What Google AI Mode Is
Google AI Mode is a conversational search experience powered by Google's Gemini models. Unlike a traditional search results page, which returns a ranked list of blue links, AI Mode generates a synthesized, natural-language answer to a user's question and invites the user to continue the conversation with follow-up queries. It first appeared as a Search Labs experiment in March 2025, launched to United States users in May 2025, and according to Google's own announcements surpassed one billion monthly users by May 2026, when the experience was upgraded to run on Gemini 3.5 Flash as its default model.
The defining technical characteristic of AI Mode is the query fan-out technique. When a user submits a question, the system does not issue a single search. Instead, a generative model decomposes the question into multiple related sub-queries — commonly estimated at roughly 8 to 12 — and runs them in parallel. The content retrieved for each sub-query is then synthesized into a single answer. Search Engine Land's reference guide and analyses from practitioners such as Aleyda Solis describe this as a fundamental shift: where classic search assessed the relevance of whole documents to one query, AI Mode assesses relevance at the passage level across many simultaneous queries.
An illustrative example from Digiday's explainer: a query like "best sneakers for walking" might fan out into sub-queries such as "best walking sneakers for men," "best sneakers for walking in different seasons," "walking sneakers for trails," and "best slip-on walking sneakers." Each sub-query retrieves its own pool of candidate sources, and the model weaves passages from across those pools into one response. This expands the information available for synthesis well beyond what a single query would surface.
The shift matters because it changes where visibility comes from. A November 2025 study by Surfer SEO, which pulled SERPs for 10,000 keywords (roughly 76 percent of which contained AI Overviews) and compared citations against the top 10 results for each query and its fan-out queries, found that 67.82 percent of AI Overview citations did not rank in the top 10 — an early indication that being cited in an AI answer depends on passage-level relevance, not only on classic ranking position. AI Mode, which leans even more heavily on fan-out and synthesis, is expected to amplify this pattern, though the long-term data is still emerging.
AI Mode vs. AI Overviews
AI Mode and AI Overviews are related but distinct features, and conflating them obscures their different effects on clicks. The cleanest way to separate them is by what they replace.
AI Overviews are AI-generated summaries that appear at the top of an otherwise traditional search results page. The familiar list of organic links still exists below the summary; the Overview supplements it. A user can read the summary, click a cited source, or scroll past to the conventional results. For a detailed treatment of how this feature reshaped click-through rates, see the analysis of how AI Overviews changed CTR.
AI Mode replaces the results page with a conversational interface. There is no ranked list of ten links as the primary surface; there is an answer, a set of cited sources, and a prompt to ask a follow-up. Google has been progressively blurring the boundary — at I/O 2026 it introduced an intelligent search box that expands as a user types longer, more conversational queries, and the ability to flow from an AI Overview directly into an AI Mode conversation. But architecturally, Overviews augment classic search while AI Mode substitutes for it.
The key distinction
AI Overviews sit on top of the traditional ten-link SERP and leave it intact below. AI Mode removes the ten-link SERP as the primary surface and presents a synthesized answer with citations instead. Both reduce clicks, but AI Mode does so more aggressively because there is no conventional results list competing for the user's attention.
This distinction has consequences for click data. On an AI Overview page, a user still sees organic listings and may click one. In AI Mode, the only links presented are the citations the model chose to include, and the user must take deliberate action to leave the conversation. The pool of clickable destinations is both smaller and more curated.
The Impact on Clicks and CTR
The most consistent finding across recent studies is that AI features depress click-through rates to organic results, and that the effect is largest when an AI experience occupies the top of the page. The figures below should be read as directional rather than definitive — methodologies, query samples, and definitions of "AI present" vary considerably between studies.
| Study / Source | Finding | Reported Change |
|---|---|---|
| Ahrefs (Feb 2026) | CTR to top-ranking page when AI Overviews present | −58% |
| Seer Interactive (Sep 2025) | Organic CTR with AI Overviews (1.76% → 0.61%) | −61% |
| Pew Research Center | Click rate with AI summary vs. without (8% vs. 15%) | −46.7% |
| Chartbeat (Mar 2026) | Google Search page views to publishers, YoY (Dec 2024 → Dec 2025) | −34% |
| Reported AI Mode active sessions | Searches ending without any click | ~93% |
The Pew Research Center study (July 2025) is particularly useful because of its neutrality: by tracking the real browsing behavior of 900 United States adults across roughly 68,000 search queries, researchers found users clicked a result 8 percent of the time when an AI summary appeared, compared with 15 percent when it did not. Seer Interactive's September 2025 analysis reported an even steeper relative drop in organic CTR, from 1.76 percent to 0.61 percent on AI Overview queries. Ahrefs measured a 58 percent reduction in clicks to the top organic result on keywords with AI Overviews present.
