What a Zero-Click Search Is, and Which of Your Pages It Actually Costs
Most Google searches now end without a click. See how to tell whether your own traffic is affected, and what to check before you change anything.
/ On this page10 sections
Most Google searches now end without a click. Yours may not be among them, and working out which of your pages are exposed will tell you more than any tactic you could apply this week.
What it does to a particular site depends almost entirely on the kind of question that site answers.
What a Zero-Click Search Is
A zero-click search is a search where the person got what they needed, or gave up, without clicking anything on the results page.
The answer came from the page itself. A dictionary definition, a weather box, a Google Maps card, a currency calculator, a featured snippet lifted from somebody's article, or an AI Overview assembled from several.
That is the usual definition, and it hides something.
A search also counts as zero-click when the person clicked nothing because they were not finished. Somebody who skims the results, decides none of them fits, and types a different query has produced a zero-click search.
So the headline figure counts two different events under one name: the question that was answered, and the question that was abandoned.
The distinction matters when you are reading your own numbers, because only one of the two is a visit you lost.
How Much of Search This Now Is
Between January and April 2026, 68.01% of US Google searches ended without a click. That is SparkToro's measurement, taken from Similarweb's desktop and mobile clickstream panel.
The same study puts 2024 at 60.45% and 2019 at 49%.
Read that as a direction rather than a precise slope. SparkToro says so itself: the 2019 numbers came from the now-defunct Jumpshot panel, 2024 from Datos, and 2026 from Similarweb, which makes the comparison, in Rand Fishkin's own words, "a bit of apples and oranges".
Two limits on the figure are worth carrying with it.
It is a US measurement, so it says nothing certain about any other market.
It also excludes the Google mobile app, covering only mobile searches made in a browser. Fishkin expects the app is worse, which would make 68% an undercount rather than a ceiling.
None of that makes the number wrong. It makes it a number about searches, and your site is a smaller thing than that.
Use this chart — embed code and citation
<a href="https://neerajjivnani.com/blog/zero-click-search/"><img src="https://neerajjivnani.com/infographics/zero-click-search/three-readings-three-panels.png" alt="Column chart of the share of Google searches that ended without a click at three readings, drawn on one linear axis starting at zero where the full column height stands for 100 percent of searches: 49 percent in 2019 from the Jumpshot panel, which has since closed; 60.45 percent in 2024 from the Datos panel; and 68.01 percent between January and April 2026 from the Similarweb desktop and mobile panel. The empty space above each column is the share of searches that did send a click. Three cards below the columns carry the limits on the comparison: the readings come from three different measurement panels and three sets of users and devices, which Rand Fishkin calls a bit of apples and oranges; the 2024 and 2026 readings are US measurements of Google, scoped that way by the study, and neither carries a claim about any other market; and the 2026 reading excludes the Google mobile app and covers only mobile searches made in a browser, which Fishkin expects would make 68.01 percent an undercount rather than a ceiling." width="1200"></a>
<p>Chart: <a href="https://neerajjivnani.com/blog/zero-click-search/">Neeraj Jivnani</a></p>Neeraj Jivnani, "What a Zero-Click Search Is, and Which of Your Pages It Actually Costs", neerajjivnani.com, https://neerajjivnani.com/blog/zero-click-search/Free to republish with a link back to this page.
Why the Results Page Stopped Sending People Away
The results page stopped sending people away because Google started answering in place, and because answering in place works.
This is not new and it did not start with AI. Google has been putting answers on the page since it introduced universal results and rich snippets, and the zero-click share was already 49% in 2019, years before any generated summary existed.
What changed is the pace.
There is also an incentive worth naming plainly, because it explains why this will not reverse.
A searcher who gets an answer without leaving is a searcher who comes back and searches again. SparkToro found the share running a second search up 7.2 points between 2024 and 2026.
More searches is more of everything Google sells.
The Feature Everyone Expects to Blame Is Not the One Doing It
AI Mode is Google's chat-style search tab, and it is the obvious suspect. On the evidence it is not the cause, at least not yet.
SparkToro found that 0.34% of searches reached AI Mode in the same January to April 2026 window.
That is a rounding error against a 68% zero-click rate, and it is a useful correction: the clicks are going to ordinary AI Overviews and ordinary search features on ordinary results pages, not to a tab almost nobody has opened.
It will not stay that way. Google said at its 2026 developer conference that AI Mode had passed 1 billion monthly users, with queries more than doubling every quarter.
