The Changing Value of a Click: The Zero-Click / High-Conversion Paradox

The majority of Google searches now end without a click to the open web. Clickstream panels measure the change at population scale. The outcome is also structural. Informational queries make up 50-70% of all search volume (Semrush measures 53%, and SE Ranking corroborates at 70%). An informational query is the intent type that resolves without a click most easily, so the base rate guarantees the outcome.

SparkToro and Datos (a Semrush company) analyzed a multi-million-device clickstream panel in March 2024 and found that for every 1,000 US searches, only 360 result in a click to the open web. 58.5% of searches end with zero clicks. 24.4% end without any click at all, not even to a Google property.

SparkToro measured the rate again in June 2026, and reported 68.01% for US searches between January and April 2026. Rand Fishkin states that ten years ago the rate was approximately 45%.

Do not subtract the two numbers. The 2024 study used the Datos panel, and the 2026 study used a Similarweb panel. SparkToro states that these "aren't the same users or devices". The 2026 measurement also excludes the Google mobile search application, so SparkToro expects the true rate to be higher.

The 2026 model also estimates three of its inputs. SparkToro assumes a mobile split of two thirds, a 10-second inactivity window, and a paid click share from a third-party benchmark. Read 68.01% as a within-panel Similarweb measurement, and read 58.5% as a within-panel Datos measurement. Both point the same direction.

By Q4 2025, the zero-click rate had climbed further. Datos' State of Search report showed zero-click searches rising to 27.2% in the US (up from 24.4%), while Google desktop searches per US user fell nearly 20% year-over-year. The two trends form a double headwind. Search volume per capita falls, and the share of searches that produce no outbound click rises.

Then there is AI Mode. Semrush measured AI Mode sessions in September 2025 and found a 93% zero-click rate. Standard search with AI Overviews runs at 43%, and standard search without AIOs runs at 34%. AI Mode is the most click-suppressive search surface Google has ever launched.

The public scale figures for AI Mode do not reconcile. Google reported 100 million monthly active users in the US and India in July 2025 (Nick Fox via TechCrunch). Search Engine Journal reported 75 million daily active users globally for Q3 and Q4 2025. A figure of 75 million daily users implies far more than 100 million monthly users, so the two counts use different scopes or different definitions. Treat both as approximate.

For comparison, AI Overviews reach 2 billion monthly users. The Gemini application reached 750 million monthly users by February 2026. AI Mode is small next to AIOs, it is growing, and its click suppression is close to total.

The CTR curve that SEO practitioners relied on for two decades, a predictable, decaying function of position, no longer describes reality. The technical patterns that once guaranteed click volume need re-evaluation.

GrowthSrc analyzed 200,000+ keywords across ecommerce, SaaS, B2B, and EdTech (approximately 74,000 keywords ranking in the top 10) and measured organic CTR in 2024 versus 2025. Position #1 CTR dropped from 28% to 19%, a 32% year-over-year decline. Position #2 fell even harder, from 20.83% to 12.60% (a 39% decline).

The counterintuitive rise in positions #6-#10 (+30.63%) is significant. Users who scroll past AI Overviews to reach traditional results are more intent-driven. They have already decided the AI summary was insufficient. The scroll itself acts as a commitment filter, concentrating purposeful clicks among lower-ranking results.

seoClarity (12 million keywords, approximately 750 million impressions, 32 million clicks) found that position #3 is the most vulnerable to AIO displacement on both desktop (-2.41 bp) and mobile (-1.31 bp). The likely mechanism: AIOs push position #3 below the fold on most viewports. Mobile AIO prevalence surged 474.9% year-over-year (September 2024 to September 2025), accelerating this displacement. However, Advanced Web Ranking's Q3 2025 data shows signs of quarter-over-quarter stabilization. Branded queries at position #1 lost only 1.52 percentage points, and positions #2-#6 gained a combined 8.71pp. Users appear to be learning to scroll past the AI features.

Ahrefs confirmed the macro trend with a 300,000-keyword study comparing December 2023 to December 2025. Position #1 CTR for AIO-triggering keywords fell from 0.073 to 0.016, a 58% decline. For non-AIO informational keywords, CTR still fell 49%. Seer Interactive (3,119 queries, 42 organizations, 25.1 million organic impressions) measured a 61% organic CTR decline and a 68% paid CTR decline when AIOs were present. Even on queries without AIOs, organic CTR fell 41%, evidence of a broader behavioral shift beyond AIO displacement alone.

