Zero-Click Search and AI Overviews: Impact on SEO Traffic

Zero-Click Search ends the query on Google’s own results while the click never reaches the site, so AI Overview citations register impressions that have no session records. A zero-click search is one in which the user finds the answer directly on the SERP (search engine results page) and does not click through to any external website. SparkToro’s Similarweb-panel analysis puts US zero-click at 68.01% across January to April 2026, up from 60.45% in 2024.

Google answers queries on the SERP through several answer surfaces that resolve the query and suppress clicks. Answer-first content structured for extraction earns a citation inside the result. Measurement then shifts to impressions-versus-clicks divergence and brand-search lift. Recovery re-targets the query types AI Overviews cannot resolve, which is where answer-first content carries weight for Answer Engine Optimization, the practice of optimizing for the systems that answer a query instead of listing results.

All panel figures and Search Console behavior on this page reflect Google Search as of 5 October.

What is a zero-click search?

A zero-click search is one that ends on the SERP because the answer appears there, and the searcher does not click through to any external website. A click counts as a tap on an organic listing, an ad, or a link that loads an external site, while in a zero-click search, the answer lives inside the SERP itself. Zero-click search is also called ‘no-click search’.

An AI Overview (Google’s AI-generated answer at the top of results) produces a zero-click search when the answer is resolved on the SERP, so the search ends without a click to an external site.

Zero-click answering extends beyond Google to eight AI platforms and assistants:

• ChatGPT resolves queries in-chat.
• Perplexity returns cited answers.
• Claude answers within its interface.
• Gemini surfaces direct responses.
• Bing Copilot composes on-page replies.
• Alexa speaks answers aloud.
• Siri returns spoken results.
• Google Assistant delivers answers directly.

Each of these ends the query inside its own interface.

An AI Overview differs from a featured snippet in how the answer is generated and cited. If you build pages for snippets, that difference is the one to read first: featured snippet vs AI Overviews.

Three different things get called “zero-click”, and the figures are not interchangeable:

A triggered AI Overview still leaves the click available. Trigger rate is a ceiling on exposure, not a zero-click rate.

Zero-click searches now span Google and multiple assistants, raising the question of why they occur.

Why do zero-click searches happen?

Zero-click searches happen because Google answers queries directly on the SERP through several answer surfaces. Google designs these surfaces to resolve intent in the SERP. This reflects Google’s product choice, not a failure of a site’s content.

Eight SERP surfaces end sessions without a click:

• AI Overviews (AI-generated answers at the top of results)
• Featured snippets (extracted answer boxes)
• Knowledge panels (entity fact boxes)
• People Also Ask boxes (expandable related questions)
• Local packs (map-based business listings)
• Instant tools and widgets (calculators, converters, weather)
• Voice search (spoken answers from assistants)
• Mobile-first layouts (compact result screens)

Semrush’s AI Overviews study measured AI Overviews triggering on 6.49% of queries in January 2025 and 13.14% in March 2025.

AI Overviews appear when specific query signals and conditions are met. Before you diagnose a traffic loss, check whether the query even produces one: the signal set is on what triggers Google AI Overviews.

AI Overviews turn on five signal classes: query intent, phrasing, topic category, suppressors, and trigger volatility over time.

AI Mode extends zero-click behavior by structure: a question sequence that would once have been several searches becomes one session with one set of citations. The effect is real and currently small. SparkToro measured 0.34% of searches transitioning into AI Mode between January and April 2026. Google reported at I/O 2026 that AI Mode had passed one billion monthly users with query volume more than doubling each quarter, so the mechanism matters more than the current share suggests. The two surfaces cite differently, which is covered on AI Overview vs AI Mode.

AI Overviews retrieve indexed passages and ground the generated answer in them, then fan a single query into sub-queries so that the retrieval reaches beyond the words the searcher typed. Both steps move the answer off the source page and into the result. That is where the click stops. For the mechanics of each step, the retrieval side is on retrieval-augmented generation and the decomposition side on query fan-out.