Those count different things, searches against people, so the second is not the first one growing. The direction is still not in doubt.
Plan for it. Do not blame it for last quarter's numbers.
What Answers the Question Instead of You
Several things on the results page can answer a question in place, and only some of them are new.
- AI Overviews. A written summary assembled from several pages, with links to some of them. The largest single cause, and the one growing fastest.
- Featured snippets. A passage lifted straight from one page and promoted above the results, with a link back.
- Knowledge panels. The box of facts about a person, place, company or thing, drawn from Google's own knowledge graph.
- People Also Ask. Related questions that expand in place, each one a small answer.
- Local packs. Three businesses, a map, opening hours and a phone number.
- Instant answers and tools. Currency conversion, unit conversion, a calculator, a definition, tonight's weather, the score.
Keep that last group in view, because it is the reminder that this is not only an AI story.
Nobody ever visited a website to find out what 40 dollars is in rupees. Those searches were never traffic, and the pages that used to catch them were catching something they had not earned.
That is worth separating from the searches you genuinely lost.
Which Searches It Takes, and Which It Leaves
Zero-click search is not spread evenly, and this is the part that decides whether it is your problem.
Ahrefs analyzed 146,122,391 desktop search results pages, captured in September 2025 and published that November, and found an AI Overview on 20.5% of them. That is the average, and the average is close to useless here, because the spread underneath it is enormous.
Where an AI Overview is likely:
- 57.9% of question queries carry one.
- 44.1% of medical queries in the your-money-or-your-life category do.
- 43.6% of science queries and 43.0% of health queries.
Where it is unlikely:
- 7.9% of local searches, the kind where somebody wants a business nearby.
- 6.3% of the most news-driven queries. If your subject moves weekly, this is barely your problem.
- Under 1% of navigational queries, where somebody is typing a brand to reach that brand.
Those are separate cuts of one census rather than a single list, so a medical question sits in two of them. Count a page once, under whichever describes its search best.
Even allowing for that, one site's exposure can be five or six times another's, on the same web, in the same month.
The Same Split, Reached a Different Way
Northwestern University's Medill Spiegel Research Center interviewed 21 digital marketing practitioners for a December 2025 report and arrived at the same conclusion from the opposite direction, without counting a single results page.
Its finding is that "AI search primarily decreases clicks for consumers with informational intent", while "consumers with commercial or navigational intent will still seek direct access to products or services on specific websites".
One of the practitioners it quotes put it more bluntly: "Information intent has become an extremely low click-through environment."
Then, on why that is: "Why would they click? They never wanted to. Now they don't have to."
Two methods, one answer. A census of what Google shows and a set of interviews with people watching their own dashboards both land on intent as the thing that decides it.
That is the only sorting we would act on. Not industry, not company size, not domain authority.
Sort your pages by the kind of question they answer, and the exposed ones separate themselves. A page that defines a term, converts a unit, states a fact or answers a common question is on the exposed side. A page that shows what you sell, proves you can do the work, or gets somebody to a login is not.
Use this chart — embed code and citation
<a href="https://neerajjivnani.com/blog/zero-click-search/"><img src="https://neerajjivnani.com/infographics/zero-click-search/nothing-is-average.png" alt="Horizontal bar chart of how often an AI Overview appears, by the kind of search being made, drawn on one linear axis starting at zero and running to a 60 percent ceiling, with a dashed line marking the 20.5 percent average across all results pages. Searches that answer a question at 57.9 percent, medical searches in the your-money-or-your-life category at 44.1 percent, science at 43.6 percent and health at 43.0 percent all sit well to the right of the dashed line. Local searches with near me in them at 7.9 percent, the most news-driven searches at 6.3 percent and navigational searches where somebody types a brand to reach that brand at under 1 percent all sit well to the left of it. No bar lands on the average, and a thick rule separates the four kinds above it from the three below." width="1200"></a>
<p>Chart: <a href="https://neerajjivnani.com/blog/zero-click-search/">Neeraj Jivnani</a></p>Neeraj Jivnani, "What a Zero-Click Search Is, and Which of Your Pages It Actually Costs", neerajjivnani.com, https://neerajjivnani.com/blog/zero-click-search/Free to republish with a link back to this page.
How Much a Page Loses
A page in first position loses roughly six of every ten clicks it used to get when an AI Overview appears above it, and that study has been run twice.
Ahrefs compared 150,000 keywords carrying an AI Overview against 150,000 informational keywords without one, using desktop Search Console clickthrough data from December 2023 and December 2025.