One nuance qualifies the trend. Advanced Web Ranking's Q3 2025 data showed a combined +5.15 pp CTR increase for desktop positions #1-#3 quarter over quarter. The rate of decline may be stabilizing after the initial AIO shock. The year-over-year decline still holds, and the worst of the displacement may have passed for the top positions.

The CTR decline has a physical cause. Google restructured the results page at the technical layer so thoroughly that the organic results no longer hold the space they once did.

Only 1.49% of Google's first page results now appear without SERP features of any kind (AccuRanker, 2021-2025). The average SERP displays 6 features above the 10 blue links, and a competitive term shows up to 15. Image packs grew from 44% to 67% of tracked keywords, and PPC ads grew from 5.3% to 28%. Shopping listings grew from 12% to 30%. Sitelinks surged 906% between January and June 2025.

AI Overviews are the most aggressive displacer. Multiple studies converge on the same finding: the average AIO exceeds 1,200 pixels in height while a standard desktop viewport is approximately 900 pixels. The first organic result sits completely below the fold on AIO-present SERPs. On mobile, AI Overviews and Featured Snippets together consume 75.7% of screen space (67.1% on desktop).

Featured snippets, the previous answer-box format, are being systematically replaced. Ahrefs tracked Featured Snippets from 15.41% down to 5.53% of tracked keywords between January and June 2025. That 64% relative decline holds a 0.9 correlation to AIO growth. Google had already removed the FAQ box, from 55% of tracked keywords in 2021 to 3% by 2023, before it launched AIOs. AI Overviews grew 598% and AIO Sitelinks grew 541% in the same six-month period.

The type of SERP matters as much as position. A #1 ranking on a Shopping-dominated SERP captures half the clicks of a #1 on a standard SERP. Position is now a necessary but increasingly insufficient condition for visibility, which is why programmatic architecture that captures granular data-driven queries outperforms editorial volume plays.

The displaced space also has a destination. SE Ranking analyzed 68,313 keywords and 1,321,398 citations in February 2026 and found that Google cites its own properties in 17.42% of AI Mode responses. That share exceeds YouTube, Facebook, Reddit, Amazon, Indeed, and Zillow combined. Self-citation tripled from 5.7% in June 2025. AI Overviews link back to Google properties in 43% of responses. Google says its AI features send traffic to the open web. The citation data points the other way, and it is one more case where Google's public statements diverge from the behavior of its systems.

The zero-click data, taken alone, looks catastrophic. The clicks that still happen in AI search are worth far more than a traditional organic click.

Superprompt analyzed 12.3 million visits across 347 businesses between January and September 2025. Superprompt measured conversion rates by traffic source. The results are stark.

ChatGPT-referred visitors convert at 14.2% compared to 2.8% for traditional organic , a 5x multiple. Claude visitors convert at 16.8%, nearly 6x. Even Perplexity visitors, at 12.4%, convert at more than 4x the organic rate.

Practitioners call the mechanism the "educated click" pattern. An AI visitor arrives pre-briefed. The answer already told them what the product does, how it compares to the alternatives, and why it might suit them. The AI system filters the intent before the click happens. That visitor engages more deeply, converts faster, and bounces less. The click is a commitment signal, because the user has decided that the answer merits a closer look.

Corroborating this, Seer Interactive found that brands cited in AI Overviews receive 35% more organic clicks and 91% more paid clicks than non-cited pages. AIO citation drives clicks rather than replacing them, for the pages that earn citation.

The largest measurement study in the field makes the same argument from the opposite direction. Conductor covered 13,770 domains, 21.9 million Google searches, 17 million AI responses, more than 100 million AI citations, and 3.3 billion sessions across 10 industries. AI referrals average 1.08% of total website traffic. ChatGPT holds 87.4% of that referral traffic. A share of 1.08% is easy to dismiss in an aggregate traffic report. Combined with a conversion premium of 5x or more, 1.08% of the traffic can carry 5-12% of the conversions. A report that shows AI as a percentage of total traffic understates what the channel does.

