Google AI Mode statistics for 2026 show how quickly conversational search has moved into Google’s mainstream product. In May, Google reported more than 1 billion monthly active AI Mode users worldwide and said queries had more than doubled every quarter since launch. Reach is only part of the story, however: those figures say little about what people do once they begin a session.
The behavioral figures reveal a different search pattern. Initial queries are longer than conventional searches, people are asking more follow-up questions, and images or voice increasingly become part of the input. Planning and product comparisons also stand out as fast-growing activities, although growth rates should not be confused with the share of all searches.
Independent research describes a less comfortable trade-off. In a preregistered U.S. field experiment, assigning people to AI Mode reduced their rate of visiting external websites by 18.8 percentage points relative to current Google Search. Separate SE Ranking datasets examine how often AI Mode cites Google-owned destinations and displays paid text ads. The studies use different samples and dates, so their results need to stay separate.
These figures should not be mixed with AI Overviews statistics. AI Mode is a distinct conversational search experience, and this article focuses on studies measuring that product rather than AI-generated summaries on conventional search results pages.
Google AI Mode Statistics: Key Numbers for 2026
| Metric | 2026 figure | What it measures |
|---|---|---|
| Monthly active users | 1B+ | Global AI Mode users reported by Google |
| Query growth | More than 2× every quarter | Growth in AI Mode queries since launch |
| Average query length | About 3× traditional Search | Relative length of an AI Mode query |
| Follow-up query growth | More than 40% average per month | U.S. follow-up queries inside AI Mode |
| Multimodal searches, U.S. | More than 1 in 6 | Voice or image inputs, May 2026 Google report |
| Multimodal searches, global | 1 in 5 | Voice, image, or document inputs, October 2026 Alphabet Q3 remarks |
| Image-input growth | More than 40% month over month | U.S. image searches since launch |
| Planning-query growth | 80% faster than AI Mode overall | Relative growth over the previous six months |
| “Which” query growth | 40% faster than AI Mode overall | Relative growth over the previous six months |
| Brainstorming-query growth | 30% faster than AI Mode overall | Relative growth since launch |
| Image-creation query growth | More than 3× | Growth since the start of 2026 |
| External website click-through effect | −18.8 percentage points | March 2026 U.S. experiment: forced AI Mode vs. current Search |
| Google.com share of citations | 17.42% | 1.32M AI Mode citations studied |
| Commercial queries with text ads | 29.45% | June 30 U.S. snapshot of 50,032 selected commercial keywords |
| Advertiser domain also cited | 11.53% | Domain-level overlap between paid placements and AI Mode citations |
Most behavioral figures come from Google’s May 2026 AI Mode U.S. Insights. The 1B+ monthly-active-user figure is global, and the newer global multimodal statistic comes from Alphabet’s October Q3 remarks. The external-website and advertising figures come from independent research rather than Google’s usage report.
Google’s keyword- and query-specific growth rates use Trends data covering May 2025 through April 2026. They describe how particular categories changed relative to AI Mode activity overall, not how much of total search demand those categories represent. The table intentionally retains dates and geographic labels where the datasets differ.
Google AI Mode Users Have Passed 1 Billion
AI Mode passed 1 billion monthly active users globally within roughly a year of its U.S. launch. The reported query volume had also more than doubled each quarter since launch, giving the conversational interface a substantial audience by May 2026. These are Google’s product-level statements, not an independently measured share of all searches.
The missing denominator matters. Google has not published the proportion of all Google searches that take place in AI Mode, and its billion-user milestone cannot fill that gap. Someone who opens AI Mode once in a month and someone who uses it daily can both count as monthly active users.
Even with that limitation, AI Mode is no longer a niche interface. The product’s scale makes its evolving behavior relevant to publishers, advertisers, and retailers, but absolute adoption and usage share remain different questions.
Google AI Mode Queries Are Much Longer Than Traditional Searches
The average AI Mode query is about three times as long as a traditional Google Search query. Instead of compressing a need into a handful of keywords, someone can specify the outcome they want, relevant constraints, and the format of the answer in one request. A longer prompt is not necessarily a harder question, but it often supplies more context at the outset.
A person searching for a product, for example, can include price, location, size, preferences, and intended use in the first prompt. A travel query can contain dates, budget, transport preferences, and a request for an itinerary without splitting those details into several separate searches.
Google’s language analysis offers a second clue. Frequent opening words include “what,” “how,” “I,” “is,” and “can,” while common terms include “find,” “information,” “identify,” “explain,” and “summarize.” Many prompts read more like instructions for a task than a conventional keyword string.
Follow-Up Queries Are Growing More Than 40% per Month
The first query is only part of the session. Google says follow-up queries in U.S. AI Mode increased by more than 40% on average per month, showing that users increasingly continue the same search instead of starting again from a blank box.
The interaction can resemble progressive research rather than a sequence of separate searches. A person might start with a broad comparison, add a price limit, rule out an unsuitable option, ask about local availability, and request a final recommendation. Each follow-up refines an earlier answer, sometimes without a return to a conventional results page.