How many searches end in zero clicks?

About two-thirds of Google searches now end without a click. SparkToro’s Similarweb-panel analysis puts US zero-click at 68.01% across January to April 2026, up from 60.45% in 2024. The share of searches producing at least one click (organic, paid, or to a Google property) fell 9.51 percentage points over those two years, a 22.9% relative decline. Meanwhile the share of searches leading to another Google search rose 7.2 points.

Zero-click search rose from 24.4% in March 2024 to 27.2% in March 2025 across the U.S., and from 23.6% to 26.1% across the E.U./UK, tracking the expansion of AI Overviews. Those figures count searches with no click at all, the second row of the table above. Zero-click exposure also varies by vertical.

Exposure is concentrated in Science, Health, and Pets & Animals, where AI Overviews trigger on 43.6%, 43.0%, and 36.8% of queries, compared with a 20.5% all-SERP baseline (Ahrefs, 146 million SERPs). Trigger rate is the ceiling on exposure, not the zero-click rate itself. A triggered overview still leaves the click available.

SparkToro reports 68.01% of January-to-April 2026 US searches ending with no click to a site outside Google, against 27.2% of March 2025 searches ending with no click at all. The gap between those two definitions is the share of clicks that go to YouTube, Maps, and other Google properties.

Each of those figures comes from a panel rather than from Google.

Is the zero-click figure a measurement or an estimate?

Every published zero-click figure is a panel estimate, and the panels have changed three times. The series does not run in one direction and it does not run on one methodology:

The number went up, then down, then up again, and the 2024 figure moved five points when the provider changed, without anything happening in Search. SparkToro states directly that comparisons across these studies are not equivalent because the providers, panels and methodologies differ.

The 2026 study also excludes searches made inside Google’s mobile app, where SparkToro says zero-click behavior may be higher still, and it assumes a two-thirds mobile / one-third desktop split rather than measuring it. Treat 68% as the best current estimate of a US desktop-and-mobile-web panel, not as a measurement of Google.

Trigger-rate figures behave the same way. Semrush and Ahrefs report different AI Overview rates because they use different denominators and time windows, not because they disagree.

Where do the clicks go when a search ends on Google?

Zero-click searches (searches that end on the results page without a click to an external site) redirect clicks toward Google-owned properties rather than the open web. Zero-click searches route 14.3% of U.S. clicks in 2025 to Google-owned properties, such as Google Maps and YouTube, rather than to the open web, per Rand Fishkin / SparkToro. The EU/UK split reached 12.6% in 2025, per the same SparkToro clickstream analysis. Search Console folds these movements into aggregate impression and click totals.

Search Console’s Generative AI report, launched 3 June 2026, separates impressions earned inside AI Overviews and AI Mode from the aggregate, broken out by page, country, device and date. It carries no clicks, no click-through rate, and no query dimension, and it blends AI Overviews with AI Mode rather than splitting them.

How do zero-click results reduce search traffic?

Zero-click results reduce search traffic by resolving the query on the SERP, so impressions still register while the click that would have reached the site never happens. Flat impressions with falling clicks mean the SERP answers the query in place, breaking the impression-to-click step of the funnel, not rankings or demand.

The presence of an AI Overview lowers organic click-through rate for affected queries. SparkToro found that AI Overviews now appear on more than 20% of Google searches, and that click-through rate drops by close to 60% when they do. AI Overviews absorb queries such as informational-intent queries, definitional or quick-fact queries, branded searches, local business queries, and high-intent transactional searches. A zero-click answer can end one search and start the next, so a single displayed answer generates queries the site never receives a click from.

Answer-first content recovers some ground because AI Overview citations drive measurable referral traffic with a different conversion quality than blue links; the full method is in how you optimize for zero-click searches, below. You confirm the damage in Search Console, where impressions climb while the click line drops. Lower CTR and interrupted sessions cut the qualified visits that a marketing funnel depends on.