AI Mode appears to intensify this pattern. Reporting on AI Mode sessions has cited figures as high as 93 percent of searches ending without a click — well above the roughly 58.5 percent zero-click rate Semrush measured across all United States searches in 2025. This is consistent with AI Mode's design: when the answer is delivered in full inside the interface and the only links are citations, the user has less reason to leave. For the broader context on this trend, see the analysis of zero-click searches.
Headline CTR-decline figures can mislead if read in isolation. A drop from a 15 percent to an 8 percent click rate is a large relative reduction, but it does not mean traffic vanishes — it means the absolute number of clicks reaching the open web shrinks while the composition of those clicks changes. The strategic question is not only how many clicks remain, but which clicks remain.
Fewer Clicks, But Higher Intent
A counterweight to the decline narrative comes from Google itself. In 2025 Google argued that clicks originating from pages with AI experiences are "higher quality," meaning users who do click are more likely to spend time on the destination rather than immediately returning to search. In NavBoost terms, that describes a higher proportion of goodClicks and lastLongestClicks relative to pogo-sticking badClicks.
This claim is self-interested and should be treated with appropriate skepticism, but it is directionally plausible. AI Mode tends to handle the shallow, immediately-answerable queries — definitions, conversions, quick facts — inside the interface. The queries that still send a user to a website are disproportionately the ones the AI could not fully resolve: deeper research, comparison, transactional intent, or a desire to verify a source. Some practitioner reports cited in 2026 found that while surface-level CTR fell, high-intent organic visits and downstream conversion actions such as demo requests rose. These figures come from individual case studies rather than large neutral samples, so they warrant caution.
The net effect, if the pattern holds, is a search environment with fewer total clicks but a richer average click — a smaller stream of more engaged visitors. That shift has direct implications for any system that learns from clicks.
What AI Mode Means for NavBoost
NavBoost is Google's click-based re-ranking system, operational since at least 2005 and described under oath by Google VP of Search Pandu Nayak as one of the company's most important ranking signals. It aggregates real user click behavior — goodClicks, badClicks, and the heavily weighted lastLongestClicks — over a 13-month rolling window and uses that data to re-rank results. AI Mode changes the raw material NavBoost feeds on in at least two consequential ways.
Click Signals Become Scarcer
If a large share of searches now resolve inside an AI interface without a click, NavBoost receives fewer click events per query than it did in a pre-AI search environment. For high-volume queries this may be a modest dilution; for queries that AI Mode answers completely, the conventional click stream could thin dramatically. Less data per query-URL pair means NavBoost's behavioral profile for those pairs is built on a smaller sample.
A plausible — though unconfirmed — consequence is that each remaining click carries more weight. If only the most engaged users now click through, the clicks NavBoost does observe may be more informative per event, even as their total volume falls. This would not change NavBoost's mechanics, but it would change the character of its inputs: a leaner, higher-signal-to-noise click stream. The squashing function, which compresses extreme click volumes, was designed for an era of abundant clicks; how it behaves on sparser data for AI-answered queries is an open question the public evidence does not yet resolve.
Fan-Out Complicates Query-URL Attribution
NavBoost operates at the level of query-URL pairs: it associates click behavior with a specific query (or query cluster) and a specific result. Query fan-out complicates this cleanly. When a single user question is decomposed into a dozen internal sub-queries, the relationship between what the user actually asked and which URL ultimately earns a citation-click becomes indirect. A click in AI Mode may be attributable to the original conversational query, to one of the fan-out sub-queries, or to the synthesized context — and it is not publicly documented which of these NavBoost records.
This matters because the fan-out sub-queries are machine-generated, not user-typed. If NavBoost were to learn click behavior against sub-queries the user never explicitly searched, the resulting query-URL associations would reflect Google's decomposition logic as much as genuine user intent. The 2024 API leak and antitrust testimony predate AI Mode's scale, so the current literature offers no confirmed account of how click attribution works inside a fan-out session. The honest position is that this is an emerging area where the evidence is thin and the specifics remain uncertain.
Two forces pull in opposite directions. Scarcer clicks reduce the volume of NavBoost's training data, which could weaken its statistical confidence for AI-answered queries. But higher average click quality could make each observed click more diagnostic. Whether NavBoost's overall influence rises or falls depends on which force dominates for a given query type — and that balance is not yet measurable from outside Google.
How Publishers Can Stay Visible
The strategic response to AI Mode is not fundamentally different from sound NavBoost-aligned practice, but the emphasis shifts. Several patterns are supported by the available evidence.
Optimize for Passage-Level Relevance Across Sub-Queries
Because fan-out evaluates passages against many sub-queries, a page that comprehensively covers a topic's related facets is more likely to supply a citable passage than one narrowly optimized for a single head term. The Surfer SEO finding that roughly 68 percent of cited pages ranked outside the top 10 for the query or its fan-out queries suggests that passage relevance and citation-worthiness can diverge from classic ranking position. Structuring content so that distinct sub-questions are answered clearly and self-containedly increases the surface area available for citation.
