Why Every Zero-Click Search Statistic You Have Seen Is Different

What is zero click search?

Searchers end a zero-click search without visiting any external site, because the results page or an AI answer already gave them what they came for. The query happened, the impression counted and the answer landed. Only the click is missing. Google, Bing and every AI assistant produce these searches, and the share of all searches that end this way is the most quoted and most disputed number in search marketing.

What ends a search on the results page?

Four result formats answer the question before the user reaches a link: featured snippets, knowledge panels, direct answers and AI Overviews. A featured snippet lifts a paragraph, list or table out of a ranking page and prints it above the results. A knowledge panel assembles facts about an entity from Google's Knowledge Graph, so a query like "population of Portugal" resolves inside the panel. Direct answers cover calculations, unit conversions, weather, sports scores and dictionary definitions, rendered by Google with no source page involved. AI Overviews merge several pages into one generated passage with inline citations. Audit which of the four appears on your top queries before you assume that ranking first still sends traffic.

Why do published zero-click numbers disagree so much?

The published figures disagree because each study defines a search differently, treats clicks to Google's own properties differently, and draws on a different panel. Semrush classified a jump to Google Images or Google News as a "Google click" and kept it out of its zero-click bucket, while SparkToro's 2026 analysis counted a visit to YouTube or Google Maps as a click that ended the search. Read the sample column before you quote the figure in the same row.

Figure	Source	Sample and counting rule	Data period and publication
68.01% of US Google searches ended without a click	SparkToro (Rand Fishkin), using Similarweb data	US desktop and mobile web clickstream panel. Clicks to Google properties, organic links and paid links all count as clicks	Data January to April 2026, published 9 June 2026
69.5% UK, 68.0% US, 65.3% France, 63.8% Canada, 63.4% Italy, 62.1% Germany	SparkToro, using Similarweb data	Same panel, split by country	Data January to April 2026, published 17 June 2026
58.5% of US searches and 59.7% of EU searches ended without a click	SparkToro, using Datos clickstream data	Datos US and EU panel, tens of millions of users. Fishkin flags thin iOS coverage and ad-blocker skew	Data September 2022 to May 2024, published 2 July 2024
25.6% of desktop searches and 57% of mobile searches ended without a click	Semrush	609,809 unique search actions. Clicks to Google properties counted in a separate bucket, not as zero-click	Data May 2022, published 25 October 2022
Users clicked a result link on 8% of visits carrying an AI summary against 15% without one. 26% of AI-summary visits ended the browsing session against 16% without	Pew Research Center	900 US adults on KnowledgePanel Digital, 68,879 Google searches, 12,593 of them producing an AI summary	Data March 2025, published 22 July 2025
"Total organic click volume from Google Search to websites has been relatively stable year-over-year"	Google, Liz Reid, VP and Head of Search	Google's internal data. No sample, panel or query set disclosed	Published 6 August 2025

Last verified: 8 September 2026

Do not average these rows. The 25.6 percent desktop figure and the 68.01 percent figure measure different things on different populations four years apart, so a blended number would describe no real search behavior at all. Quote the row whose counting rule matches the question you are answering. For the adoption numbers behind these click studies, ChatGPT's reach and AI Overviews usage among them, see the sourced AI search statistics.

What has Google said about the zero-click figures?

Google disputes the third-party decline reports on the record. Liz Reid, VP and Head of Search, wrote on 6 August 2025 that "total organic click volume from Google Search to websites has been relatively stable year-over-year" and set that against "third-party reports that inaccurately suggest dramatic declines in aggregate traffic," which she attributed to "flawed methodologies, isolated examples, or traffic changes that occurred prior to the roll out of AI features in Search." Google named no study, disclosed no sample and published no query set, so nobody outside Google can check the claim against the panels above. Cite Google's position beside the panel figures, not in place of them.

How do you calculate a zero-click rate?

Divide the searches that produced no click by the total searches for the same query set, then multiply by 100.

Zero-click rate = (searches - clicks to any destination) / searches * 100

Take a keyword drawing 10,000 monthly searches that produces 2,400 clicks across every result on the page. That gives (10,000 - 2,400) / 10,000 * 100, a 76 percent zero-click rate. The rate moves with the query, not with your position: a definition query answered by a knowledge panel runs far above a query where the searcher wants to compare vendors. Calculate it per query group instead of site-wide, because one blended figure hides the buying queries inside a mass of question queries.

Turning a keyword's numbers into a zero-click rate

1

Start with total searches. The keyword draws 10,000 monthly searches.

2

Count clicks to any destination. The same keyword produces 2,400 clicks across every result on the page.

3

Subtract clicks from searches. 10,000 minus 2,400 leaves 7,600 searches with no click.