Two other studies measure the same premium by different methods. Semrush recorded conversion rates of 15.9% for ChatGPT, 10.5% for Perplexity, 5% for Claude, and 3% for Gemini. Semrush puts the average AI visitor at 4.4 times the value of an organic visitor. Microsoft Clarity analyzed more than 1,200 publisher and news sites over 8 months. Clarity measured 1.66% sign-up conversion from AI referrals against 0.15% from search, a factor of 11.

The three studies disagree on the size of the effect. The multiple runs from 4.4x to 11x, and each study defines a conversion differently. Superprompt uses standardized goal values. Semrush uses platform-reported conversions. Clarity counts sign-ups and subscriptions on publisher sites only.

The studies also disagree on Claude. Superprompt puts Claude at 16.8%, and Semrush puts it at 5%.

Superprompt publishes little about how it defined a conversion across 347 different businesses. The direction still holds across independent sources. An AI-referred click converts at a multiple of an organic click. The Seer Interactive AIO citation data corroborates the mechanism from another methodology entirely. Read the direction as reliable, and read the specific multiple as an estimate.

There is also a demand-side explanation. Profound's study of tens of millions of real ChatGPT interactions found transactional intent surging from 0.6% to 6.1% , a 9x increase relative to traditional search. Users approaching AI platforms are already skewing toward purchase-adjacent behavior at dramatically higher rates. On top of this, 37.5% of ChatGPT interactions fall into a wholly new category, generative intent, where the user wants output, not a link. The conversion paradox is both supply-side (AI filtering produces educated clicks) and demand-side (AI users are coming with higher purchase intent to begin with).

For as long as SEO has existed, impressions and clicks moved in the same direction. More visibility meant more traffic. That correlation has broken.

Ahrefs documented the break on their own blog. The impressions-to-clicks correlation flipped from +0.425 (late 2024) to -0.352 (first half 2025). A page can rank well and log impressions from both the regular SERP and the AIO box, and still lose clicks. The AIO answers the query first.

The diagnostic pattern is identifiable in any Google Search Console account: stable or rising impressions alongside declining clicks. If you see this in client data, AIOs are the likely cause, a pattern that surfaces early in systematic layer-based diagnostic triage. The signature often appears in GSC before any external tool confirms an AIO on the relevant queries.

An AIO does not drive every case of rising impressions with falling clicks. Intent drift, where Google's classification of a query shifts from one format to another during a core update, produces the identical GSC signature. A documented SaaS case found intent drift on 11 of 40 top-ranking pages over 24 months. After a rebuild to the 2024 intent format, 7 of the 11 recovered within 60 days. Lily Ray's analysis of 130 websites hit by the September 2023 HCU confirmed the pattern. Stable impressions with falling clicks signal an intent mismatch as often as an AIO displacement.

The distinction sets the remedy. AIO displacement needs AI citation optimization. Intent drift needs a rebuild of the content format.

AIO citation instability compounds the problem. Authoritas via Press Gazette found that approximately 70% of pages cited in AIOs change over a 2-3 month period, and those changes tracked no movement in the traditional organic rankings. Authoritas submitted the research to the UK Competition and Markets Authority.

The click loss from the same research needs a careful reading. Press Gazette reports a 47.5% drop in the desktop clickthrough rate for the top organic link when an AIO appears, and a 37.7% drop on mobile. Secondary coverage of the same research quotes figures up to 79%. The Authoritas report that holds the higher numbers is not publicly available. Cite 47.5% as the measured figure, and treat 47.5% to 79% as the range.

Two studies now quantify the divergence between AI citation and organic ranking. Surfer SEO (February 2026) found that 67.82% of sources cited in AI Overviews do not rank in Google's top 10 organic results for the same query. This corroborates Shashko's finding (74.7% of cited URLs not in organic top 10, across six platforms and 42,971 citations) from different methodology and sample. The ranking factors that determine AI citation are diverging from traditional organic signals. Optimizing for one does not guarantee the other.

The trajectory is accelerating. Ahrefs data shows AIO-organic top-10 overlap declined from ~76% (July 2025) to 38% (February 2026), attributed to Google's query fan-out process and the Gemini 3 upgrade. Google's own AI features also diverge from each other. Ahrefs compared 730,000 response pairs and found 86% semantic similarity between AI Mode and AI Overviews, with only 13.7% citation overlap. The two products agree on what to say, and they cite different sources to say it. Total AIO citations dropped 46% from November 2025 to January 2026 (Tinuiti/Profound Q1 2026), though some of the decline may reflect tracking tool trigger-rate changes.