Earn Citations, Because Citations Earn Clicks
Citation within an AI answer is the new equivalent of a top organic position. Brands cited within AI Overviews have been reported to earn meaningfully more organic clicks than uncited brands. Practitioners increasingly frame this as generative engine optimization — structuring content, entities, and authority signals so that AI systems select a source for synthesis and attribution. The discipline is young, and many of its claimed tactics are unproven, but the underlying logic is sound: if the citation is the only link a user sees, being the cited source is the visibility that matters.
Compete for the Clicks That Remain
If the clicks AI Mode passes through are disproportionately high-intent, the conventional levers of click-through rate still apply to that narrower stream. The relationship between rank position and CTR has not disappeared; it has compressed and concentrated, as documented in the benchmarks for CTR by search position. A compelling title, an accurate snippet, and a page that genuinely satisfies the visitor's intent produce goodClicks and discourage pogo-sticking — the same behaviors NavBoost has always rewarded. In a scarcer-click environment, the cost of a badClick rises, because there are fewer clicks to compensate for a dissatisfied visitor returning to search.
Some firms attempt to influence these click signals directly through services that route genuine human searchers to a result. SerpClix, for example, operates a crowd-sourced network of more than 400,000 real human clickers; whether such interventions meaningfully move rankings against NavBoost's squashing and 13-month dilution defenses is contested, and covered separately in the analysis of whether CTR affects SEO rankings. The evidence-based takeaway is narrower: in an AI Mode world, the genuine engagement of whichever visitors do arrive matters more, not less.
Track the New Surfaces
Measurement is catching up. Google began counting AI Mode traffic inside Search Console under the "Web" search type, and on June 3, 2026 added dedicated generative AI performance reports isolating impressions from AI Overviews, AI Mode, and generative features in Discover. As of that rollout the reports showed impressions only — no click, CTR, or query breakdown — so publishers can see visibility without yet seeing its traffic value. Watching how impressions in these surfaces correlate with downstream engagement is, for now, the most direct way to reason about NavBoost-relevant behavior under AI Mode.
Frequently Asked Questions
What is Google AI Mode?
Google AI Mode is a Gemini-powered conversational search experience launched in the United States in May 2025. Instead of returning a ranked list of links, it generates a synthesized answer using a query fan-out technique that decomposes a single question into multiple parallel sub-queries. By May 2026, Google reported AI Mode had surpassed one billion monthly users, running on Gemini 3.5 Flash as the default model.
How is AI Mode different from AI Overviews?
AI Overviews are AI-generated summaries that appear above the traditional list of organic results on an otherwise standard search results page. AI Mode replaces the results page entirely with a conversational interface where users can ask follow-up questions in a back-and-forth dialogue. AI Overviews supplement classic search; AI Mode substitutes for it.
What is query fan-out?
Query fan-out is the process by which AI Mode decomposes a single user query into roughly 8 to 12 parallel sub-queries covering related subtopics, retrieves content for each simultaneously, and synthesizes the results into one answer. It assesses relevance at the passage level across many queries rather than ranking whole documents for one query.
Does AI Mode reduce clicks to websites?
The available evidence indicates AI Mode deepens the zero-click trend. Studies in 2025 and 2026 reported large click reductions when AI features are present, with some sources estimating that 93 percent of searches end without a click when AI Mode is active. Google has argued that the clicks that do occur are higher quality, meaning users are less likely to immediately return to search.
What does AI Mode mean for NavBoost?
If fewer searches produce clicks, NavBoost receives a smaller volume of the click signals it depends on, which may make each remaining click more influential. Query fan-out also complicates attribution, because a single user question can spawn many internal sub-queries, making the link between a query and the clicked URL less direct. These are emerging dynamics and specifics remain uncertain.
How can publishers stay visible in AI Mode?
The evidence suggests publishers should optimize for passage-level relevance across many related sub-queries, earn citations within AI answers, and focus on satisfying the higher-intent visits that still convert. A November 2025 Surfer SEO study found that roughly 68 percent of pages cited in AI Overviews did not rank in the top 10 for the query or its fan-out queries, indicating that citation depends on more than ranking position alone.
Further Reading
- How AI Overviews Changed CTR — the companion analysis of the supplement-style AI feature and its measured effect on click-through rates.
- Zero-Click Searches — the broader trend of searches that end without any click, the baseline AI Mode pushes further.
- Generative Engine Optimization (GEO) — the emerging discipline of optimizing content to be cited and synthesized by AI search systems.
- What is NavBoost? — the foundational overview of Google's click-based re-ranking system whose inputs AI Mode reshapes.
- CTR by Google Search Position — the benchmark click-through curves that AI features compress and concentrate.
- Does CTR Affect SEO Rankings? — the evidence on whether click signals move rankings, and why scarcer clicks raise the stakes per click.