Zero-Click and the Vanishing Referral

Zero-click is not new. The speed of the last two years is.

Search results pages have been answering questions directly, without requiring a click, since long before generative AI arrived: featured snippets, knowledge panels, and local packs have quietly absorbed a share of search intent for the better part of a decade. What is new is the slope of the line. SparkToro's clickstream panel research, analyzed by Digital Applied, measured the US zero-click rate at 60.45 percent in 2024 and 68.01 percent across January through April 2026, a jump of nearly eight points in roughly two years. Split by device, mobile zero-click sits near 77 percent against about 50 percent on desktop, a gap that matters because mobile is where most local, near-me, and recommendation-style searches happen.

This figure deserves a caveat rather than a headline treatment. It excludes the Google Search app entirely, which likely means the true zero-click rate is understated rather than overstated. And the underlying panel vendor has changed twice in recent years, first from Jumpshot to Datos and then to Similarweb, which weakens any claim of a single, clean long-run trend line running through the data. That is exactly why this study tiers the SparkToro figures as emerging rather than established: the direction is real and corroborated elsewhere, the precise percentage is one vendor's proprietary measurement.

Only 276 of every 1,000 US Google searches now end in a click to the open web, down from 374 two years earlier.

When an AI Overview appears, the click drops hardest

The clearest signal in this study is not the aggregate zero-click rate. It is what happens specifically when a generative answer appears on the page. SparkToro's panel data found searches that trigger a Google AI Overview running near an 83 percent zero-click rate, against roughly 60 percent for a standard query without one. Ahrefs reached a directionally similar conclusion through an entirely different method: analyzing 300,000 keywords in Google Search Console data, split evenly between queries with and without an AI Overview, and comparing March 2024 to March 2025, Ahrefs found the presence of an AI Overview correlated with a 34.5 percent lower average click-through rate for the number-one organic ranking position.

The most rigorous evidence on this specific question comes from Pew Research Center, which did not analyze aggregate search logs but tracked the actual browsing behavior of 900 US adults across 68,879 real Google searches in March 2025, including 12,593 that returned an AI summary. People clicked a traditional search result in just 8 percent of visits when an AI summary appeared, compared with 15 percent when it did not, very nearly halving the click rate. More striking still: only 1 percent of visits to an AI-summary page produced a click on one of the sources the summary itself cited, meaning even the sites AI Overviews credit as their source are not reliably receiving the visit. And people were more likely to simply stop searching altogether after seeing a summary, ending their session entirely 26 percent of the time versus 16 percent for a standard results page.

Three independent methods, three different organizations, three different data sources, one consistent direction. That convergence is what makes this the most confidently established finding in the whole study, even while the exact percentage a business owner should expect to lose remains a live, disputed question.

Publishers are already living the downstream effect

If the click is disappearing at the search results page, the next place to look for the effect is the traffic logs of the sites that depend on that click. Reuters Institute for the Study of Journalism, working with underlying Chartbeat traffic data across more than 2,500 tracked publisher websites, found Google search referral traffic down 33 percent globally and 38 percent in the US, year over year, to November 2025. Google Discover referrals, a separate traffic source, fell 21 percent globally over the same window. The same report surveyed 280 media leaders across 51 countries and found they expect a further 43 percent average decline over the next three years, which reads less like a one-time correction and more like an organization bracing for a permanent structural shift.

Smaller, more granular data points confirm the same direction with useful texture on who gets hit hardest. Digiday, citing a Digital Content Next trade-body survey of 19 member publishers, found a median year-over-year Google Search referral decline of 10 percent over an eight-week window in May and June 2025, with results ranging from roughly minus 1 percent to minus 25 percent across the group, and non-news brands (down 14 percent) losing more than news brands (down 7 percent). A UK trade group, the PPA, separately reported member click-through-rate declines of 10 to 25 percent even where rankings had stayed stable, and cited one lifestyle publisher whose click-through rate fell from 5.1 percent to 0.6 percent with an unchanged page-one ranking. Ranking well is no longer the same thing as being visited.

The dissent: not everyone agrees the sky is falling

A study built to inform real decisions has to include the evidence that complicates its own headline, and this one does. Graphite, working in partnership with Similarweb, analyzed more than 40,000 of the largest US websites and found overall US organic search traffic down just 2.5 percent year over year, comparing February through December 2024 against January through November 2025. The researchers explicitly framed this as a rebuttal to the widely circulated 25-to-60 percent decline figures, arguing those numbers are artifacts of small, unrepresentative samples or of recalled-behavior survey methodology, neither of which directly measures actual traffic the way a large panel does.