Do the clicks that survive an AI Overview convert better?

The mechanism is plausible and the outcome is unproven. Zero-Click Search (a search that ends on the SERP with no outbound click) removes the informational tail, so the remaining clicks should skew toward users who are ready to compare, download, or buy. SparkToro found that AI Overviews now appear on more than 20% of Google searches, and that click-through rate drops by close to 60% when they do.

Fewer clicks do not automatically mean better clicks. An answer already given on the results page removes the visitor who only wanted that answer. Whether what remains converts better is a separate question, one that requires conversion data from a pre-2024 baseline to answer.

If the surviving clicks cannot be counted on to convert harder, the page has to be built to win something else.

How do you optimize for zero-click searches?

Publish answer-first content structured for extraction, so Google cites the page in the SERP instead of routing a click to it. Zero-click strategy trades the click for citations and brand exposure, so the payoff is citations, brand mentions, and assisted conversions rather than visits. The core moves are 40 to 60 word direct answers opening each section, structured formatting, and lists and tables.

Three sub-methods carry the framework.

Retrievability and passage quality. AI Overview cites sources that meet retrievability and passage-quality criteria. For the selection criteria themselves rather than the traffic consequence, see how to be cited in AI Overviews.

Structured data. Structured data increases eligibility for rich results and for AI extraction. Six schema types carry most of that weight:

• FAQPage
• HowTo
• LocalBusiness
• Product
• Review
• Organization

Implementation per type, including which of the six applies to which page, is on schema markup for AI extraction.

Topical authority. Pillar pages and cluster content raise the odds of being selected as an AI Overview source, which Semantic SEO (organizing content around entities and their relations) structures. If you need the build order rather than the principle, it is on topical authority.

Whether link authority conditions that selection depends on the vertical, and the split is sharper than any single number suggests.

Across 100 queries checked on 31 August 2026, 91 returned an AI Overview. On those, cited pages and uncited top-10 pages from the same SERPs were compared at the page level, by referring domains. Split by topic, the two behave nothing alike, and neither behaves the same at the median as in the tail.

Across the 20 health and finance queries, cited pages carried a median of 19 referring domains against 3 for the uncited pages ranking alongside them, and at the 90th percentile 1,819 against 365. The advantage holds throughout the distribution.

For the other 80 queries the median advantage disappears: 0 referring domains for cited pages against 1 for uncited. The tail does not disappear. At the 90th percentile, cited pages carried 1,382 referring domains against 203, a 6.8× gap wider than YMYL’s.

Read whole, the set shows no link advantage at all: median 1.0 on both sides, a ratio of 1.0×. Read split by topic and by percentile, it shows two mechanisms. The aggregate number would have been the wrong answer, and it would have looked like a clean one.

Outside health and finance, AI Overviews reach pages with no links at all and pages with a great many, and comparatively few in between.

Both groups came from the same result pages, and the ranking positions were close enough to read the comparison cleanly, with cited pages at a median rank of 4 versus 5 for uncited ones.

Of the 684 cited pages, 323 did not appear in the organic top 10 of the same SERP: 47.2% of citations went to pages that were not ranking on the page they were cited on.

Which zero-click queries are worth pursuing?

Pursue the ones where citation is reachable and the click still has somewhere to land. Two tests from the citation study decide most cases.

Is it health or finance? In those verticals cited pages carried a median of 19 referring domains against 3 for their uncited neighbours. Without link authority, citation there is unlikely regardless of content quality.

Is it anything else? Outside health and finance the median cited page carried no referring domains at all, and 47.2% of citations went to pages outside the organic top 10. Citation is reachable on content alone.

Then check what survives extraction. A definitional query with nothing on the page but the definition converts a citation into nothing. A query whose answer points at a calculator, a dataset or a tool keeps the visit even after the summary takes the explanation.

