AI search referral traffic is still a small part of the overall web, but it is growing quickly and often arrives with strong engagement. Semrush measured AI traffic at 0.14% of total web traffic in 2025, while Similarweb estimated more than 1.13 billion AI-platform referral visits in June 2025, up 357% from June 2024. The figures below show where those visits come from, how they compare with other channels, and why crawling volume does not automatically translate into referrals.
Contents
- AI referral traffic scale and growth
- Which AI platforms send referrals
- AI referrals compared with other channels
- Engagement, conversion, and revenue quality
- AI referrals by industry
- Consumer use and clicking behavior
- AI crawling versus referral efficiency
AI referral traffic scale and growth
Semrush’s 2025 traffic-channel study found that AI traffic grew 66.02% during 2025 and represented 0.14% of total web traffic. AI traffic increased in 16 of the 17 industries included in that analysis. The same study recorded 544.2 billion web visits across its 2025 comparison, down 0.43% from 546.5 billion.
The growth rate is much more pronounced in Similarweb’s platform-referral estimate. Similarweb reported more than 1.13 billion AI-platform referral visits worldwide in June 2025, a 357% increase from June 2024. In a later period covering June 2025 through May 2026, Similarweb reported an average of 770.7 million AI referral visits per month worldwide, up 117.4% year over year. These estimates use different periods and should not be treated as a single continuous series.
Google AI Mode also expanded sharply within the Semrush dataset. Its traffic rose from 1,600 visits in January to 38.2 million in December 2025, although it still accounted for 0.01% of total traffic in December. By comparison, organic search generated more than 1 trillion visits in 2025 and grew 2.38%.
The scale matters because a small percentage of a very large web can still produce meaningful traffic. It also matters because growth percentages are sensitive to a small starting base. A 66.02% increase and a 357% year-over-year increase describe rapid expansion, but neither says that AI referrals have replaced organic search.
Which AI platforms send referrals
Semrush’s July 2025 analysis of 10 major AI platforms showed a highly concentrated referral market. ChatGPT generated 5,244,855,278 visits and held an 85.79% share among those platforms. Gemini generated 287,535,861 visits, or 4.70%, while Perplexity generated 173,581,751 visits, or 2.84%.
Grok generated 153,011,519 visits and a 2.50% share. Claude generated 136,577,834 visits and a 2.23% share. Microsoft Copilot generated 97,802,733 visits and a 1.60% share. These shares are specifically for the 10-platform comparison in July 2025, rather than all possible AI sources.
| AI platform | Visits in July 2025 | Share among 10 platforms |
|---|---|---|
| ChatGPT | 5,244,855,278 | 85.79% |
| Gemini | 287,535,861 | 4.70% |
| Perplexity | 173,581,751 | 2.84% |
| Grok | 153,011,519 | 2.50% |
| Claude | 136,577,834 | 2.23% |
| Microsoft Copilot | 97,802,733 | 1.60% |
Similarweb’s mid-2025 analysis likewise found that ChatGPT accounted for more than 80% of AI referrals to its top 1,000 global domains. Semrush’s clickstream analysis from January 2025 through January 2026 found that more than 30% of ChatGPT referral traffic went to 10 domains, and more than 20% went to Google. This concentration means that visibility in one platform can have a disproportionate effect on reported AI referral totals.
ChatGPT’s search feature was enabled on 34.5% of queries in February 2026, according to Semrush, compared with 46% in late 2024. Search activation is not the same as a site referral, but the figures show why platform behavior and referral reporting can move independently.
AI referrals compared with other channels
AI traffic’s 0.14% share in Semrush’s 2025 dataset was smaller than several established channels. Organic search represented 16.04% of total traffic and grew 2.38%. Referral traffic represented 9.60% and grew 53.39%. Organic social represented 1.47% and declined 8.86%.
Some smaller channels had higher growth rates than AI traffic. Paid search grew 75.84% in 2025 but represented 0.49% of total traffic. Display advertising grew 62.76% and represented 0.09%. The comparison shows two separate dimensions: current scale and rate of change. AI traffic grew rapidly while remaining much smaller than organic search and general referral traffic.
For SEO teams, this supports measuring AI referrals as a distinct acquisition source rather than folding them into a broad referral total. A channel that represents 0.14% of visits can still deserve attention if its growth and engagement differ from the site average. At the same time, the organic-search baseline remains much larger in the 2025 Semrush data.
