AI Referral Traffic Statistics 2026: What the Public Datasets Actually Show

How much traffic does AI send to websites?

AI assistants send well under 1% of website visits in every public dataset we read: between 0.14% and 0.43%, depending on whose panel is counting and when. Ahrefs measured 0.43% across 115,254 sites in September 2026. Semrush measured 0.14% across 2025. A peer-reviewed study of 973 e-commerce sites found less than 0.2%. The channel is small, growing quickly, and dominated by ChatGPT. The popular claim that AI visitors convert far better than search visitors is contested.

This article lines up the primary datasets, puts each method next to its number, and keeps our interpretation in separate "Our reading" sections. It covers human clicks from AI answers, often called AI referral traffic, and separately covers requests from AI crawlers.

Disclosure: This analysis is published by CiteLadder, which sells an AI visibility product. Every figure below is third-party data, read from the original publisher's page on 20 October 2026. We did not collect any of these statistics ourselves.

Key numbers at a glance

Five datasets, five answers on size

Ahrefs: 0.43% in September 2026

Ahrefs' AI vs Search benchmark tracks 115,254 sites that run its free Web Analytics script. It covers June 2025 to September 2026 and was last updated on 19 October 2026. For September 2026 it puts AI search at 0.43% of tracked traffic, ChatGPT at 0.34% and Google at 23.96%.[1] Ahrefs notes that its categories overlap: "Search" and "AI Search" are defined by referring source, while direct, paid and social are channel types, so its six categories add up to 103.1%.

An earlier Ahrefs reading, published on 15 May 2026, used a different cohort of 74,752 sites. All AI chatbots combined sent 3.5 million visits in March 2026, or 0.28% of traffic. Google sent 345.2 million, a 28.12% share.[2]

Semrush: 0.14% across 2025

Semrush's channel-mix study covers more than 50,000 websites in 17 industries from January to December 2025, using its Traffic & Market Toolkit. It defines AI traffic as referral visits from tools such as ChatGPT, Perplexity and Copilot. AI was 0.14% of total traffic, against 16.04% for organic search and 64.69% for direct.[3] Semrush says it reassigns some visits that site owners would see as direct to more specific channels, so its denominator is not the same as an analytics tool's.

StatCounter: 0.24% of referrals, not of all visits

StatCounter's "Search vs Social vs AI Chatbot" chart put AI chatbots at 0.24% in July 2026, with search at 91.29% and social at 8.47%.[5] Those three shares sum to 100%, so the denominator is referrals from those three source types only. Direct, email and paid visits sit outside it. When it launched its AI chatbot data in June 2025, StatCounter described its panel as over 3.8 billion page views a month across over 1.5 million websites.[6]

Kaiser and Schulze: under 0.2% in e-commerce

The only peer-reviewed study we found is by Maximilian Kaiser and Christian Schulze in Marketing Science (volume 45, issue 4; online 21 April 2026). They used first-party Google Analytics data from 973 e-commerce websites with $20 billion in combined revenue, covering August 2024 to July 2025. Organic LLM traffic was less than 0.2% of all traffic, about 200 times smaller than Google organic search.[7]

Adobe: growth rates, but no share

Adobe Analytics covers more than 1 trillion visits to US retail sites. Its public releases report growth and conversion, but the materials we read do not state AI's share of total retail visits. For the 2025 holiday season, 1 November to 31 December, Adobe reported that traffic to retail sites from generative AI tools rose 693.4% year on year. In the same release, Adobe noted that the base of users is still modest.[8]

Our reading

The spread is a method story, not a contradiction. The panels differ: sites running Ahrefs' script, Semrush's traffic estimates, StatCounter's tracker, e-commerce Google Analytics accounts and US retailers on Adobe. The denominators differ: all visits, three referral types, or e-commerce sessions only. The periods differ by up to a year in a channel that moves month to month. Every dataset that reports both channels agrees on one thing: AI referrals are dozens to hundreds of times smaller than search.

Growth is steep, from a small base, and volatile

Semrush found that AI traffic grew 66.02% in 2025, from 462 million to 767 million monthly visits, faster than any other channel. Organic search grew 2.38% and declined in 13 of the 17 industries. AI traffic grew in 16 of 17 industries; computer software and development fell 26%. Semrush tracks Google AI Mode as a separate channel: it went from 1,600 visits in January 2025 to 38.2 million in December, still 0.01% of total traffic.[3]

Adobe reported US retail AI traffic up 393% year on year in the first quarter of 2026.[10] For the 2025 holiday season it also reported growth outside retail: travel up 539%, financial services 266%, tech and software 120%, and media and entertainment 92%.[9]

Month to month, the line is not smooth. In Ahrefs' benchmark, visitors from ChatGPT rose 59.4% from April to May 2026, fell 12.7% the next month and 9.0% the month after, then rose 33.2% and 14.1%. Across the whole table, ChatGPT averaged +5.5% a month while Google averaged -3.0%.[1]

Our reading

Triple-digit growth rates from a base under half a percent still produce a small channel. Plan around the absolute number of AI visits your own site receives, and expect single months to swing.