Google has not disclosed what percentage of all AI Mode sessions contain follow-ups. The published 40% number refers to the growth of follow-up activity, not its overall share.
More Than 1 in 6 AI Mode Searches Are Multimodal
Google’s May report said more than one in six U.S. AI Mode searches used voice or images. In Alphabet’s October 2026 Q3 remarks, the company put the global figure at one in five searches using voice, images, or documents. Both are useful observations, but they have different geography, dates, and input definitions; the difference is not a clean growth rate.
In its U.S. analysis, Google reported image searches growing by more than 40% month over month since launch. People can photograph an object, upload a screenshot, or show a product, then ask for identification, alternatives, compatibility details, or troubleshooting. Document-based questions also matter in the newer global statistic, though Google has not published a separate global percentage for document inputs.
This creates search behavior that conventional keyword datasets do not fully describe. A visual request may contain little text, while the user’s actual requirements emerge only after the system interprets the image and receives a follow-up.
Visual creation is growing too. Google says image-creation queries in AI Mode more than tripled from the start of 2026 to the period covered by its report. That makes visual interaction a two-way activity: people ask Search to analyze what they see and increasingly use the same interface to generate something new.
Google AI Mode Usage Is Shifting Toward Planning and Decisions
Some of the fastest-growing AI Mode categories involve organizing information rather than retrieving a single fact. In Google’s Trends comparison, planning-related queries grew 80% faster than AI Mode overall during the previous six months. That is a relative growth comparison, not an 80% share of AI Mode activity.
These searches include schedules, restaurant planning, travel itineraries, budgeting, exercise plans, to-do lists, and similar tasks where the user expects Search to organize several pieces of information into something usable. AI Mode is particularly suited to that format because the plan can be revised through additional prompts rather than rebuilt from the beginning.
What matters for publishers is the form of the task. A travel itinerary or purchase shortlist may synthesize information that previously required visits to several specialist pages. The statistic does not establish whether those users would otherwise have followed external links, so it cannot be converted directly into a traffic forecast.
“Which” Searches Are Growing as Users Narrow Their Options
Queries beginning with “which” grew 40% faster than AI Mode queries overall over six months. Google highlighted patterns such as “which of” and “which one,” language that fits a decision stage where the user is weighing alternatives instead of collecting a list of possibilities.
That matters for commercial searches because product and service evaluation may now happen deeper inside Google. A shopper can ask which laptop fits a certain workload, which hotel is better for a particular trip, or which service provider matches a set of requirements before visiting any external site.
Brainstorming queries showed a separate 30%-faster growth rate relative to AI Mode overall since launch. Together the two patterns point to AI Mode serving both ends of a decision: generating options and narrowing them. Their different comparison windows mean the percentages should not be ranked as if they measured identical periods.
Shopping Research Often Continues Inside AI Mode
Google’s research shows a common shopping pattern in which users begin with traditional Search and then enter AI Mode to investigate further. Electronics, books and media, apparel, health and beauty, automotive, home and garden, groceries, home improvement, toys and games, and sports and outdoor products were among the main categories generating shopping follow-ups.
As the shopper narrows the purchase, the questions become increasingly practical. Rather than asking only for product names, users look for details that determine whether a specific choice will work. Common follow-up topics include:
- price and product availability;
- brand, size, material, style, and quality;
- nearby stores carrying the product;
- replacement parts and compatibility;
- financing options;
- whether an item is available to buy online.
For ecommerce sites, the important shift is where comparison happens. Users may narrow their options inside an AI Mode conversation before deciding whether to visit a retailer. That does not mean every AI-assisted session ends with a purchase; the distinction between adoption, referrals, and transactions is developed further in our AI shopping statistics for 2026.
AI Mode Reduced External Website Click-Through by 18.8 Percentage Points
Google’s usage figures cannot tell us what happens to external websites. A preregistered field experiment conducted in March 2026 examined how search behavior changed when people were assigned to an AI Mode experience. Its main click-through outcome covered external sites, not publishers alone.
Researchers from the University of Pennsylvania and Northeastern University recruited U.S. adults who primarily used Google Search and Chrome. After a three-day baseline, participants were randomly assigned to current Google Search, a “No AI” condition, or a condition that redirected searches into AI Mode. The behavioral analysis included 1,100 participants who searched at least once during the seven-day treatment period.
The preregistered AI Mode field experiment found that assignment to AI Mode reduced click-through to external websites by 18.8 percentage points compared with current Search. Its reported 95% confidence interval ranged from −22.2 to −15.3 percentage points. The result applies to the study’s intervention and participants, not to every AI Mode session in 2026.
This is an 18.8-point difference, not an 18.8% relative decline. The distinction matters because percentage points describe a change between rates, while a relative percentage requires a known starting rate. Nor does the experiment provide a universal organic-search CTR benchmark for publishers.
