Google vs ChatGPT: Who's Sending Traffic to Your Site?

 Previsible's report splits AI discovery into two separate measurement buckets, and the distinction matters for how marketers should read the results. The first bucket is AI activity inside Google itself — AI Overviews and AI Mode — which the study excluded from its standalone LLM referral count because Google's AI search features don't produce trackable referral sessions the same way an outbound click from ChatGPT does. The second bucket is referral traffic from standalone assistants that a Google Analytics 4 property can actually attribute to a named source. 

Measured this way, Google's AI surfaces account for more AI-influenced traffic than every standalone assistant combined. That's a meaningful finding on its own, since it means the biggest AI discovery channel by volume isn't a chatbot at all — it's still Google, just wearing a different interface.

Key Points

• Previsible's third AI Traffic Study analyzed 6.77 million LLM-driven sessions across 166 websites from November 2024 to May 2026
• Google's AI Overviews and AI Mode combined drive more AI-influenced discovery than all standalone LLM platforms put together
• Among standalone assistants, ChatGPT accounts for 92.4% of measurable referral traffic, up from roughly 84% in December 2025
• Claude grew 64 times over the study period and passed Perplexity in March 2026, with strength among developers and professional services audiences
• AI traffic penetration varies widely by industry, with e-commerce up 37 times and health the only vertical to decline

ChatGPT's Grip on Standalone Referrals Is Tightening

Within the standalone assistant category, ChatGPT's dominance grew during the study window rather than eroding. Its share of measurable referral traffic rose from about 84% in Previsible's December 2025 review to 92.4% in the current one, and its monthly referred sessions climbed from 47,606 in November 2024 to 610,910 in May 2026 — a 12.8-fold increase.

The more interesting movement is underneath ChatGPT. Gemini grew 3.2 times over the same period and holds the second position. Claude grew 64 times, overtaking Perplexity in March 2026 and holding that spot through May, with referral traffic concentrated among developers, technical buyers, and professional services audiences — a distinct pattern from the more consumer-facing traffic Perplexity and ChatGPT tend to draw.

Copilot moved in the opposite direction, falling from 8,651 monthly sessions in August 2025 to 339 in May 2026. The study also recorded a sharp, temporary drop across standalone referrals in November 2025 — ChatGPT alone fell from 448,412 sessions in October to 213,345 in November before recovering the following month — though the report didn't attribute the dip to a confirmed cause.

Where AI Traffic Lands Depends on the Industry

The report's page-level data is where the practical guidance lives. E-commerce saw a 37-fold increase in AI referral traffic, with product pages as the primary landing surface and ChatGPT as the near-exclusive driver. Insurance grew 18.9 times to 1.51% of total sessions, and education grew 5.4 times, with course pages absorbing over half of that category's LLM referrals. Health was the sole vertical to decline, dropping from 0.23% to 0.17% of sessions, though "about" pages still captured over 40% of the health referrals that did occur — consistent with users checking a source's credibility after an AI assistant pointed them there.

Across every industry in the dataset, roughly a quarter of AI-referred traffic landed on internal search results pages rather than a specific piece of content, a pattern worth noting for anyone auditing where their site's AI traffic is actually going.

What This Means for How Teams Prioritize AI Visibility

The report's authors frame the practical order of operations simply: build the site architecture and content signals that make Google's AI results want to cite you first, then treat ChatGPT as the leading standalone surface worth winning next. That sequencing matters for teams deciding where to put effort, since it suggests traditional SEO fundamentals and generative engine optimization aren't competing priorities — they're the same foundation serving two different discovery layers.

For teams building out their AI-informed content strategy, the industry-specific landing page data is a useful starting point for auditing which pages are actually positioned to catch AI referral traffic today. And for anyone shaping a broader growth strategy around where AI discovery is heading, Claude's growth among technical and professional audiences is worth watching closely if that's the buyer profile in question — it's a smaller number today, but the growth rate is the fastest in the dataset.

More detail on the full study is available via Marketing Tech News.