Search changed more between 2024 and 2026 than in the decade before it, and the numbers are genuinely striking: most US Google searches now end without a click, AI answers push that figure past eight in ten, and the visitors who do arrive from AI assistants convert better than the ones from traditional search. Below are the figures worth knowing, each with its source and date, followed by something most statistics roundups leave out: how to read them critically, because several widely repeated numbers are shakier than they look.
Zero-click search: the headline numbers
- Around 68% of US Google searches ended without a click between January and April 2026 (SparkToro, June 2026).
- The share of searches producing at least one click fell 9.51 percentage points between 2024 and 2026, a relative decline of roughly 23% (SparkToro, 2026).
- Zero-click behavior is not new. Roughly 49% of Google searches ended without a click as far back as 2019 (SparkToro, using the now-defunct Jumpshot panel).
- When an AI Overview appears, about 83% of those searches end without a click, against roughly 60% when one does not (Bain and Dynata, December 2024; Similarweb).
- In Google's AI Mode, the zero-click rate reaches roughly 93% (Semrush, September 2025).
- Mobile and desktop behave very differently: approximately 77% of mobile searches end without a click, against 46.5% on desktop (SparkToro and Datos, 2024).
- Intent matters more than any average: around 74% of informational queries are zero-click, against roughly 31% of transactional ones (Semrush intent study).
That last figure is the one business owners should sit with. If you sell something, your queries live in the healthier half of the distribution, which is the point we make at length in our guide to winning search when nobody clicks.
AI search: adoption and reach
- ChatGPT reported around 900 million weekly active users in February 2026, up from roughly 400 million a year earlier (OpenAI, February 2026).
- Google reported at I/O 2026 that AI Mode had passed one billion monthly users, with query volume more than doubling each quarter (Google, 2026).
- Yet AI Mode accounted for only about 0.34% of searches during SparkToro's January to April 2026 study window, meaning its effect on behavior was still small during that period even as adoption grew quickly (SparkToro, 2026).
- Referral share is concentrated but disputed: ChatGPT has been measured at roughly 92% of AI referral traffic (Grips Intelligence via eMarketer), about 87% (Conductor), and as low as 63% in another panel (Goodie, 2026). The direction is consistent; the precision is not.
The traffic that still arrives is performing better
- AI-referred visitors converted about 42% better than non-AI traffic in March 2026, a reversal from converting roughly 38% worse twelve months earlier (Adobe Digital Insights, Q1 2026).
- They also engaged more: around 48% more time on site, 13% more pages per visit, and 37% higher revenue per visit (Adobe Digital Insights, 2026).
- AI-driven traffic to US retail sites grew roughly 393% year over year in the first quarter of 2026 (Adobe Digital Insights via eMarketer).
- Volume remains small. Across most B2B sites, AI referrals still represent on the order of 1% of total sessions, which is why the quality gap matters more than the quantity today.
- Most teams are not measuring it. Only around 14% of marketers track AI search performance, and an estimated 70% of AI-referred traffic may arrive without referrer data attached, landing in direct or other buckets (The Digital Bloom, 2026).
Read those two blocks together and the strategic picture is clear enough: fewer clicks, better clicks, and almost nobody counting them properly.
How to read search statistics without being misled
This section is why we published this page. Search statistics circulate faster than they are checked, and four habits will keep you honest:
- Check what is being measured. Zero-click studies typically count clicks to the open web, meaning a search that ends on a Google-owned property like Maps or YouTube may be counted as a non-click even though the user went somewhere. That single definitional choice moves the headline number several points.
- Check the panel. Clickstream studies infer national behavior from a sample of users on particular devices in particular countries. SparkToro itself cautions that its own historical comparisons are not directly equivalent, because the underlying data providers changed between studies. When the source warns you, believe the source.
- Check the date, then check the drift. A 2024 figure describing a market that changed in 2025 is not evidence about today. Notice how many roundups repeat the same two-year-old numbers under a 2026 headline.
- Prefer ranges to decimals. When three credible datasets put ChatGPT's referral share at 63%, 87%, and 92%, the honest statement is "the large majority, somewhere between roughly two thirds and nine tenths." A single decimal point implies a precision the methods cannot support.
The practical version: use published statistics to understand direction, and use your own Search Console, analytics, and citation checks to understand your situation. Nobody else's average describes your business, which is the whole reason we built the 15-minute AI visibility self-test.
What the numbers add up to
- Traffic-only measurement now understates performance. Impressions, citations, branded search, and direct visits carry the signal that sessions used to.
- Transactional and local intent remains comparatively protected. The zero-click wave hit informational lookups hardest.
- Being named in the answer is now a distinct objective, not a byproduct of ranking, which is the argument in our SEO versus AEO comparison.
- Measurement is the cheapest competitive advantage available, because most of your competitors are not tracking any of this yet.
Frequently asked questions
What percentage of searches end without a click?
Approximately 68% of US Google searches in early 2026, rising to roughly 83% when an AI Overview is present and around 93% in AI Mode. Mobile runs far higher than desktop, and transactional queries run far lower than informational ones.
Does AI search traffic convert better than Google traffic?
On current retail data, yes: AI-referred visitors converted about 42% better than non-AI traffic in March 2026, having converted worse a year earlier. Volumes remain small, so treat this as a quality signal rather than a channel replacement.
Are these statistics reliable?
They are directionally reliable and precisely uncertain. Different panels, countries, devices, and definitions produce meaningfully different figures, which is why every number above carries its source and date, and why the section on reading them critically sits in the middle of this page rather than at the end.
How often do these figures change?
Materially, every few months in the current market. We review this page quarterly. If you are citing any search statistic in a pitch or a board pack, check its original publication date first, because a surprising share of widely quoted numbers are older than the article quoting them.
The bottom line
The direction is not in dispute: clicks are becoming scarcer, answers are absorbing them, and the visitors who still arrive are worth more than they used to be. The precision is in dispute, and anyone quoting these figures to two decimal places is selling certainty that does not exist. Use the numbers to set direction, then measure your own position, because that is the only dataset that describes you. If you want that measurement done properly, our free SEO and AEO audit reports where your business stands in both search results and AI answers, with your numbers rather than the industry's.