What content format survives an AI Overview summary?

Content formats that survive an AI Overview summary are the assets Google cannot reproduce inside the SERP. Answering fully earns the AI Overview citation. The residual value stays off-SERP, held inside experiences that the summary cannot copy. Five content formats survive:

• Calculators and interactive tools
• Quizzes
• Proprietary data and original research
• Downloadable resources
• In-depth analysis and visuals

Each of these formats keeps something on the page that a summary cannot carry across: a computation the reader has to run, a dataset they have to query, a tool that only works where it lives. The extraction takes the explanation and leaves the function behind.

A passage that front-loads the direct answer works with these formats by placing a 40 to 60 word answer at the top of each section, which the AI Overview extracts, while the interactive asset holds the click. For how to write that opening passage, see answer-first content.

How do you measure zero-click search impact?

To measure zero-click search impact, you track three signals: impressions-versus-clicks divergence, zero-click checkers, and brand-search lift. Google Search Console shows impressions and clicks in aggregate, so a keyword with high impressions and near-zero clicks is the zero-click signal, but the report can mislead, because it folds AI Overview impressions into totals without isolating them. Search Console isolates AI Overview impressions from standard totals in its Generative AI report. Zero-click keywords are identified by SERP-feature ownership and impression/click divergence, using Position Tracking’s SERP feature filter alongside the Search Console impressions-versus-clicks comparison.

Five KPIs replace the click as the unit of account. Each needs a place to read it and a number to read it against:

• AI Visibility Score: how often your domain appears across AI answers. There is no Search Console equivalent; it comes from a tracking tool.
• SERP Feature Share: the percentage of tracked queries where your domain owns a SERP feature. Read it in Position Tracking with the SERP feature filter.
• Brand Mention Sentiment: the tone of unlinked brand references, read in a mention tracker.
• Question Coverage Rate: the share of the question set around your topic that your pages answer directly.
• Multi-Platform Visibility: the same citation check runs outside Google, through Prompt Tracking and a visibility overview.

The benchmark for the first two comes from your own SERPs rather than a published average. In the 100 queries measured here, 47.2% of AI Overview citations went to pages outside the organic top 10, so a feature-share number read only against your ranking positions understates what is reachable. For the tools that record citations per surface, see AI visibility tracking.

Once zero-click impact is measured, the open question is how you attribute conversions consumed inside an AI Overview.

How do you attribute conversions consumed inside an AI Overview?

To attribute a conversion consumed inside an AI Overview, you credit three proxy signals rather than a last-touch session: assisted conversions, branded search demand lift, and direct traffic lift. No session means no last-touch row exists to fill, so credit moves to assisted conversions (conversions where the channel appeared earlier in the path) and branded demand signals.

Since 3 June 2026 the impression side no longer needs a proxy. Search Console’s Generative AI report shows which of your pages appeared inside AI Overviews and AI Mode, broken out by page, country, device and date, for the subset of properties in the rollout. What it does not show is clicks, click-through rate, or the query behind the impression, and it blends AI Overviews with AI Mode. So the three proxies above still carry everything downstream of the appearance. They are no longer carrying the appearance itself (Google Search Central).

Can a zero-click AI Overview build brand recall?

An unclicked AI Overview citation builds brand recall by placing your brand name in front of the searcher even when no session is recorded. The AI Overview citation creates brand exposure not captured as a session. Three proxies measure that exposure, such as branded search demand, brand mention sentiment (tone of unlinked brand references), and share of voice (portion of AI Overview citations your brand holds).

AI Overview source trust rises with E-E-A-T and branded mentions, built through expert bylines, updated references, clear sourcing and subject-matter experts. For how those mentions are tracked and built, see brand entity signals. Unclicked impressions are a measurable value through those three proxies. But the exposure cannot be fully controlled because the AI Overview selects and truncates sources. Can a brand remove itself from that surface entirely?

Can you opt out of AI Overviews?