Which assistants send the clicks

Every dataset puts ChatGPT first by a wide margin. The order behind it changes with the panel and the month.

Sources: StatCounter[4], Ahrefs[1][2], Kaiser and Schulze[7].

StatCounter cannot count Grok, because it does not pass referral data in its header.[6] The gap behind ChatGPT has narrowed: StatCounter's launch data for May 2025 had Gemini at 2%, behind Perplexity at 11.8% and Copilot at 5.2%.[6]

Our reading

By our arithmetic, ChatGPT accounts for roughly 77% to over 90% of AI referrals across these datasets (2.7M of 3.5M is about 77%; 0.34 of 0.43 is about 79%). Rankings among the smaller assistants move between months and panels, so a single month's order is weak evidence.

Do AI visitors convert better? The evidence is split

Semrush: 4.4 times, sample not stated

A 2025 Semrush study said the average AI search visitor, tracked to a non-Google source such as ChatGPT, was worth 4.4 times an average organic search visitor, measured by conversion rate.[11] The post does not state the sites, sample size or period behind that figure. Its notes describe the study's traffic and value projections as extrapolations.

Adobe: from worse to better in US retail

Adobe reported that AI referrals converted 31% more than other traffic sources during the 2025 holiday season.[9] In March 2026, AI-referred retail visitors converted 42% more than non-AI traffic, and revenue per visit was 37% higher. A year earlier, AI visitors converted at nearly half the rate of non-AI traffic, and non-AI visits were worth 128% more.[10] The Adobe materials we read report aggregate differences and do not describe controls for differences between sites.

Kaiser and Schulze: below most channels, but improving

The peer-reviewed study compared channels within the same websites, with website, device and month fixed effects. Organic search was 13% more likely to convert than ChatGPT referrals, affiliate 86% more likely, and paid social 53% less likely. ChatGPT traffic converted and earned per session above paid social but below every other traditional channel. The gap to organic search became statistically insignificant in five of the authors' robustness checks. Conversion rates rose over the year while average order values fell, and results were stronger for complex products. The authors note that last-click attribution understates any discovery role.[7]

Our reading

The three answers come from different periods and methods. The academic data ends in July 2025, and Adobe's own figures show AI converting worse in March 2025, so the two do not clash on timing. Adobe compares AI with all non-AI retail traffic in aggregate; Kaiser and Schulze compare channels within each site. Semrush's multiplier has no published denominator. A conversion advantage for AI visitors is plausible and appears to be growing in US retail. A universal 4.4x is not supported by the data we could inspect. Measure it on your own site.

How much AI crawlers take versus send back

Cloudflare computes a crawl-to-referral ratio by dividing HTML page requests from a platform's crawlers by HTML page requests whose referrer is that platform's hostname.[12]

• For 19-26 June 2025, ratios ranged from Anthropic's 70,900:1 down to Mistral's 0.1:1.[12]
• Anthropic's ratio fell from about 286,000:1 in January 2025 to about 38,000:1 in July 2025. Perplexity was at 194:1 in July.[13]
• In the first week of August 2025, Cloudflare reported Anthropic at nearly 50,000:1, OpenAI at 887:1 and Perplexity at 118:1.[14]
• Over the 12 months to mid-2025, 80% of AI crawling was for training, 18% for search and 2% for user actions.[13]

Cloudflare adds an important caveat. Traffic referred by Claude's native app carries no referrer header, and Cloudflare believes the same holds for other native apps. That means the ratios may be overstated, by an amount Cloudflare says is unclear.[12] Cloudflare has since added site-level crawl-to-referral ratios, including per bot operator, to an Attribution Business Insights dashboard for its Bot Management customers.[15]

Our reading

These are 2025 snapshots across Cloudflare's network; Cloudflare Radar publishes the current table. Your own ratio depends on what you publish and which bots you allow. A high ratio is a reason to look at your logs, not proof that a crawler is worthless to you.