Its February 2026 conclusion: an AI Overview correlates with a 58% lower clickthrough rate where one appears.
That is up from the 34.5% the same study found when it was first run in April 2025.
Hold two things apart when you read it.
It is a correlation across a large sample of keywords, not a prediction about a page. And it applies only to results that carry an overview, which is one results page in five.
Some of What Looks Like a Loss Is a Completed Action
Not every uncaptured click is a customer who went elsewhere.
A local pack that shows your opening hours, your rating and a call button can end a search without a session in your analytics, and end it with a phone ringing.
The same is true of a booking module, a directions request and a click to your listing rather than your site. The activity is real and your analytics never sees it, which makes it easy to file under losses.
Check the Google Business Profile side of that before concluding a decline in sessions is a decline in business.
How to Find Out Whether It Is Happening to You
Open Search Console. It will now tell you how often your pages appeared inside AI answers, and reading it takes a few minutes.
Google added Search Generative AI performance reports to Search Console on 3 June 2026 and finished rolling them out to every site worldwide on 31 August 2026.
The report gives you a dedicated view of how often your pages appeared inside AI Overviews, AI Mode and the generative features in Discover, broken down by page, country, device and date.
Note what is in that list and what is not.
Google gives you impressions and it does not give you clicks. The announcement says the reports show impressions, pages, countries, devices and dates, and that it is considering "adding additional metrics over time".
So the instrument reproduces the problem it measures. You can now see that you were shown, and you still cannot see whether being shown cost you anything.
Comparing Your Own History
Comparing your own history is the way around the gap, because it needs no number Google withholds.
- Open the Performance report and set a long window, sixteen months if you have it, so you are comparing seasons rather than weeks.
- Sort your queries by impressions and look at the clickthrough column, not the position column.
- Find the queries where impressions held or rose while clicks fell. That shape means the results page changed while your position did not.
- Split those queries by kind. Questions and definitions in one pile, brand names, product terms, comparisons and anything local in another.
- Read the generative AI report against the first pile. If your exposed queries are the ones drawing AI impressions, you have your answer. If your clicks fell on brand and product terms, look elsewhere, because this is not what did it.
Step four is the one that turns a general worry into a finding you can act on, and it is the easiest to leave out.
What to Do About It
There are three things worth doing about it, in this order, and one popular answer that will not solve the problem you have.
Stop investing in the searches that were never going to visit. If a page exists to define a term or answer a question in two sentences, and the results page now answers it in two sentences, that page has lost its job. Retiring or repurposing it is cheaper than defending it.
Make the pages on the unexposed side better. Commercial, comparison, navigational and local searches still send people to websites, and they are where the clicks that remain now sit. That is where a rewritten title, a clearer promise and a faster page still pay.
On the pages you keep, write so the answer can be lifted. Lead a section with its answer, in one or two sentences.
That is not a contradiction of the first point. Where the click was never coming, being named in the answer is the only thing left to win, and it is ordinary practice rather than a new discipline.
The one to be careful with is brand building as the answer to a traffic problem.
It is real, and it is what makes somebody type your name instead of a question. It also works on a horizon of years rather than quarters.
Offering it to a site that lost a third of its sessions last quarter solves a different problem from the one being described.
The Questions People Ask About This
What is a zero-click search? A search where the person clicked nothing, because the results page answered them, or because they searched again instead.
Is it true that 60% of searches are zero clicks? That was SparkToro's 2024 figure for US Google, at 60.45%, and its most recent reading is higher still. Treat any bare percentage you are quoted as needing a year attached.
What are zero-click searches on Google? They are searches resolved by AI Overviews, featured snippets, knowledge panels, People Also Ask boxes, local packs and the small tools like calculators and converters, plus the ones where nothing on the page fitted and the person searched again.
What is a zero-click strategy? For most sites it is a sorting exercise before it is a strategy. Work out which of your pages answer the kind of question Google now answers itself, stop investing there, and spend the difference on the pages that still get visited.
Where That Leaves You
Zero-click search leaves you with one question you can settle yourself in an afternoon, and it is worth settling before you spend anything.
Two thirds of Google searches ending without a click is a fact about Google. Whether it has cost you anything is a different fact, about you.
If your pages answer questions, this is happening to you, and the honest response is to move investment rather than to defend the ground.
If your pages sell something, prove something, or serve a place, the numbers say you are mostly watching somebody else's weather.