Since 3 June 2026 a Search Console setting removes a domain from AI Overviews, AI Mode and Discover’s generative features. Google added the control under a UK Competition and Markets Authority conduct requirement, began honoring the setting on 17 June, published documentation in July, and has since extended the rollout past the UK to the US and other international properties. It sits under Settings, in a section called Search generative AI, and offers a domain-level choice: include the site in AI Overviews, AI Mode, and Discover’s generative AI features (the default), or exclude it. Excluded sites stop appearing as links and stop grounding those answers. Google states that the setting is not used as a ranking or inclusion signal anywhere else in Search. Changes take one to two days to propagate. Page-level control is scheduled for March 2027.

Two things sit outside the toggle. It does not govern model training, which still runs through Google-Extended. And the “Google-Extended” token does not govern AI Overviews either, which is the reverse mistake and the more common one. Google-Extended applies to Gemini apps and Vertex AI grounding. Disallowing it does not change whether a page appears in an AI Overview.

The “nosnippet” directive removes the page from featured snippets and text previews across Search, not from AI Overviews in isolation.

Leaving is not a clean exit. The cost is not only AI traffic. Top Stories carousels have been observed rendering inside AI Overviews on US trending news results, so a news site opting out may be opting out of both. Publishers who depend on informational queries absorb this cost the most. Accuracy is the other half of the decision: what an AI Overview gets wrong about a brand is covered on AI Overview accuracy.

Are zero-click rates higher on mobile than on desktop?

Zero-click rates are higher on mobile than on desktop. Zero-click search, a search resolved on the SERP without a click to a website, reached 77% on mobile devices against 47% on desktop, per Up & Social. The device gap runs close to thirty points in favor of mobile.

Mobile and desktop cannot be compared on click-through rate alone. Ranking position differs between the two on most properties, often by several positions, and position moves click-through rate far more than device does. Search Console reports no AI-Overview-specific click data on either device, so a device split measures the query mix and the position gap together with whatever the AI Overview is doing, and separates none of them.

Both figures are panel-based clickstream estimates, not Search Console data, and no property owner can reproduce them from their own account.

A device gap alone does not confirm the cause of a traffic drop. How do you tell AI Overview traffic loss from other causes?

How do you tell AI Overview traffic loss from other causes?

To tell AI Overview traffic loss from other causes, you compare impressions against clicks in Google Search Console, check whether an AI Overview actually triggers on the affected query, and match clickstream data against Google Search Console figures. An AI Overview loss holds impressions steady or rising while clicks fall, because Google still displays the page but answers the query on the SERP. A ranking drop pulls both down together; a technical fault removes impressions outright. Clickstream data (panel-based records of what users click after searching), the kind Rand Fishkin publishes through SparkToro, confirms the on-SERP resolution behind Zero-Click Search.

Those are two different diagnoses in one account. Desktop lost impressions and position together, which is a ranking problem. Mobile held its ranking, gained visibility, and still lost clicks. That combination is the one a ranking drop cannot produce.

One page shows it in isolation. The site’s midi-dress collection page held position 2.4 to 2.5 across the two windows while impressions rose from 169 to 247 and clicks fell from 8 to 2. Checked on 31 August 2026, the head query for that page returns an AI Overview in position 1, citing four competing collection pages before any organic results appear.

Google displays the AI Overview (Google’s generative answer block at the top of the SERP), so the query resolves without a session.

How do you recover traffic lost to AI Overviews?

To recover traffic lost to AI Overviews, publishers re-target the query types AI Overviews cannot resolve. Publishers apply three recovery moves: shifting to bottom-funnel and transactional queries (searches with buying intent), expanding beyond summarizable informational content, and diversifying to non-Google surfaces (platforms such as ChatGPT, Perplexity and Bing). LLM SEO extends reach onto those non-Google surfaces. Rand Fishkin’s SparkToro clickstream data confirms the drain, so publishers rebuild volume where extraction cannot compete.