4

Divide, then multiply by 100. 7,600 divided by 10,000, times 100, gives a 76 percent zero-click rate.

Does a zero-click search still count for a brand?

A zero-click search still moves a brand when the answer names the brand, because the searcher reads the name whether or not a visit is recorded. AI Overviews, ChatGPT, Claude and Gemini all print brand names and sources inside the answer text, so the exposure lands somewhere your analytics cannot see it. Pew's March 2025 data shows how quiet that side goes: users clicked a link inside the AI summary itself on 1 percent of visits. Track AI referral traffic as its own line rather than folding it into organic, and treat a named mention in an answer as the unit you are counting.

What should a brand do about zero-click search?

Move the target from sessions to citations, and write pages in the shape answer engines quote. Answer engine optimization covers the writing side: make the heading the question a person types, put the complete answer in the first sentence under it, and keep each claim short enough to lift whole. The measurement side needs its own instrument, since your analytics never sees an answer that sent no visit. LLM visibility tracking closes that gap by recording which assistants name you, on which prompts, and which sources they cite next to you.

RedReplier tracks the Reddit, X, Bluesky, Facebook and Hacker News threads naming your brand in one place, and shows which of those threads rank on Google for your keywords. You add a website and your keywords, and RedReplier surfaces the mentions ranked by buying intent, with an AI explanation of why each one was flagged. Email alerts arrive on your plan's interval, so a thread asking which tool to buy reaches you while it is still open.

Frequently asked questions

Is a zero-click search the same as a lost customer?

No. A zero-click search means no visit was recorded, not that no influence occurred. When an AI Overview or a snippet names your brand, the searcher carries that name into a later branded search or a direct visit. The loss is in your attribution, and you close it by counting mentions in answers, not only sessions in analytics.

Which zero-click statistic should I quote?

Quote the one whose counting rule matches your question, and name the source and date in the same sentence. For "how much US search traffic reaches the open web in 2026", SparkToro's 68.01 percent for January to April 2026 is the closest fit. For "what an AI summary does to click rates in a controlled panel", Pew's 8 percent against 15 percent is the better citation because its methodology is published in full.

Do AI Overviews cause zero-click searches?

AI Overviews raise the rate, and they are not the origin of it. Zero-click searches predate them by years, which is why the 2022 Semrush study already found 57 percent of mobile searches ending without a click. Pew measured the added effect directly in March 2025: 26 percent of visits to a results page carrying an AI summary ended the session, against 16 percent of visits without one.

Can Google Search Console show my zero-click rate?

Not the true rate. Search Console reports impressions and clicks for your site alone, so the gap between them mixes real zero-click searches with clicks that went to a competitor. Use it to find queries where impressions climbed while clicks fell, then open those SERPs by hand and record which feature took the answer.

Do zero-click searches happen outside Google?

Yes. Bing, DuckDuckGo, YouTube search, Amazon search and every AI assistant resolve queries without sending anyone to an external site. ChatGPT, Claude and Gemini push the pattern furthest, since the default outcome of a prompt is a written answer with citations that most users never open.

Which queries have the highest zero-click rates?

Definitional, factual, local and one-word head terms sit at the top, because the answer fits in a snippet, a panel or a map card. Similarweb's 10 August 2026 analysis put the head term "sunglasses" at a 77 percent zero-click rate and the query "Stockholm" at 85 percent. Comparison, review and buying queries hold the lowest rates, which is where the traffic worth chasing still lives.

What counts as a click to a Google property?

Semrush and SparkToro disagree on exactly this point. Semrush treats a jump to Google Images or Google News as a "Google click" and excludes it from its zero-click bucket, while SparkToro's 2026 analysis counts a visit to YouTube or Google Maps as a click that ended the search. That single decision changes which rows in the table look bigger or smaller, before any other methodology difference even enters the picture.

Why do zero-click rates differ so much by country?

SparkToro's Similarweb panel found the UK at 69.5 percent and Germany at 62.1 percent, with France, Canada and Italy landing between the two. The numbers come from the same panel, the same counting rule and the same period, January to April 2026, so the gap reflects real differences in how each country searches rather than a change in method.

How often do users click a link inside an AI Overview itself?

Pew's March 2025 data puts that figure at 1 percent of visits. Almost all of the exposure an AI Overview gives a brand happens in the reading, not in a click, which is why counting a named mention matters more than waiting for a visit that these numbers say rarely arrives.

What is the highest and lowest zero-click rate published?

SparkToro's 2026 Similarweb analysis puts the ceiling at 68.01 percent of US searches ending without a click. Semrush's 2022 study puts the floor at 25.6 percent for desktop searches specifically, measured four years earlier on a different panel with a different counting rule. Both numbers are correct for what they measured, and the gap between them is a methodology gap, not a trend line.