AI Referral Traffic 2026: Growing Fast, Still Tiny

Reviewed 13 August 2026. AI platforms are sending more visitors to websites, but their growth rate can obscure their size. A channel that grows 66% from a small base can remain tiny beside organic search. The latest evidence suggests businesses should measure AI referrals now without treating them as a replacement for search.

Key findings

  • Semrush analysed more than 50,000 sites across 17 industries and found AI referral traffic grew 66% during 2025 but represented only 0.14% of total visits.
  • In the same study, organic search generated more than one trillion visits and represented 16.04% of the measured channel mix.
  • Similarweb estimated generative-AI platforms averaged 9.5 billion monthly web visits between June 2025 and May 2026, up 70% year on year.
  • An independent 2026 research paper estimated outbound clicks occurred in only 5.2% of ChatGPT conversation sessions in its dataset.
  • AI usage, citations and referral visits are different metrics. None should be used as a substitute for leads or revenue.

Three datasets describe different stages of the journey

The Semrush traffic-channel study measured worldwide visits across more than 50,000 websites and 17 industries from January to December 2025. It separated AI referrals from Google AI Mode and reported median channel patterns.

Similarweb’s 2026 Generative AI Landscape analysis focuses on traffic to AI platforms as well as their citation and referral behaviour. It found 9.5 billion average monthly web visits across generative-AI platforms between June 2025 and May 2026, with 655 million unique visitors.

The paper Answering Without Referring examined the economic relationship between AI answers and outbound website traffic. Its 5.2% ChatGPT outbound-click estimate is about sessions inside the assistant, not the share of destination-site traffic.

StudyUnitHeadlineDoes not mean
SemrushVisits to 50,000+ sitesAI referrals +66%; 0.14% shareAI use is only 0.14%
SimilarwebVisits to AI platforms9.5b monthly visits; +70%9.5b visits clicked to sources
Academic paperChatGPT conversation sessions5.2% produced outbound clicks5.2% of users bought something

Momentum and scale tell different stories

Semrush found monthly AI referral visits grew from 462 million to 767 million during 2025. Organic search grew only 2.38%, but its absolute volume exceeded one trillion visits across the year. A manager looking only at percentage growth could overfund the smallest channel; one looking only at current share could miss an emerging source.

The practical response is a barbell: protect the SEO programme that captures meaningful search demand, while establishing low-cost measurement and selective experiments for AI discovery. Our AI SEO services can set up that tracking and testing.

AI platforms themselves are fragmenting

Similarweb estimated ChatGPT’s share of generative-AI website visits fell from roughly 76% in June 2025 to about 53% in May 2026. Gemini increased from under 9% to around 27–28%, while Claude approached 9%. ChatGPT did not necessarily shrink; the category expanded faster and competitors took more of the growth.

This makes a single-platform “ChatGPT optimisation” plan too narrow. Content should be accurate, accessible, well-supported and easy to discover across search and the open web. A broader search engine marketing strategy can compare paid search, organic search and emerging referral sources using common commercial outcomes.

How to measure AI referral traffic

  1. Create a channel definition. Group known assistant referrers in analytics, but retain the source platform for comparison.
  2. Expect undercounting. Some journeys lose referral information, move between apps and browsers, or lead to a later branded search.
  3. Compare landing pages. Identify whether AI visitors land on deep educational pages, homepages or product pages.
  4. Measure quality. Compare engagement, qualified leads, ecommerce conversion, average order value and assisted revenue.
  5. Track visibility separately. Citation frequency and brand mentions can change without producing a click.

Businesses can use paid search campaigns as a controlled demand test while organic and AI visibility mature, but attribution models should not award every conversion to the final click automatically.

Limitations

  • The studies use modelled panels, tracked-site populations and classifications that may not represent an individual Australian business.
  • AI referrals can be misclassified as direct or referral traffic and may exclude app-to-browser journeys.
  • Worldwide platform traffic is not the same as Australian adoption.
  • Fast percentage growth from a small base can appear more commercially important than it is.
  • Referral sessions do not measure citations, brand influence or offline conversions.

The bottom line

AI referral traffic is real and growing quickly, but it remains small beside organic search in the broadest current dataset. Businesses should instrument it, learn which pages and platforms produce qualified visits, and avoid false either-or decisions. The sensible 2026 strategy is to maintain search visibility while building measurement and content practices that also work across assistants.

i 3 Table Of Content

SEO Summer is a digital agency based in Melbourne Australia. We've helped 100+ clients reach the first page of Google Search. Get in touch to see how we can help you.

Our Marketing Services

Social Media Marketing
Email Outreach

Written By Victor Volynski

Vic is the Founder and Lead SEO at SEO Summer. He has over 10 years experience in building brands through digital marketing. Vic has worked with businesses in all niches, all the way from local businesses to international brands.

Vic focuses on the cutting edge of SEO, employing the most effective techniques in the space to achieve the best results for clients.