68% of Google searches now end without a click. What the 2026 data shows, why AI answers without sending traffic, and how to become the cited source.
By , founder and lead strategist at Flowup
Direct answer
The “end of the click” is the consolidation of a behavior: most searches now end without a visit to any website. In the U.S., about 68% of Google searches ended without a click in early 2026 (SparkToro/Similarweb), and when an AI summary appears, the click-through rate on traditional results drops by nearly half (Pew Research Center). This does not mean the end of search or of organic: it means visibility and traffic are no longer the same thing. The new contest is not for the link’s position; it is for being the source the answer cites. Brands that AI systems don’t understand and trust don’t just lose clicks: they leave the conversation where the buying decision begins.
A zero-click search is one that ends without the user visiting any website: the answer is consumed in the interface itself, at the top of Google, inside an AI Overview, in a knowledge panel, or in a conversation with ChatGPT, Gemini, Claude, or Perplexity. The phenomenon isn’t new; featured snippets and panels retained clicks for years. What changed between 2024 and 2026 was scale and speed.
The economic model of the open web was built on an implicit trade: sites supply content to the search engine, the search engine returns visitors. Generative AI broke that trade. Systems keep consuming site content (to train, to retrieve information in real time, to ground answers), but return fewer and fewer visits. The information circulates; the click does not.
Zero-click is when a human searches and doesn’t click. Agentic search is when the human doesn’t even search: they delegate the task to an AI agent that researches, compares, and in some cases transacts on their behalf. These are two different phenomena, and the second, newer one grows on top of the first. This article deals mainly with the first, but the strategy recommended at the end prepares a brand for both.
Many numbers circulate on this topic, and not all of them measure the same thing. Before citing any statistic in a budget decision, it pays to know exactly what each study measured. The table below gathers the main independent studies, with source, period, and object of measurement.
| Source and date | What it measured | Core result |
|---|---|---|
| Pew Research Center (Jul/2025) | Real behavior of 900 U.S. adults, 68,879 Google searches (clickstream panel) | With an AI summary on the page, users clicked a traditional result in 8% of searches; without it, 15%. Clicks on sources cited inside the summary itself: 1%. Sessions ending right after the search: 26% with a summary vs. 16% without. |
| SparkToro / Similarweb (Jun/2026) | Clickstream of U.S. Google searches, Jan-Apr/2026 (desktop and mobile web, excluding the Google app) | 68.01% of searches ended without a click, up from 60.45% in 2024, the fastest acceleration in the series. On mobile, the rate approaches 77%; on desktop, around 50%. |
| Ahrefs (2025) | Large-scale Search Console data (study spanning 590 million searches) | 58% drop in position-one CTR when an AI Overview is present; average click reduction around 34.5% on affected queries. |
| Randomized field experiment (Apr/2026) | Users randomly assigned to see AI Overviews or not: causal design, not correlational | Estimated 38% cut in organic clicks when the AI summary is shown, with more searches ending without a click and lower reported satisfaction in the AI Mode group. |
| BrightEdge (Feb/2026) | Presence of AI Overviews across monitored queries, by industry | AI Overviews on about 48% of monitored searches, 58% year-over-year growth. In health, education, and research, presence approaches 80% of queries. |
| Similarweb (2026) | Referral rate by interface | Google’s AI Mode generates referrals on only 1.6% to 2.5% of queries, versus 17% to 19% for traditional search: the interface was designed to keep users inside the answer. |
Three readings matter more than any single number. First: the studies use different methods and reach the same direction, with behavioral panel (Pew), clickstream (SparkToro/Similarweb), Search Console data (Ahrefs), and a randomized experiment converge. When independent methodologies point to the same place, the trend is real. Second: the most underrated data point is the 1% of clicks on sources cited inside the summary. It shows that even the cited get little direct traffic: the value of being cited lies in presence inside the answer, not in the click it generates. Third: intent segments everything. Informational queries concentrate the zero-click; transactional and bottom-of-funnel queries still generate clicks at far higher rates.