Random assignment makes this stronger evidence of a treatment effect than a comparison between voluntary AI Mode users and nonusers. But the treatment was unusually intense: AI Mode accounted for only 0.6% of participants’ baseline searches, whereas 94.7% of searches in the intervention group were successfully redirected to it. The experiment also preceded the broader 2026 expansion of ads in AI Mode.
That makes the result informative about enforced exposure, not a direct estimate of how voluntary adoption will affect every website. Separate AI Overview CTR research addresses a different search interface and should not be blended with this AI Mode treatment effect. The shared warning is methodological: the interface, population, and measurement period determine what a statistic can mean.
AI Mode Citations Often Keep Users Inside Google
SE Ranking examined 68,313 keywords across 20 niches and collected 1,321,398 AI Mode citations on February 12, 2026. Google.com was the most frequently cited domain, accounting for 17.42% of all citations in the dataset.
The SE Ranking AI Mode citation study also looked at where those Google citations led. About 59% pointed to organic Google search results displayed in the AI Mode side panel, while 36.1% went to Google Business Profiles. The remaining share was spread across other Google properties.
A citation inside an AI Mode response is therefore not necessarily a direct link to an external publisher. If it points to a Google results page or Business Profile, the user can continue navigating Google’s own surfaces. The study measured citation destinations, not the probability that users actually clicked them.
The pattern was broad but not uniform. Google.com ranked first among cited domains in 19 of the 20 categories SE Ranking analyzed, yet its share differed substantially between niches. A single overall citation percentage is therefore a poor benchmark for individual industries.
For site owners, a single domain-level citation rate hides several different outcomes. A citation may lead to a source website, a Google result, or a local profile; those destinations create different opportunities. Citation frequency and destination type should therefore be examined separately from any later site visit.
AI Mode Ads Appeared on 29.45% of Commercial Queries
Paid placements are an established part of AI Mode, but the available prevalence figures are snapshots. SE Ranking analyzed 50,032 selected commercial keywords across 20 U.S. niches on June 30, 2026. Text ads appeared for 14,733 queries, or 29.45% of that sample, which was specifically assembled around keywords capable of triggering website text ads.
The SE Ranking study of ads in AI Mode found two ad items in 71.1% of responses where an ad appeared. In those ad-bearing results, an advertiser’s domain also appeared in the AI Mode citation list for the same query in 11.53% of cases. That is domain-level overlap with AI citations, not a measure of ordinary organic rankings.
Paid visibility and cited-source visibility are therefore different layers. A domain can appear in a sponsored placement without appearing in the AI answer’s source list, while a cited website may have no paid placement. The 11.53% overlap does not establish that buying an ad improves the chance of a citation.
The 29.45% figure should not be described as the current rate for all AI Mode searches. It excludes broader informational activity, covers only selected U.S. niches, and predates later product changes. In October 2026, Google announced AI Mode ads in 10 additional countries and the launch of multi-turn ads in the U.S., which can surface within follow-up questions. Google also said it planned more markets during Q4; those plans should not be treated as completed launches.
For advertisers, a paid placement is not the same outcome as a cited source, and neither establishes commercial value. A visible ad may lead to an inquiry without generating a qualified opportunity; the distinction between cost per lead and cost per qualified lead remains important as these ad formats expand. Comparing the June ad dataset against a later rollout also requires care because the underlying product and geographic coverage changed.
The Search Session Is Becoming More Important Than the Single Query
The most consistent behavioral pattern is that more of the task unfolds inside one conversational search experience. Users write detailed prompts, add follow-up questions, bring images or documents into the interaction, and use AI Mode for planning and comparing alternatives. That helps explain why a single keyword or a single page view says less about the full decision process than it once did.
The March experiment provides another useful distinction. Assignment to AI Mode increased time per search session by approximately 0.43 minutes while reducing search sessions per day by approximately 0.92, relative to current Search. Longer sessions and fewer sessions are not contradictory: a more involved interaction can contain more work without creating more separate search occasions. These are experimental treatment effects, not measured population-wide trends.
The resulting visibility landscape has several layers. A company might rank in conventional Search, appear in an AI Mode citation, be represented through a Google-owned result, or show an ad. Even when two of those placements relate to the same question, the underlying mechanisms and likely downstream actions differ; a brand mention alone does not tell us whether the user reached the brand’s site.
Large measurement gaps remain. Google has not disclosed AI Mode’s share of all Google Search queries or absolute volumes for the Trends categories in its May report. There is also no universal external-site click-through benchmark: the field experiment tested a particular intervention, while the SE Ranking studies sampled different queries on different dates. The May behavior figures are mostly U.S.-based, whereas the monthly-user and October multimodal figures are global.
Traditional Google Search has not disappeared, and none of these datasets shows that it has. Instead, the evidence describes a larger conversational product where research, comparison, and advertising increasingly coexist. The clearest takeaway for publishers and marketers is to keep adoption, query behavior, citations, external visits, and paid visibility as separate measures rather than forcing them into one headline statistic.