A methodological note on the Surfer SEO finding above: AirOps (March 2026, 15,000 queries, 82,108 citations) found that 55.8% of ChatGPT-cited pages do rank in Google's top 20 for at least one query including fan-out sub-queries. The apparent contradiction with Surfer's 67.82% figure resolves on inspection. The two studies use different denominators. AirOps asks how many cited pages rank, and Surfer asks what share of all citations comes from outside the top 10. They also use a different query scope (AirOps includes the fan-out queries) and a different rank depth (top 20 against top 10). Shared quality signals confound the relationship between organic ranking and AI citation, and no AI system pulls a citation straight from the SERP.

Wikipedia provides a controlled case study. Researchers (arXiv:2602.18455) studied 161,382 matched article-language pairs and found that AIO exposure reduces daily traffic to English Wikipedia by approximately 15%. The Wikimedia Foundation separately reported an 8% year-over-year decline in human page views. Wikipedia is the single most-cited source across AI platforms. If even Wikipedia, which dominates AI citations, loses 15% of its traffic from AIO exposure, the displacement effect on less-cited sites is likely larger.

Not everything is losing. Three categories are structurally advantaged in the current transition.

1. Forums and Reddit. GSQI analyzed 97 forums and found visibility increases above 100% on 88% of them, and above 500% on many. Reddit rose to the #2 most visible site in Google US search results by early 2025. Reddit appeared in 97.5% of Google product review queries (February 2024). Google renamed its "Perspectives" tab to "Forums" in March 2024, explicitly acknowledging that users trust peer discussion over brand-published content for experience-based queries. User-generated content now displaces traditional editorial and commercial content directly.

2. Breaking news. Define Media Group (64-site portfolio, GSC data at scale) measured a +103% increase in breaking news traffic across all Google surfaces between November 2024 and February 2026. Over the same period, evergreen content declined 40%, and web search overall declined 43% against the pre-AIO baselines. The split is why a data-driven programmatic architecture like Banksparency targets the query types that AI cannot fully resolve. The cause is structural, because the Top Stories carousel stays AIO-protected for a time-sensitive query. LLMs cannot retrieve or generate summaries at breaking news speed. Frontier models still risk hallucinations on critical, fast-moving information. News queries trigger AIOs at only 15% compared to 45%+ for health and science content.

3. Google Discover. Define Media Group also found that Discover now sends equal traffic to web search for their publisher portfolio, the first recorded instance of Discover/search parity. Discover's growth rate is +30% while web search declined 43%. Nearly all the breaking news growth is Discover-driven, doubling after the December 2025 Core Update. For publishers, Discover is partially compensating for organic search losses. For non-publishers, Discover remains largely inaccessible.

4. Review and directory platforms. Active profiles on Trustpilot, G2, and Capterra correlate with a 3x higher ChatGPT citation chance (ConvertMate, directional). The mechanism: AI systems triangulate claims across independent sources, and review platforms provide structured, third-party corroboration that brand-owned pages cannot.

The concentration dynamics are extreme. Goodie AI (September 2025) analyzed 5.7 million citations across ChatGPT, Gemini, Claude, and Perplexity and found that the top 50 domains capture approximately 53% of all citations. Of the most-cited domains, 74% are susceptible to marketing influence: reviews, directories, community platforms, and PR-driven sites. The AI citation economy is not a level playing field. It is dominated by a small number of platform types that function as citation hubs.

Organic search is losing clicks to both AI answers and paid results at an accelerating rate.

Aleyda Solis (Similarweb data, top 5,000 queries across four verticals, January 2025 versus January 2026) found that combined paid results (text ads plus PLAs) now capture 36% of clicks, up from 16%, more than doubled in one year. Text ads alone went from 3% to 16% click share (+13 percentage points). Organic click share declined 11-23 percentage points by vertical. Total click volume also declined (headphones: 2.5 million to 1.8 million clicks).

Ads are also infiltrating AI Overviews directly. Semrush tracked ads appearing on less than 1% of AIO SERPs in March 2025 to 25.56% by November 2025, a 394% increase in eight months. Over 80% of AIO ads appear below the AI box, and 18-20% appear above it. A separate Semrush study (10 million keywords) shows ads alongside AIOs grew from approximately 3% to 40% between January and November 2025.