In February 2024, Gartner published one of the market’s most-quoted predictions: traditional search volume would drop 25% by 2026, with search marketing losing share to chatbots and virtual agents. The line became a slide in thousands of decks, almost always without the full reading.
With 2026 here, the prediction did not literally come true. Google search volume did not plunge 25%: Google reacted by folding AI into search itself (AI Overviews, AI Mode) and still holds over 90% of the search market. Analyses published in 2026 revisiting the prediction show search evolved rather than collapsed: a relevant share of questions migrated to conversational interfaces (ChatGPT passed the hundreds of millions of active users mark), but Google absorbed the shift in-house.
Why dedicate a section to a prediction that missed its target? Because honesty with the data is exactly what separates analysis from fear marketing. Gartner’s prediction missed the magnitude and got the direction right: people keep asking as much as before (or more) but the answer increasingly arrives ready, synthesized, and click-free. The risk to brands was never the disappearance of search. It is their disappearance inside the answers.
Practical rule for consuming statistics in this market: always check whether a number measures search volume, click-through rate, referred traffic, or a forward-looking projection. They are four different quantities, frequently presented as interchangeable.
Zero-click is not a technical anomaly: it is the sum of three self-reinforcing forces.
For most informational questions, the AI-generated summary does the job. Pew observed that questions starting with “who,” “what,” “when,” and “why” trigger AI summaries about 60% of the time: exactly the query type that used to distribute clicks across blogs, portals, and institutional pages.
Every answer interface (AI Overviews, AI Mode, chatbots) is designed to resolve the journey inside itself. AI Mode’s referral rate (1.6% to 2.5% of queries, per Similarweb) is not a flaw: it is the product working as designed.
A generation of users now starts the discovery journey in conversational assistants (asking, refining, comparing) instead of typing keywords. Queries got longer, more contextual, and closer to the decision. And the longer the query, the more likely the answer arrives synthesized.
The combined effect is a change in the architecture of discovery: Google’s first page stopped being the market’s front door. The front door is now the answer, and it cites few brands.
Brazil, Flowup’s home market, is not behind this curve. It is one of its epicenters, and a useful preview for any market where AI adoption outpaces corporate response.
On the other side of the counter, corporate adoption hasn’t kept pace: 2026 surveys indicate the vast majority of companies have not yet structured their presence for this environment. The asymmetry is the opportunity: consumers migrated faster than brands, and the answers about each category are being formed now, based on the sources available today. Whoever structures first tends to become the reference the other answers repeat.
If the story ended at the traffic decline, the conclusion would be grim. But the same data showing the click disappearing shows something else: the click that survives is worth more.
| Source | Measurement | Result |
|---|---|---|
| Semrush (2025) | Conversion of visitors from AI search vs. traditional traffic | AI-referred visitors convert roughly 4.4x more. |
| Adobe (2026) | AI-referred traffic in U.S. retail | Conversion 42% higher in Q1, rising to 54% higher in a later 2026 measurement. |
| Leadster: Panorama PRO 2026 (Brazil) | Conversion rate by channel on Brazilian websites | AI search as the channel with the highest median conversion (7.8%) among those analyzed. |
The explanation is structural, not statistical: AI took over the middle of the funnel. The work users did by clicking ten links (comparing, filtering, understanding criteria, discarding options) now happens inside the conversation. When they finally click, they’ve been through triage, they’ve compared, they’re closer to the decision. AI isn’t stealing your funnel; it’s running the top and middle of it on its own, and handing the bottom to whoever it cites.
The indicator that matters is no longer “how many visits organic generates” but “how much qualified demand is born in answers that cite the brand.” A site can lose 30% of informational visits and, at the same time, grow organic-origin revenue, if it is present in the answers doing the triage. The reverse also holds: keeping top-of-funnel traffic while being absent from the answers means watching the competitor receive the decided visitor.
Put the two sides of the data together (the rare click and the valuable click) and the strategic conclusion draws itself: the unit of visibility moved from ranking position to presence in the answer.