Where the numbers on this page come from

Published figures and first-party measurements are kept separate throughout.

Panel figures come from SparkToro’s clickstream studies, most recently a Similarweb panel covering US Google searches from January to April 2026. Trigger-rate figures come from Semrush’s AI Overviews study (March 2025) and Ahrefs’ analysis of 146 million desktop SERPs (September 2025). Where those sources disagree, both are shown.

Two first-party tests:

The YMYL bucket was defined positionally as the first twenty keywords in the source file, which were the ten health and ten finance queries. Referring domains were counted at the page level; a domain-level count would have measured which large sites appear in the set rather than which pages earned links. URLs appearing in both the cited and uncited sets were deduplicated before fetching.

What limits these numbers

The panel figures are estimates, not measurements. Three providers across seven years, not directly comparable, and the 2026 study excludes Google’s mobile app and assumes rather than measures the device split.

The citation study is one snapshot of 100 queries. AI Overview citations change between generations, so a single pass understates how much the cited and uncited pools overlap. The YMYL bucket was defined positionally rather than semantically.

The retailer case is one account. It shows a pattern a ranking drop cannot produce; it does not establish how common that pattern is.

Everything above is about a click that does not happen and what to do about it. The questions below sit just outside that: each comes up whenever zero-click is discussed, and each links to the page that covers it properly.

Related questions about zero-click search

Does zero-click affect paid search too?

Yes. The decline SparkToro measures covers paid clicks as well as organic ones. The 9.51-point fall between 2024 and 2026 is in searches producing at least one click of any kind: organic, paid, or to a Google property. A query resolved inside an AI Overview removes the ad impression’s downstream click along with the organic one.

Which industries lose the most to zero-click?

Exposure is highest where AI Overviews trigger most, and Science, Health and Pets & Animals lead at 43.6%, 43.0% and 36.8% against a 20.5% baseline. Health and finance carry a second cost: in the citation study here, cited pages in those verticals held a median of 19 referring domains against 3 for their uncited neighbours, so exposure is high and citation is gated. The category-by-category rates are on what triggers Google AI Overviews.

Was zero-click a problem before AI Overviews?

Zero-click predates AI Overviews by at least five years. The 2019 Jumpshot figure was 49%, revised to 50.33%, and the 2020 Similarweb figure was 64.82%. AI Overviews launched in May 2024. Featured snippets, knowledge panels and local packs were ending searches on the SERP long before a generative answer did. AI Overviews accelerated the trend rather than creating it.

Zero-click search FAQ

What is zero-click content?

Zero-click content is content written to be resolved where the reader already is, on the SERP or inside an AI answer, rather than to pull a click. On a website that means a direct answer at the top of each section, with the part a summary cannot carry across kept on the page.

Do zero-click searches hurt rankings?

No. A zero-click loss holds impressions steady or rising while clicks fall, which is the opposite of a ranking drop. The page still ranks; the query is answered before the click happens.

Can you track AI Overview clicks in Search Console?

No. The Generative AI report launched on 3 June 2026 shows impressions by page, country, device and date, and it carries no clicks, no click-through rate, and no query dimension.

Does opting out of AI Overviews affect organic rankings?

Google states that the Search generative AI setting is not used as a ranking or inclusion signal anywhere else in Search. What it does cost is the AI Overview and AI Mode surfaces themselves, and on trending news results possibly the Top Stories carousel that renders inside them.

What to do with this

Zero-click is not a ranking problem and it does not have a ranking fix. Confirm it first: impressions holding or rising while clicks fall, on a query that actually returns an AI Overview. Then decide what you are optimizing for. In the 100 queries measured here, 47.2% of AI Overview citations went to pages that were not in the organic top 10 of the same SERP, and outside health and finance most cited pages carried no links at all. Citation is reachable without rank and without authority. What it is not is a click. Plan for the citation, and keep something on the page a summary cannot carry across.