Engagement, conversion, and revenue quality
Adobe’s retail analysis for May 2025 found that visits from AI referrals lasted 38% longer than non-AI visits, produced 10% more page views per visit, and had a bounce rate 27% lower than non-AI traffic. Desktop devices accounted for 82% of AI-driven retail visits, compared with 34% of all retail traffic.
The retail conversion gap also narrowed over time. AI referrals converted 91% lower than non-AI traffic in July 2024, but the gap was 22% lower in May 2025. Revenue per visit rose from 3% of non-AI revenue per visit in July 2024 to 70% in May 2025. Those figures describe relative retail performance, not a universal conversion rate for every site.
Adobe’s holiday-season report showed an even stronger retail result. From November 1 through December 31, 2025, AI referrals converted 31% better than non-AI sources. AI referral conversions were 54% higher on Thanksgiving 2025 and 38% higher on Black Friday 2025. AI-driven traffic also showed 14% higher engagement during the November–December holiday period.
These measurements suggest that referral volume alone is an incomplete KPI. The useful comparison is the full path from source to visit quality, conversion, and revenue per visit. Results can vary by season, device mix, industry, and the type of user intent represented by the referral.
AI referrals by industry
Travel saw some of the largest increases in Adobe’s holiday 2025 analysis. AI-driven visits to travel sites grew nearly 539% year over year from November 1 through December 31, 2025, and were up 405% from January 2025. The travel conversion gap narrowed from 86% below non-AI traffic in October 2024 to 13% below it during the 2025 holiday season.
Travel referrals also showed strong on-site behavior during November and December 2025. Engagement was 16.8% higher than for traditional sources, the bounce rate was 33% lower, and pages per visit were 21% higher. Time on site was 65% longer than for non-AI sources.
Other sectors also recorded substantial growth during the same holiday period. AI-driven visits to financial-services sites grew 266% year over year, while visits to media and entertainment sites grew 92% year over year. These are Adobe measurements for November 1 through December 31, 2025, so they should not be generalized to every month or business model.
The industry pattern is consistent with a referral channel that may be especially useful when users need comparison, planning, or explanation before taking action. The available statistics quantify growth and engagement, but they do not establish that one content format or optimization tactic caused the results.
Consumer use and clicking behavior
Adobe’s Holiday 2025 survey found that more than one-third of shoppers used AI Assistants during the holiday season, while about half used them for holiday shopping. In the same survey, 81% of consumers said AI Assistants improved their shopping experience, 47% reported trusting them, and 64% said they were using AI Assistants more than they used to.
A separate Adobe survey from August 2025 found that 65% of consumers using AI for online shopping said it made them more confident in their purchase. Another 68% said they were less likely to return the product afterward. These are survey responses, not observed referral or transaction records.
More than 55% of surveyed consumers said they click links provided by AI Assistants. That result helps explain why AI visibility can matter even when an assistant answers part of a question directly. A link click creates an opportunity for a site visit, but the eventual value still depends on landing-page relevance and the quality of the user’s intent.
AI crawling versus referral efficiency
Cloudflare measured AI bots at an average 4.2% of HTML requests across its customer base during 2025. Over the 12 months through July 2025, 80% of AI crawling was for training, 18% for search, and 2% for user actions. Crawling volume therefore represented a different activity from referral traffic.
The mix changed between July 2024 and July 2025. Training’s share of AI crawling rose from 72% to 79%, while search fell from 26% to 17%. GPTBot’s share of AI-only crawler traffic increased from 11.9% to 28.1%, and ClaudeBot’s share increased from 15.0% to 23.3%.
Cloudflare’s crawl-to-refer ratios show why bot requests should not be used as a proxy for visits. Anthropic’s ratio fell from 286,930.1 crawled pages per referred visit in January 2025 to 38,065.7 in July, an 86.7% decline. OpenAI’s ratio fell from 1,217.4 crawled pages per referred visit to 1,091.4 over the same period, a 10.4% decline. Perplexity moved in the opposite direction: its ratio rose from 54.6 crawled pages per referred visit in January to 194.8 in July, a 256.7% increase.
For reporting, the distinction is practical. AI crawler requests describe access to content by automated systems; AI referral sessions describe people arriving from an AI platform. The two measures can move in different directions, so analytics dashboards should keep them separate and label the measurement period, geography, and source methodology whenever AI traffic is reported.