ChatGPT dominates AI referral traffic: study

A year ago, the industry was still betting on which AI platform would dominate web discovery. Perplexity looked like the search-native challenger, Microsoft Copilot like the enterprise Trojan horse. Both bets failed. Previsible's third AI Traffic Study analyzes 6.77 million LLM-driven sessions and shows clear consolidation: monthly LLM sessions grew 9.9x, reaching around 644,478 visits in May 2026. Of that traffic, 92.4 percent comes from a single platform.

For SEO and GEO teams, this means AI referral traffic is no longer a fringe phenomenon but a measurable channel with its own volatility, landing page patterns, and platform-specific priorities. Anyone planning visibility in generative search surfaces needs to understand where users land and which product decisions can halve traffic overnight.

The plateau was a pause

In mid-2025, AI traffic looked like it was plateauing in some sectors. The data shows the opposite. Sessions rose from 65,249 in November 2024 to 396,278 in August 2025, dropped sharply in November 2025, and hit new highs of 428,203 in February 2026 and 644,478 in May.

The November drop needs context: sessions fell 50 percent in one month, almost entirely because ChatGPT referrals dropped from 448,412 to 213,345. Other platforms held steady. Similar effects appeared when many sites lost half their ChatGPT traffic because the model increasingly favored Wikipedia and Reddit. By December, values recovered to 442,609.

The lesson for marketing and SEO leaders: one vendor's product decision can halve AI traffic overnight. Roadmaps should plan for volatility, not just growth.

Consolidation, not competition

In December 2025, ChatGPT still held roughly 84 percent market share, followed by Perplexity at 8.9 percent, Gemini at 4.5 percent, Copilot at 2.1 percent, and Claude at 0.6 percent. Six months later, the field has collapsed further toward the leader.

Across the full dataset, ChatGPT sends 92.4 percent of measurable LLM referral traffic and grew 12.8x over 19 months. It is the only platform delivering referral volume at meaningful scale. GEO strategies that do not prioritize ChatGPT often optimize for an abstraction.

Important framing: this measures standalone LLM referral traffic. AI discovery inside Google, including AI Overviews, almost certainly drives more AI traffic than all standalone platforms combined, but it follows a different measurement paradigm and is excluded here.

The challengers flipped

The surprise is not at the top but underneath. Claude grew 64x from 133 sessions in November 2024 to 8,528 in May 2026 and overtook Perplexity for the first time in March 2026. After a flat 2025, Claude accelerated 4x in two months as agentic tools and enterprise integrations gained traction.

Gemini, Perplexity, and Copilot

Gemini is the quiet number two: 3.2x growth with almost no volatility. Workspace and Android integrations suggest referral numbers undercount real discovery reach. Perplexity peaked at 17,507 monthly sessions in March 2025 and has fallen 61 percent since. Copilot collapsed from 8,651 sessions in August 2025 to 339, a 96 percent drop.

Neither is much of a growth bet for traffic acquisition anymore. Both increasingly keep users inside their own browsers, agents, and modes where external referrals are unnecessary. For technical audiences, developers, and professional services, Claude visibility is now material, and the window for early positioning remains open.

Where LLMs send users

The most actionable finding is not market share but landing page distribution. ChatGPT sends 28.8 percent of its traffic to internal search results pages. Across industries, roughly 25 percent of AI-referred traffic lands on internal search.

The model trusts the domain but cannot confidently pick the right page, so it sends users to internal search. This pattern persists across verticals and time periods and appears structural to retrieval-augmented generation, not a temporary quirk. Internal search becomes an acquisition surface, not just navigation.

• SaaS: 34.6 percent of sessions on search pages.
• Publishers: 54 percent on news pages, with 0.11 percent penetration versus organic traffic.
• E-commerce: product pages with purchase intent already formed.
• Education: 52 percent directly on course pages.
• Health: 42.1 percent on About pages for source evaluation.
• Legal: spread across blog, about, contact, and location pages.

Platforms behave differently: ChatGPT and Gemini follow search patterns with domain trust and page-level uncertainty. Perplexity and Claude select specific content pages and over-index on long-form formats. If your strategy depends on editorial content, Perplexity and Claude matter disproportionately to their share.

Action steps for SEO teams

• Optimize for ChatGPT first and expand to other platforms when volume justifies it.
• Monitor Claude and position early because early visibility compounds.
• Treat product pages as AI entry points; structured product data is now required.
• Make pricing machine-readable instead of contact-only pricing pages.
• Prioritize internal search and build it as a conversion lever.
• Track AI traffic by page type, not only site-wide averages.

The dataset covers 166 GA4 properties from November 2024 through May 2026 across SaaS, e-commerce, finance, legal, health, insurance, education, publishing, and ticketing. All properties are present throughout the full 19-month window, so trends reflect behavioral change rather than sample expansion.