ChatGPT dominates AI referral traffic, but page type decides whether visits convert
Previsible’s 6.77 million-session study shows AI referrals are growing fast, consolidating around ChatGPT and behaving differently by page type. For GEO teams, the headline is not only traffic share; it is where AI sends users and whether those pages are ready.
Editor’s Note by Dageno: Dageno AI’s editorial team structurally distilled Search Engine Land’s article “ChatGPT commands 92% of AI referral traffic. Here’s what 6.77 million sessions reveal.” to help GEO teams understand AI referral concentration, volatility, Claude’s growth and landing page readiness.
Understand the article in 60 seconds
- Previsible analyzed 6.77 million LLM-driven sessions across 166 GA4 properties from November 2024 through May 2026.
- Monthly LLM sessions grew 9.9x to 644,478 in May 2026, while ChatGPT accounted for 92.4% of trackable standalone LLM referral traffic.
- AI referral traffic is volatile: ChatGPT referrals dropped from 448,412 to 213,345 in one month before recovering to 442,609 by December.
- The most actionable finding is landing page type: ChatGPT sends 28.8% of its traffic to internal search pages, and roughly 25% of AI-referred traffic lands on internal search across industries.
Standalone AI referral traffic is consolidating around ChatGPT
The article reports that ChatGPT now accounts for 92.4% of trackable standalone LLM referral traffic in the Previsible dataset, up from roughly 84% in December 2025.
For GEO teams, that means “AI visibility” cannot be an abstraction. ChatGPT deserves first-priority monitoring where referral traffic is the outcome, while Google AI surfaces require separate measurement because they operate inside search results.
AI traffic can disappear because one platform changes behavior
Sessions dropped 50% in one month when ChatGPT referrals fell from 448,412 to 213,345, before recovering to 442,609 by December. The article links the pattern to model-related product shifts.
This is a risk-management signal. AI referral growth is real, but it is not stable enough to treat as a normal channel without volatility planning.
Claude is becoming material for technical and professional audiences
Claude grew 64x, from 133 sessions in November 2024 to 8,528 in May 2026, overtaking Perplexity in March 2026 and staying ahead.
If the buyer audience includes technical buyers, developers, professional services or enterprise users, Claude visibility should be monitored even if its total share remains smaller than ChatGPT.
Dageno view: The strategic question is not only what the article says, but which answer paths, source gaps and business metrics should change because of it.
Perplexity and Copilot are weaker traffic bets
Perplexity peaked at 17,507 monthly sessions in March 2025 and fell 61% afterward. Copilot dropped 96% from its August 2025 peak, from 8,651 sessions to 339.
That does not mean these platforms are irrelevant, but teams should not expect them to produce the same referral value. They may matter more for influence, citations or enclosed agent workflows than traffic acquisition.
Internal search is now an acquisition surface
ChatGPT sends 28.8% of its traffic to internal search results pages, and roughly a quarter of AI-referred traffic across industries lands on internal search.
The model trusts the domain but may not choose the exact page. That makes internal search UX, result relevance, merchandising and conversion paths part of GEO performance.
Page type reveals different buyer intent by vertical
The article notes that SaaS traffic often lands on search pages, publishers on news pages, ecommerce on product pages, education on course pages, health on About pages and legal traffic across evaluation pages.
A site-wide AI referral average hides the work. GEO teams should evaluate AI traffic by page type because each type reflects a different buyer job and conversion path.
Product pages and pricing need machine-readable detail
The article recommends treating product pages as AI entry points and making pricing machine-readable. “Contact us” pages give AI systems little to summarize, compare or recommend.
For ecommerce and SaaS, structured product data, comparison-ready attributes, pricing logic and availability can determine whether an AI referral lands on a page that converts.
The unanswered question is conversion rate by LLM platform
The study says the next question is which platforms send users who buy and which send users who bounce. The data already shows answers will change quickly.
Dageno would build this into reporting now: platform, page type, intent family, conversion rate, assisted revenue and bounce behavior should be separated before AI referral traffic becomes too large to untangle.
Facts to convert into AI evidence
These source facts should become reusable evidence blocks: short, specific, current and easy for AI systems to retrieve without guessing.
6.77 million LLM-driven sessions across 166 GA4 properties were analyzed.
The study spans 19 months and multiple verticals, making it a useful directional benchmark.
ChatGPT generated 92.4% of trackable standalone LLM referral traffic.
Optimization should prioritize ChatGPT when referral traffic is the KPI.
ChatGPT referrals halved in one month before recovering.
Teams need monitoring that can distinguish channel growth from product-change shocks.
Internal search receives roughly a quarter of AI-referred traffic.
Internal search should be treated as an acquisition and conversion surface.
How teams should respond across the AI visibility workflow
The response is cross-functional because AI visibility is shaped by content quality, source access, commercial context, third-party proof, tooling decisions and measurement discipline.
| Owner | Execution focus | Signal to watch |
|---|---|---|
| GEO team | Turn prompt and answer-surface findings into source, citation, recommendation and competitor-share priorities. | Prompt-family visibility, citation accuracy and recommendation share. |
| SEO team | Connect technical accessibility, internal linking, commercial page updates and AI-readable evidence to business outcomes. | AI Overview inclusion, crawl access, internal link support and revenue-page movement. |
| Content strategy | Build topic clusters from competitive gaps, first-party data and buyer intent rather than keyword similarity alone. | Content gap priority score, existing authority signals and engagement by cluster. |
| Analytics / BI | Tie AI referral traffic, organic revenue, profit, margins and page type behavior into one measurement layer. | Organic profit, AI referral sessions, conversion by page type and attribution confidence. |
| Procurement / leadership | Evaluate AI vendors by business problem, proof, data policy, implementation cost and team adoption burden. | Pilot value, contract risk, data governance clarity and internal lift. |
Signals to monitor next
Measurement should show whether the brand is being found, understood, cited, recommended and converted as answer surfaces change.
- AI referral traffic by platform, not only total AI traffic.
- AI referral landing pages grouped by page type and vertical intent.
- Internal search conversion rate for AI-referred sessions.
- ChatGPT referral volatility and recovery after product shifts.
- Claude visibility for technical, developer and professional-service audiences.
Turn AI visibility signals into execution.
Dageno AI is a complete execution platform from data monitoring to strategy, content generation and result attribution. Use it to monitor AI answer surfaces, find strategy gaps, generate content actions and connect visibility shifts to business impact.
Alexa for Shopping