AI Referral Traffic Is 1 Percent of Total Visits. Here Is What to Measure Instead.
ChatGPT sends 92% of standalone AI referral traffic, but AI referrals are roughly 1% of total visits. Similarweb research shows AI recommendations drive 2.5x more visits through branded search. Here is how to measure citation presence alongside referral sessions.
AI-driven referral traffic from standalone assistants accounts for a small fraction of total website visits, and ChatGPT sends 92.4% of that. But new research from Similarweb shows AI recommendations make users 2.5x more likely to visit a brand's site — through branded search, not referral links. If you are only tracking AI referrals in GA4, you are measuring the smallest slice of AI's real influence on your pipeline.
ChatGPT Sends 92 Percent of Standalone AI Referral Traffic
Previsible analyzed 6.77 million LLM-driven sessions across 166 sites from November 2024 through May 2026. ChatGPT sent 92.4% of all trackable standalone AI referral traffic. Claude quietly overtook Perplexity in March 2026 after growing 64x in referral sessions. Perplexity's referral sessions dropped 61% from their March 2025 peak.
The part most coverage skips: Google's own AI Overviews and AI Mode already produce more AI-influenced traffic than ChatGPT, Claude, Gemini, Perplexity, and Copilot combined. That 92.4% figure describes ChatGPT's share of the smaller standalone-assistant pool — not its share of all AI-influenced traffic.
When I see teams report "AI traffic" as a single line item, they are usually looking at this small pool and missing the larger picture entirely.
AI Recommendations Drive 2.5x More Site Visits Through Branded Search
This is the finding that should change how you think about AI measurement. Similarweb tracked thousands of real user journeys across finance, travel, and beauty. Users who received an AI recommendation from ChatGPT were 2.5x more likely to visit that brand's website within seven days.
They did not arrive through AI referral links. Over 56% came through branded search queries — up from the standard 40% for non-AI-influenced visits. AI-influenced visitors also viewed nearly twice as many pages and spent twice as long on site compared to standard visitors.
The attribution gap is straightforward: when someone asks ChatGPT for a recommendation, reads the answer, and searches for that brand three days later, your analytics logs a branded search visit. The AI recommendation gets zero credit. As Rand Fishkin noted in Similarweb's study: "Whether we're talking about a brand's visibility on the side of a highway in 1926 or its presence in an AI tool's response in 2026, it's clear that influence is happening."
Each AI Engine Cites Different Brands
The referral traffic numbers tell you who sends clicks. They do not tell you who mentions your brand. Those are different metrics, and they diverge sharply by platform.
| Platform | Referral traffic behavior | Citation behavior |
|---|---|---|
| ChatGPT | 92.4% of standalone referrals | Often answers without including clickable source links |
| Claude | Grew 64x, passed Perplexity in March 2026 | Smaller citation footprint, growing fast |
| Perplexity | Down 61% from referral peak | Cites sources on every response by design |
| Google AI | More AI-influenced traffic than all standalone LLMs combined | Pulls from existing index authority |
Ahrefs analyzed roughly 75,000 brands across ChatGPT, AI Mode, and AI Overviews and found that YouTube mentions showed the strongest correlation with AI visibility at 0.737 on the Spearman scale. Branded web mentions followed at 0.66 to 0.71 depending on the engine. Backlink count and content volume — the traditional levers — fell below 0.30. AI engines reward brand presence and third-party mentions more than link authority.
Your citation presence across individual engines matters independently from whether any of them sends you a click.
How to Set Up Dual Measurement This Week
I have been running this dual setup for client campaigns since Q1. Here is the practical version.
Referral tracking (what sends clicks): Build a custom channel grouping in GA4 that maps chatgpt.com, perplexity.ai, claude.ai, gemini.google.com, and copilot.microsoft.com to an "AI" channel. Out of the box, GA4 scatters these across Direct and Referral with no useful segmentation. Track sessions, engagement rate, and conversion rate by AI source separately.
Citation monitoring (what mentions you): Track where your brand appears in AI answers regardless of whether users click through. This requires monitoring queries against ChatGPT, Perplexity, Claude, and Gemini to see if and how your brand is cited. The Machine Relations Index tracks citation rates across six AI engines at the segment level — citation rate measurement captures the influence signal that referral dashboards miss entirely.
Branded search lift: Watch for increases in branded search volume that correlate with AI visibility gains. Similarweb's research shows branded search is where most downstream value actually appears — 56% of AI-influenced visits arrive this way.
Compare per-platform conversion quality, not just volume. A platform that sends 50 sessions at a 12% conversion rate is more valuable than one that sends 500 at 0.3%.
Why Citation Presence Compounds and Referral Numbers Fluctuate
Referral sessions move with product changes, link placement decisions, and platform growth trajectories. Perplexity's 61% referral decline does not mean brands cited by Perplexity lost influence. Those users still saw the recommendation. The downstream visit came through branded search instead.
Citation presence compounds differently. When AI engines consistently cite a brand, three things happen: consumers encounter that brand across the 2.4 platforms they now check before purchase decisions. Branded search volume grows as AI-primed users search by name. Entity authority strengthens across the graph that AI engines use to decide who to cite next.
Fractl's research presented at SMX Advanced categorized GEO tactics into three tiers. Original data and digital PR form the moat — the tactics hardest for competitors and AI to replicate, as the Ahrefs correlation data independently confirms. FAQ optimization, the most popular tactic at 49% adoption, is the highest risk because AI can reproduce it.
The execution move: stop reporting "AI traffic" as one number. Report referral sessions by platform, citation presence by engine, and branded search lift over time. That is where the compounding value lives.
FAQ
How much referral traffic do AI search engines send to websites?
AI-driven referral traffic from standalone assistants like ChatGPT, Claude, and Perplexity remains a small fraction of total website visits for most sites. Previsible's analysis of 6.77 million LLM-driven sessions showed AI referral traffic grew 9.9x over 19 months, but Google's AI Overviews and AI Mode send more AI-influenced traffic than all standalone assistants combined.
Which AI platform sends the most referral traffic to websites?
ChatGPT sends 92.4% of all trackable standalone AI referral traffic, according to Previsible's 2026 AI Traffic Report. Claude overtook Perplexity in March 2026, growing 64x in referral sessions. Perplexity's referral sessions declined 61% from their March 2025 peak.
What is the difference between AI referral traffic and AI citation presence?
AI referral traffic measures visits where a user clicks a link in an AI-generated answer. Citation presence measures how often AI engines mention a brand in answers regardless of clicks. Similarweb's 2026 research shows AI recommendations make users 2.5x more likely to visit a brand through branded search — not through referral links — making citation presence the larger influence on pipeline.
How do I track AI referral traffic in Google Analytics 4?
Create a custom channel grouping in GA4 that maps chatgpt.com, perplexity.ai, claude.ai, gemini.google.com, and copilot.microsoft.com referrers to an "AI" channel. By default GA4 scatters these across Direct and Referral categories. Segment by AI source to compare sessions, engagement rate, and conversion rate across platforms.