The monetization squeeze extends to local search, previously considered structurally resilient because the local pack's conversion model (call, directions, visit) bypasses website clicks. Sterling Sky (Joy Hawkins, 1,200 mobile ranking reports via Places Scout) tracked sponsored placements inside the local 3-pack. They rose from about 1% to 22% of tracked mobile queries between October 2025 and January 2026, a 733% increase in three months. A paid ad in the local pack strips the direct call button and the website button from the organic listings. Those two buttons carried the conversion.

Local Services Ads above the pack also grew from 11% to 31% in the same period. A local business now faces paid displacement from above the pack and from inside it. A #1 organic local pack position no longer carries its historic conversion advantage.

The zero-click paradox forces a measurement and diagnostic rethink. Traffic volume no longer correlates with business value, and it sometimes correlates inversely. The KPIs that dominated SEO reporting for two decades need a revision.

Impressions are no longer a proxy for opportunity. The Great Decoupling (impressions +0.425 correlation to clicks flipping to -0.352) means a page can set impression records while it delivers less value. Read the GSC impression data against click-through and conversion data, or it means nothing.

Most AI traffic arrives under the wrong channel. GA4 has no AI channel group. AI visits divide across Direct, Referral, and Organic by platform. Free-tier ChatGPT often strips the referrer. The ChatGPT and Claude mobile applications strip it completely. Gemini clicks frequently land as direct.

Seer Interactive estimates that 70.6% of AI traffic lands as Direct in GA4, and that true AI referral volume is 2 to 3 times what GA4 reports. The AI-referred conversion value already sits in your data under the wrong channel.

The fix is a custom channel group in GA4. Match the source against the AI platform domains, and place the new channel above Referral in the rule order. GA4 evaluates the rules from the top down. A channel below Referral never receives the traffic, because the Referral rule matches first. Custom channel groups also apply to the historical data, so the correction is retroactive.

Google AI Mode is the exception that most reporting gets wrong. AI Mode links carried noreferrer from May 20 to May 28, 2025. Maga Sikora identified the fault, Aleyda Solis corroborated it, and John Mueller confirmed it was an unintentional bug. Google fixed it in about 8 days.

AI Mode traffic now passes a referrer, and GA4 classifies it as organic. The bug skewed only the data inside that 8-day window. Practitioners still repeat the claim that AI Mode is permanently invisible. The claim is wrong.

AIO citation is volatile, not a position to hold. With 70% of cited pages turning over in 2-3 months, AIO presence requires ongoing content engineering for AI extraction and monitoring. A citation is not a stable ranking position, and a reporting framework that treats it as a fixed asset misrepresents what happens. Freshness weighting varies by platform: ConvertMate (via Surfer SEO, February 2026, methodology undisclosed) estimates that freshness accounts for approximately 40% of Perplexity's ranking factors, with content labeled "updated two hours ago" cited 38% more often than month-old content. The estimate is directional, but it aligns with the measured AIO citation churn and points toward a structural freshness advantage for maintained content.

Traditional SEO metrics have near-zero predictive power for AI citation. Profound (250M+ AI responses, 3B+ citations tracked across 8 answer engines) found that traffic explains 5% of citation behavior (r²=0.05) and backlinks explain 3.8% (r²=0.038). The metrics practitioners track for organic ranking, the ones that populate every SEO dashboard, are functionally irrelevant to whether an AI platform cites a page.

Revenue per visit should replace total traffic as the primary organic KPI. The conversion paradox data (5x conversion rate for AI-referred traffic) means a declining traffic trendline can mask improving business outcomes. If leadership is panicking over traffic drops without examining per-visit conversion and revenue, the strategic response will be wrong. The likely error is too much investment in volume recovery, and too little in the signals that earn AI citations, including allowing LLM crawlers access to content.

Brand investment is an organic search strategy. A brand carries 45.7% of search volume (Ahrefs, 150 million keywords), and navigational queries collapse from 32% to 2% in an AI chat context (Profound). Brand recognition is the only reliable bypass of AI intermediation. Users who know your name search for you directly. Those queries are not subject to zero-click because the user's intent is explicitly navigational.