbers to what it calls “click quality.” Reid asserts, “average click quality has increased, and we’re actually sending slightly more quality clicks to websites than a year ago.” In Google’s view, a quality click is one where users don’t quickly return to the search results a sign, they argue, of more satisfied and engaged visitors. Still, with no public metrics available, the volume and pattern of these so-called quality clicks are unknown.
One of the most significant changes in user behavior, Google says, is the increasing call for “authentic voices and first-hand perspectives.” The pattern can be seen in Google’s 2024 alliance with Reddit, which caused a spectacular jump in Reddit’s visibility in search results. Reddit’s active users increased by 21 percent over the last year to more than 110 million. Consequently, increasing clicks are being redirected to big players like Reddit, and smaller, niche publishers are experiencing a significant drop in organic traffic share. This is nothing new; small review sites in past years had their traffic cut into by big brands and SEO-optimized aggregators. Now, AI Summaries seem to be speeding this lopsided distribution along, preferencing sites that resonate with Google’s shifting sense of “authenticity” and user interest.
The dynamics of click-through rates (CTR) in this new world are complicated. Traditionally, the #1 organic result in Google had an average CTR of 27.6 percent, with a precipitous decline for lower-ranked results. The introduction of AI Overviews breaks this curve because the user is more satisfied with the summary created by AI and less likely to click any link whatsoever. For news searches, the proportion of zero-click queries has increased from 56 percent to 69 percent since May 2024 when AI Overviews were introduced. For publishers, this would mean that even a first-ranking citation within an AI Overview might no longer be fruitful in terms of traffic as recent research indicates.
Google insists that AI Overviews are intended to “draw attention to the web, not diminish the need to click.” Admittedly, though, Reid allows that for certain questions, users might “get what they need from the AI answer and will not click further.” This double nature AI as both door and doorway raises deep questions about the web’s future visibility. Though Google boasts billions of daily clicks in total, the redistribution of those clicks is far from equal. Top sites such as Reddit and big publishers might gain, but most smaller sites experience plateauing or declining traffic even while their search impressions increase.
For digital marketers and search professionals, the message is plain: the measurements and dynamics of natural traffic are being rewritten by AI. Those traditional SEO tactics Keyword optimization, rankings, even CTR now have to compete with a probabilistic, LLM-based system that rewards passage-level coherence, semantic salience, and user engagement signals above plain old rank position. As AI Overviews proliferate, the game will not only be to show up in search, but to get cited, believed, and most importantly clicked. Until Google makes available clear, site-level information, the argument about who benefits and who loses from the AI era of search is bound to escalate.
