- Tension: Google now sends 32 out of every 100 searches to the open web, down from 45 a decade ago, and the gap keeps widening
- Noise: This has been filed as an AI story, obscuring that the pattern of Google retaining more search traffic with each product cycle predates AI Overviews by a decade.
- The Direct Message: Marketing built its search assumptions when Google sent 45 clicks. It still operates on those assumptions. The channel is now sending 32.
To learn more about our editorial approach, explore The Direct Message methodology.
In the first four months of 2026, 68% of Google searches in the United States ended without a click. That figure, from a SparkToro analysis of Similarweb panel data covering hundreds of millions of searches, means that less than one in three Google searches now sends a user to any external website. The number that corresponds to “external website receives a visitor” is 32 out of every 100 searches. A decade ago, it was roughly 45.
Thirteen clicks per hundred sounds like an abstraction. At the scale Google operates — approximately 8.5 to 14 billion searches per day globally — it is an abstraction that translates to billions of visits that the open web used to receive and no longer does. The marketing infrastructure built around organic search — the SEO strategies, the content investments, the channel attribution models, the benchmarks against which performance is measured — was constructed in a world where 45 out of 100 searches produced a click. That world no longer exists, and the transition happened slowly enough that most of the infrastructure built for it is still running unchanged.
A decade of incremental retention
The story that AI Overviews broke Google search for publishers is a compressed version of a longer and more deliberate story. The trajectory of zero-click search runs from approximately 45% in 2016 to 49% in 2019, to 60.45% in 2024, to 68% in the first four months of 2026. AI Overviews, which began rolling out at scale in 2024, accelerated the final phase of that curve. They did not initiate it.
What initiated it was a series of product decisions Google made over a decade, each individually defensible and each contributing to the same aggregate effect: more of the user’s query resolved on the search results page, fewer users needing to go anywhere else. Featured snippets, introduced in 2014, put the answer directly in the results. The Knowledge Graph, expanded continuously from 2012 onward, answered factual questions without a click. Rich results for weather, sports scores, flight information, and stock prices gave users the data they were looking for without ever loading an external page. Local results with hours, phone numbers, and maps replaced the visit to the business’s website. Each feature was presented, accurately, as an improvement to user experience. The cumulative effect on the publishers whose content fed those features was a decade-long decline in the traffic that content was supposed to generate.
The methodology note matters here: SparkToro’s historical data stitches together three different measurement panels — Jumpshot for 2016 and 2019 (now defunct), Datos/Semrush for 2024, and Similarweb for 2026. SparkToro flags this itself; the direction of the trend is consistent across all three panels, but the measurements are not perfectly apples-to-apples. The magnitude of the decline is best understood as directional rather than precise. The direction is not in dispute.
What AI Overviews added
AI Overviews accelerated the trend with unusual speed. Zero-click searches grew by 7.5 percentage points in the two years from 2024 to 2026, the fastest acceleration in the decade of data available. The mechanism is direct: when an AI Overview appears, it answers the query at the top of the page with a synthesized response that draws on multiple sources. The user’s question is resolved. The sources are credited. The visit to those sources does not occur. According to available data, AI Overviews now appear on more than 20% of Google searches, and when they do, click-through rates for results below them fall by nearly 60%.
AI Overviews are, in this sense, the most efficient expression of the product philosophy that has governed Google’s search development for a decade: resolve the query in the results, keep the user on Google. They are more capable than featured snippets and more comprehensive than Knowledge Graph entries, but they are the same product, doing the same thing, at greater scale and with greater effectiveness. The novelty is in the capability, not the strategy.
The dashboard problem
The marketing industry’s failure to fully absorb what this means for channel strategy is partly a measurement problem. Most standard analytics implementations track sessions, users, and conversions from organic search. They measure what Google sends, not what Google retains. A marketing dashboard that shows stable organic search traffic over the past several years may be accurately reflecting what Google has sent — which in absolute terms may have declined, or remained flat against rising search volume, or grown modestly — while completely missing the denominator: how many searches in the company’s relevant categories Google handled, and what percentage of those produced a visit.
The question that most search measurement frameworks are not designed to answer is: what share of people who searched for something in our category came to us, versus resolved their query on Google’s page? That question requires visibility into search volume and query behavior that most companies don’t have, or don’t routinely examine. Without it, a channel that is delivering fewer visits per search can still appear healthy in absolute traffic terms if overall search volume is growing. The channel looks stable. The underlying share is declining.
This measurement gap has real consequences for how budgets are set. Organic search is allocated resources — content teams, technical SEO infrastructure, link-building programs — based on its perceived contribution to marketing outcomes. If that perceived contribution is based on absolute traffic numbers measured against a baseline established when the channel worked differently, the resource allocation is being made on outdated premises. The content that was worth producing at 45 clicks per hundred may not be worth producing at 32 clicks per hundred on the same economics.
What 13 missing clicks actually represents
The specific categories most affected by zero-click growth are instructive. Informational queries — the kind that ask “what is,” “how to,” “who is,” or “what causes” — are most likely to be resolved by AI Overviews or knowledge panels without a click. These are also, not coincidentally, the categories that content marketing has most heavily targeted, on the premise that informational content attracts organic search traffic that can be converted downstream. If the informational query is increasingly resolved on the results page, the top-of-funnel content strategy built around it is reaching fewer people than the traffic numbers suggest.
Navigational queries — searches for a specific brand or website — are less affected; users searching for a company by name are usually going to that company’s site. Transactional queries occupy a middle ground: product searches with commercial intent still generate significant click volume, but they face increasing competition from Google Shopping units, local inventory ads, and AI-generated product summaries that can complete a significant portion of the consideration process without a site visit.
The marketing response to zero-click growth has been predictable: optimize for featured snippets and AI Overviews, on the theory that appearing in the answer generates brand visibility even without a click. This is not irrational. Visibility in AI-generated answers is a real form of influence, even if it is currently unmeasurable in most analytics systems. But optimizing for zero-click visibility is a different activity, with different costs and different returns, from optimizing for organic traffic. It is, in a meaningful sense, optimizing for a different channel — one that didn’t exist when the current marketing infrastructure was designed.
The structural question
The 13-click decline per hundred searches is not a problem to be solved by better SEO. It is a description of a platform that has, over a decade, systematically redesigned its product to retain more of the value that users come to it to find. Google’s business model — which depends on advertising revenue generated by users who are on Google’s pages, not on publishers’ pages — aligns perfectly with maximizing the proportion of queries resolved without a click. The features that drove the zero-click rate from 45% to 68% over a decade were not accidents. They were product improvements, from Google’s perspective, because Google’s users experience them as improvements. The fact that they represent a structural shift in the channel for everyone relying on it is a consequence, not an objective.
Understanding this is more useful than resisting it. The channel that sent 45 clicks per hundred in 2016 is not coming back, and the features responsible for the current 32 are continuing to develop. The trend line runs in one direction, and it has run in that direction consistently for a decade. Marketing strategies built on what Google used to be are increasingly expensive ways to reach fewer people. The honest accounting starts with the number: 32 clicks per hundred, down from 45. Everything else follows from deciding to take that number seriously.
