Research

ChatGPT's AI search share falls from 76% to 53% as Gemini gains

ChatGPT's share of generative AI web traffic fell from roughly 76% in June 2025 to about 53% by May 2026, according to Similarweb's generative AI statistics page, published 29 July 2026 and drawn from the company's proprietary worldwide traffic panel combining clickstream data with dynamic

6 min read
A stylized crown symbolizing ChatGPT's slipping dominance is overshadowed by Gemini's rising crown.
Illustration: Prompt the Market
James Dore
By James Dore · 2026-09-24

TLDR

ChatGPT's share of generative AI web traffic fell 23 percentage points in a year while Gemini tripled its share to nearly 28%. The catch: most AI referral traffic still lands on homepages, not the deep pages the engines actually cite, meaning a citation does not automatically mean a click.

KEY TAKEAWAYS

01ChatGPT's share of generative AI web traffic fell from 76% to 53% over twelve months to May 2026.
02Gemini tripled its share from under 9% to nearly 28% in the same period.
0358.8% of ChatGPT's referral traffic lands on homepages, despite most cited URLs sitting deeper in site architecture.
04Travel and Hospitality prompts carry citations 22.6% of the time, nearly six times the rate for Professional Services at 3.9%.
05Monthly generative AI web visits reached 9.5 billion by May 2026, up 70% year on year.

How the market split

ChatGPT's share of generative AI web traffic fell from roughly 76% in June 2025 to about 53% by May 2026, according to Similarweb's generative AI statistics page, published 29 July 2026 and drawn from the company's proprietary worldwide traffic panel combining clickstream data with dynamic click-through-rate models and search engine result page crawls.[1] The platform still leads, but its dominance has been cut almost in half in a year.

Gemini's share rose from under 9% to around 27 to 28%, while Claude climbed from about 2% to close to 9% over the same period.[1] Google pushed Gemini into its own products aggressively. Anthropic expanded its enterprise footprint. Together they took roughly 30 percentage points off ChatGPT's lead.

Average monthly web visits across generative AI platforms reached 9.5 billion by May 2026, up 70% year on year. Unique visitors grew 57% to 655 million and app downloads rose 58% to 4.4 billion.[1] The category grew fast enough that even a platform losing share is still handling more traffic than it was twelve months ago.

Google's AI Overviews, the summary boxes that appear above standard search results, now trigger on more than 40% of United States searches.[3] That compresses organic click-through rates further and pushes brands toward a question they have not yet answered cleanly: which engine matters most for referrals?

Citations versus clicks

Earning a citation from an AI engine and earning a visit from one turn out to be two different things. Aleyda Solis, founder of search consultancy Orainti, said: "The data shows a clear disconnect between the pages ChatGPT cites and those receiving referral traffic. 65% of cited URLs sit two or three folders deep, with folder depth two alone accounting for 41.7% of citations. In contrast, 58.8% of AI referral traffic lands on homepages."[3]

The engine cites a product page or a category guide buried three levels into a site's architecture. The person who acts on that answer ends up on the homepage. The deep page earned the citation; the homepage absorbed the traffic.

ChatGPT's citation rate in the United States rose from about 1.6% in June 2025 to roughly 6.8% by May 2026, so more answers are carrying source links than before.[1] Brands monitoring AI appearances need a second measure alongside citation-rate tracking: where the traffic actually lands.

For Australian brands, this changes what homepage authority means. It is less about converting first-time visitors from conventional search and more about being the logical destination when an AI engine sends someone who already trusts the brand from a citation. The homepage becomes a transfer point.

Which categories get cited most

In May 2026, Travel and Hospitality prompts carried citations 22.6% of the time. Automotive sat at 19.5%. Professional Services came in at 3.9%.[2] A travel brand is almost six times more likely to be cited in a given AI response than a professional services firm.

Travel queries tend to resolve around specific facts, prices and destinations, things an AI engine can cite cleanly. Professional services queries often resolve around general reasoning, where the engine synthesises rather than quotes. Brands in low-citation categories need to create structured, citable content that gives an AI engine something concrete to point at.

Rand Fishkin, co-author of Zero Click Marketing and founder of SparkToro, said the shift demands new measurement rather than old attribution models: "It's clear that AI influence is happening. What marketers need now is a new way to measure and attribute that impact."[3]

The practical read for Australian brands: track all three major engines, not just ChatGPT. Monitor homepage referral traffic as a proxy for AI-driven visits. In high-citation categories like travel, the citation-to-click gap Solis identified is the most immediate thing to close. Similarweb's full generative AI landscape report covers the twelve months to May 2026.

FREQUENTLY ASKED QUESTIONS

Why has ChatGPT's share of AI search traffic fallen so sharply?
Gemini and Claude have both grown their user bases significantly. Gemini tripled its share from under 9% to nearly 28% as Google pushed it into its own products. Claude grew from about 2% to close to 9% as Anthropic expanded its enterprise reach. The overall category grew 70% in monthly visits, so the pie got bigger even as ChatGPT's slice shrank.
What does the citation-versus-traffic disconnect mean for a brand's AI search strategy?
An AI engine may cite a deep product or category page while the person who acts on that answer navigates to the homepage instead. Brands need to track two things: citation appearances across the engines, and where referral traffic actually arrives. Homepage authority matters as much as the quality of deep pages.
Which industries benefit most from AI search citations?
Travel and Hospitality leads at 22.6% citation rate per prompt, followed by Automotive at 19.5%. Professional Services sits at 3.9%. The gap reflects how AI engines handle different query types: factual, specific queries in travel and automotive are easier to cite cleanly than general advisory queries in professional services.
James Dore

James Dore

James Dore is Strategy Director at Bushnote, a leader in SEO and AEO in Australia. He writes on search, AI and the technology decisions shaping government and industry.

The prompt
Every weekday morning: the five stories moving marketing, media and money, with the number that matters in each.
Sending your link…
Check your inbox to confirm.
That didn't go through. Try again.
Related topics

Make us a preferred source on Google

Tap once and our reporting shows at the top of your Google search results and AI answers. You can change this at any time.

Add as a preferred source on Google
Subscribe, it's free