In the previous model, you competed for the top of a list of links, and the reward was the click. In the current model, you compete for something narrower and more decisive: being one of the few sources the system understands, trusts, and cites when assembling the answer. That contest has its own rules:
For B2B companies and businesses that depend on organic acquisition, the end of the click reorders priorities on four fronts:
Editorial planning stops starting from keyword volumes and starts from the map of the market’s real questions: the ones customers ask assistants, in the context and language they use. Each strategic question needs a proprietary, direct, verifiable answer published on a brand asset.
The intermediate goal shifts from “attracting visits” to “being correctly understood by the systems”: well-defined entities, coherent structured data, a consistent narrative across site, press, and platforms, named and findable experts. This is the layer where an official brand knowledge base, structured for AI consumption, becomes a strategic asset, not a technical accessory.
Budget migrates from scaled production to evidence density: proprietary data, studies, cases with results, grounded opinions from real experts. Fewer pieces, more weight per piece, because AI synthesis compresses the generic and preserves the distinctive.
The dashboard gains new columns: impressions in AI experiences (already reported by Google Search Console), Copilot citations (Bing Webmaster Tools), brand presence in structured prompt tests, share of voice in the category’s answers. Rankings are still measured, but stop being the mother metric.
The sequence below is the one we apply in diagnostics and projects, and the order matters, because each layer supports the next.
A search that ends without the user clicking any result: the answer is delivered on the search page itself or inside an AI assistant. According to SparkToro, using Similarweb data, about 68% of U.S. Google searches ended without a click in early 2026, up from roughly 60% in 2024. On mobile, the rate approaches 77%. People are still asking; the click is what stopped happening.
No. What lost centrality was the click as the single metric. SEO fundamentals keep feeding both traditional results and AI answers. What changes is the goal of the upper layer: beyond ranking, being understood and cited as a source. Visibility and traffic stopped being synonyms: strategy must measure and work on both.
It depends on the measurement. Pew Research Center observed the click rate falling from 15% to 8% when an AI summary is present. Ahrefs measured a 58% CTR drop at position one on queries with an AI Overview. A 2026 randomized experiment estimated a 38% cut in organic clicks. The numbers differ because they measure different quantities, but all independent studies point the same way.
Not literally. Google search volume didn’t fall 25% by 2026: Google folded AI into search itself and kept the lead. What the prediction got right was the direction: billions of questions migrated to conversational interfaces and the click became rare. Search didn’t collapse; it changed shape.
Because the visitor who still arrives is worth more. Semrush measured roughly 4.4x higher conversion for AI-search visitors; Adobe reported 42% to 54% higher retail conversion across 2026; in Brazil, Leadster’s Panorama PRO 2026 found AI search to be the highest-median-conversion channel. AI does the triage and delivers a more decided visitor, to the brands it cites.
Start with first-party measurement: Google Search Console reports performance in generative AI experiences, and Bing Webmaster Tools reports Copilot citations. Complement with structured prompt tests: a fixed set of category questions, run repeatedly, recording when and how the brand is cited and alongside which competitors.
Diagnosis before tactics. Then, in order: technical foundation; answer-format content; the brand’s official knowledge base; external authority (mentions, proof, visible experts); and continuous citation measurement. Each layer supports the next: skipping steps is the most expensive way to not work.
Flowup’s B.I.N.A. Diagnostic measures, across four layers, how well search engines and artificial intelligences find, understand, and cite your company, with an immediate result and an in-depth report prepared by specialists.
Founder and lead strategist, Flowup Agency
Guto Bertoncini is the founder and lead strategist of Flowup Agency, which he has run since 2011. He is the author of the B.I.N.A. Method, Novo SEO and the Base Informacional Semântica (Semantic Information Base), and leads the agency's SEO for AI, GEO and AEO practice, preparing companies to be found on Google and cited by artificial intelligence platforms. He writes about search and AI on the Flowup blog and on his official website.
Sources verified on July 30, 2026. This article prioritizes independent measurements with declared methodology (behavioral panels, clickstream, Search Console data, and a randomized experiment) and explicitly identifies when a number is a projection, a correlation, or a causal measurement. Market statistics on search and AI change quickly; the figures cited reflect studies available up to the cutoff date and are presented with their source and collection period. No data point was extrapolated beyond what the original source reports.