This study tiers that finding as contested, not because it is less credible than the larger-decline figures, but because it directly disagrees with them using a comparably large sample and a defensible method. Both the Reuters Institute/Chartbeat publisher data and the Graphite/Similarweb panel data can be accurate descriptions of different populations. Large, established media publishers with heavy reliance on informational, top-of-funnel search queries appear to be losing referral traffic fast. A broader cross-section of the web's largest 40,000 sites, weighted toward sites with more transactional or brand-specific query mixes, is losing far less. Where a specific local or independent business sits on that spectrum is not something either dataset was built to answer.

A 2.5 percent panel finding and a 38 percent publisher finding can both be true. They are measuring different populations, not different realities.

Attention is migrating past the click, not just away from it

The clicks that used to land on a website are not simply vanishing into the void. Reuters Institute's 2026 Digital News Report, surveying roughly 2,000 respondents in each of 48 markets, found that for the first time at the global level, social media and video networks (54 percent) are more widely used than publishers' own websites and apps (51 percent) as a way of accessing online news. That is a genuine crossover, not a marginal shift, and it means the audience is not merely clicking less; it is choosing to discover in a different place entirely.

A second, smaller but faster-moving migration is visible in the same report: 16 percent of adults under 35 report using AI chatbots for news, against a 10 percent all-ages global average. That gap between a younger cohort and the overall population is usually the leading edge of where broader adoption is headed next, not a plateau.

For a local or independent business, the more directly relevant version of this same migration shows up in BrightLocal's 2026 consumer survey of 1,002 US adults. Use of ChatGPT and similar AI platforms for local business recommendations rose from 6 percent to 45 percent of respondents in a single year, making AI tools the third most-used recommendation source behind only Google and Facebook. Google's own share of recommendation usage, over that same period, fell from 83 percent to 71 percent. A business that is invisible to those tools is not losing a click. It is losing the recommendation itself, before a search ever happens.

Does the traffic that still arrives convert better? Nobody independent has checked

The one claim in this space that should be treated with the most skepticism is also the one most often used to reassure worried business owners: that the traffic AI tools do send converts at a premium high enough to offset the volume lost. Similarweb reported roughly a 40 percent conversion premium for AI-referred visitors over traditional organic-search visitors, 7 percent against 5 percent. Semrush reported AI visitors converting several times more often. Ahrefs found AI search driving a disproportionate share of signups relative to its still-small share of total traffic.

Every one of those figures comes from a vendor that sells AI-search-optimization or analytics services, and this study tiers all of them as contested for exactly that reason. The reported premium ranges from roughly 1.4 times to more than 20 times depending on whose method you use, which is too wide a spread to be describing the same underlying phenomenon with any precision. It may well be true that a visitor who arrives via a specific, high-intent AI recommendation converts unusually well; a small, self-selected group of highly motivated visitors converting better than a large, mixed pool of casual searchers would not be surprising on its own terms. But no independent party outside the vendors selling into this trend has replicated the finding, and until one does, the reasonable position is that this is a plausible, commercially convenient claim rather than a settled fact you should plan around.

The market is already repricing the value of the referral

When the economics of a relationship change enough, the parties inside it eventually renegotiate it, and there is early evidence that is starting to happen at scale. Reddit's content-licensing deal with Google, reportedly worth around 60 million dollars a year, signed in 2024 to supply content for AI model training, is up for renewal. Wall Street Journal reporting, aggregated by ppc.land, indicates Reddit executives are weighing non-renewal as the declining referral value from Google's AI-mediated search changes the underlying economics of the arrangement.

This is a single, market-level data point, not a trend line, and it is included here for what it signals rather than what it proves: one of the largest content platforms on the internet is treating the shrinking value of the click as a live commercial fact worth walking away from a deal over. A local business does not have Reddit's bargaining power, but it is operating in the same underlying market, where the value that used to accrue automatically from being findable is now something that has to be actively won on more surfaces than search alone.

What this means for where you look next

Every figure in this study, without exception, was measured at the level of a large publisher panel, a national consumer survey, or an aggregate keyword database. None of it was measured at the level of a single local business's phone, calendar, or cash register, and that gap is not an oversight in the research; it is a genuine, acknowledged limit of what commercial clickstream and survey panels are built to see. If you are reading these numbers as a business owner, you are right to ask what any of it means for your specific market, your specific category, your specific competitors. No existing study can tell you.

That is the precise reason the Visibility Corpus exists as a different kind of measurement: not another aggregate percentage borrowed from a media industry panel, but a direct, dated read of where a specific market's attention actually sits today, across the search results page, the AI answer, social and video discovery, and the recommendation a customer asks an AI tool for directly. Search Surface Optimization is the discipline of winning attention across that full surface rather than chasing the click alone, and a Machine-Readiness Score is the measured number that tells you where you stand on that surface right now, not where a vendor's aggregate panel says the industry stands on average.