Why every count misses some AI traffic

1. App clicks often lack a referrer. Cloudflare found Claude's native app sends none.[12] Ahrefs warns that some AI visits, especially from mobile apps, will appear as direct.[2] Semrush's projections explicitly model AI clicks without referrer data.[11]
2. Google's AI features count as search. GA4's AI Assistant channel covers sources such as ChatGPT, Gemini, Deepseek, Copilot and Grok, and excludes Google's AI Overviews and AI Mode, which fall under Organic Search.[16] Semrush, by contrast, tracks AI Mode as its own channel.[3] Two "AI share" numbers can differ simply because one includes Google's AI features.
3. Influence without a click is invisible. An answer can shape a purchase that later arrives through branded search or direct. Last-click measurement, used in the academic study, does not credit that.[7]
4. Each panel represents its own customers. Sites using Ahrefs' script, US retailers on Adobe and e-commerce stores in a Google Analytics sample are not the whole web.

What to measure on your own site

Benchmarks tell you the channel is small and growing. They cannot tell you what it is worth to you. Four measurements can:

• AI Assistant sessions in GA4. Use the session default channel group, then review landing pages and key events. Keep Organic Search separate, since it includes AI Overviews and AI Mode clicks.[16]
• Conversion, with enough volume. Kaiser and Schulze found that the median ChatGPT observation, measured per site, device and week, contained zero transactions.[7] Compare conversion rates over longer windows before drawing conclusions.
• Crawler requests in your server logs. Count requests by AI user agent and compare them with AI referrals to the same pages. That gives you your own crawl-to-referral picture instead of a network average.
• The answers behind the clicks. Referrals only come from answers that link to you. Track which prompts mention your brand and which sources the answer cites, using AI citation tracking and brand mentions in ChatGPT as starting points.

CiteLadder reports GA4 referrals from recognized AI assistants and, on paid plans, AI crawler requests from a site's own logs. It also tracks how brands appear in ChatGPT, Claude, Gemini and Google AI Overviews answers and which sources those answers cite.[17] The source is public under AGPL-3.0 and can be self-hosted from the open source repository, or used as a hosted version at citeladder.com.

Frequently asked questions

How much traffic does ChatGPT send to websites?

In Ahrefs' September 2026 benchmark of 115,254 sites, ChatGPT accounted for 0.34% of tracked traffic. In Ahrefs' March 2026 reading of 74,752 sites, ChatGPT sent 2.7 million of the 3.5 million visits from AI chatbots. Your share depends on your audience and topics.

Is AI referral traffic worth tracking if it is under 1% of visits?

Our reading is that tracking is cheap and worth doing, while large budget shifts should wait for evidence from your own site. The channel is growing quickly, but the conversion evidence is mixed and differs by industry and period.

Why does GA4 not count Google AI Overviews as AI traffic?

Google's default channel definitions put clicks from AI Overviews and AI Mode under Organic Search. The AI Assistant channel covers referrals from assistants such as ChatGPT, Gemini, Copilot, Deepseek and Grok.

What is a crawl-to-referral ratio?

It is the number of pages a platform's crawlers request for each visit that platform refers back. Cloudflare calculates it from HTML requests. Because native apps often send no referrer, published ratios may overstate the imbalance.

Does AI traffic convert better than organic search?

It depends on the dataset. Adobe reports AI referrals converting better than non-AI traffic in US retail since late 2025. A peer-reviewed study of 973 e-commerce sites, covering August 2024 to July 2025, found ChatGPT referrals converting below organic search, with the gap narrowing over time.

Sources

All pages were read on 20 October 2026.

1. Ahrefs, AI vs Search Traffic Analysis, updated 19 October 2026, read 20 October 2026.
2. Ahrefs, AI Chatbot Traffic: What It Is, and How to Get More, 15 May 2026.
3. Semrush, We analyzed billions of web visits: How AI is reshaping traffic channels.
4. StatCounter, AI Chatbot Market Share Worldwide.
5. StatCounter, Search vs Social vs AI Chatbot Market Share Worldwide.
6. StatCounter, New Statcounter AI data finds ChatGPT sends 79.8% of all chatbot referrals to websites, 11 June 2025.
7. Kaiser and Schulze, Frontiers: ChatGPT Referrals to E-Commerce Websites: How Do LLMs Compare Against Traditional Channels?, Marketing Science 45(4), 2026.
8. Adobe, Holiday Shopping Season Drove a Record $257.8 Billion Online, 7 January 2026.
9. Adobe, AI-driven traffic surges across industries with retail experiencing biggest gains.
10. Adobe Digital Insights, Quarterly AI Traffic Report, April 2026.
11. Semrush, We Studied the Impact of AI Search on SEO Traffic.
12. Cloudflare, The crawl before the fall of referrals: understanding AI's impact on content providers.
13. Cloudflare, The crawl-to-click gap: Cloudflare data on AI bots, training, and referrals.
14. Cloudflare, A deeper look at AI crawlers: breaking down traffic by purpose and industry.
15. Cloudflare, Unmasking the crawls with Attribution Business Insights.
16. Google Analytics Help, Default channel group.
17. CiteLadder product information, reviewed 20 October 2026.