For years the maths was simple: whoever ranked at the top of the search engine got visitors. But in 2026 a second route to the website has opened up. More and more people ask their questions not of a search box but of an AI assistant such as ChatGPT, and these assistants send real visitors onward: they name sources, set links and refer users to the pages their answer is drawn from. AI referral traffic has grown rapidly since 2024, by around 16 times (Trakkr) according to one analysis, with roughly 1.13 billion (Trakkr) referral visits per month by now. ChatGPT accounts for by far the largest share. For small and medium-sized businesses this creates a new, predictable channel that many competitors still overlook. This article shows why AI assistants now deliver visitors, how large the channel really is, why these visitors are especially valuable and with which content and structures a company ends up as a cited source in AI answers. It describes the implementation from a web agency's perspective and turns a trend into a concrete task for your own website.
Key takeaways
- AI assistants like ChatGPT became a real traffic source in 2026: they name sources, set links and refer users straight to the cited websites.
- AI referral traffic is growing fast, with ChatGPT accounting for about three quarters of all AI referrals. The channel is still small, but the growth is substantial.
- Visitors from AI search arrive with clear intent and, in our experience, buy more often than visitors from other channels.
- Only those who get cited become visible: generative engine optimisation ensures AI systems choose your content as a source.
- Clear structure, concrete facts with figures and original data raise the chance of being named in AI answers.
- Anyone who measures AI traffic without cookies spots the new channel early and can grow it deliberately instead of overlooking it.
Why AI Assistants Now Send Visitors
Not long ago AI assistants were seen as a dead end for traffic: they summarised knowledge and kept users to themselves. That has turned around. The major assistants now work with live access to the web, they search current pages, form an answer from them and openly state the sources used, usually with a clickable link. Anyone who wants to know where a statement comes from, or who wants to dig deeper, clicks onward exactly there. The supposed dead end has become a junction that guides people with a concrete need to the right website.
The process resembles classic search yet differs fundamentally. Instead of a list of ten blue links, the AI delivers a finished answer and bases it on a few selected sources. For companies this means it is no longer the ranking position alone that decides, but whether your page is drawn on as evidence at all. The same shift shows in the AI answer boxes of classic search, the AI Overviews: there too only a few pages are cited, and whoever is not among them loses visibility. The technical foundation for machines to read and classify a page cleanly is laid, among other things, by structured data following the Schema standard.
Search gains a second stage
How Large the Channel Really Is
AI traffic starts from a small base but grows unusually fast. According to an analysis by Similarweb, the share of AI traffic in the total web doubled roughly every two months since September 2024 and rose from around 0.02 percent (Similarweb) in 2024 to about 0.15 percent (Similarweb) in 2025, more than sevenfold. The five largest AI surfaces together accounted for about 25.77 billion (Similarweb) visits between January and April 2026. In May 2026 alone, ChatGPT's referral traffic rose by 36.7 percent (SE Ranking) over the previous month, according to an analysis of over 101,000 websites (SE Ranking).
The market split is clear: in 2026 ChatGPT accounted for around 76 percent (SE Ranking) of all AI referral traffic and, on another count, delivers about 92 percent (Similarweb) of standalone AI referrals. Measured against total visits, AI traffic still makes up only about 1 percent (Similarweb), yet the trend points steeply upward. For small and medium-sized businesses this is precisely the opportunity: the channel is not yet crowded, and whoever becomes visible as a source now secures a lead that would later be expensive to buy. This early positioning belongs in a coherent online marketing from one source that thinks of search and AI search together.
Fast growth
AI traffic doubled at times roughly every two months (Similarweb). A channel growing like that deserves attention now, not only once it is large.
ChatGPT dominates
Around three quarters of AI referrals come from ChatGPT in 2026 (SE Ranking). Anyone taking the channel seriously aligns content to the largest answer machine first.
Still a small share
About one percent of all visits so far comes from AI (Similarweb). That is exactly what makes an early entry cheap, because competition for citations is still manageable.
Why AI Visitors Are More Valuable
A one percent share sounds small, but the quality of these visitors lifts the channel well above its sheer volume. Anyone arriving via an AI assistant has usually already sharpened their need: the AI has pre-sorted the options, answered open questions and sent the user to the page with clear intent. This shows in the numbers. According to Similarweb, visitors from ChatGPT bought in e-commerce at around 11.4 percent (Similarweb) conversion, versus 5.3 percent (Similarweb) from organic search, more than twice as often. Adobe reported for May 2026 a rise in retail traffic from AI sources of 138 percent (Adobe Analytics) year on year, combined with longer time on site, more page views and more revenue per visit.
An analysis by Adobe points the same way, finding that buyers referred from AI assistants completed a purchase around 42 percent (Adobe Analytics) more often than visitors from classic channels. The reason is obvious: the AI provides advice before the click, so ill-fitting prospects get through less often and the remaining visitors match the offer better. For shops, well-qualified traffic also often means more deliberate purchase decisions, which can have a favourable effect on metrics such as the return rate in the online shop. The AI channel therefore brings not just visitors, but the right ones.
Why intent runs so high
Becoming a Cited Source in AI Answers
For an AI assistant to refer to your page, it must select it as evidence. Influencing that selection deliberately is the job of generative engine optimisation, GEO for short. Unlike classic search engine optimisation, which aims at good placements, GEO works towards being cited and woven into the generated answer. The central metric here is Share of Model: it measures how often a brand is named in AI answers to relevant questions, relative to the competition. Anyone wanting to improve this figure combines proven SEO craft with a structure that makes citing easy for machines.
Which factors favour citation is now well described. According to analyses from GEO practice, what counts above all is content comprehensiveness, clear structure, verifiable facts with concrete figures, credible sources and consistency across several platforms. Original data and original content are especially effective: whoever publishes a figure, a comparison or a guide that exists nowhere else gives the AI a reason to cite precisely this page instead of one of many similar ones. Solid ground for this is laid by the basics of search engine optimisation on which GEO builds.
Facts with figures
Concrete, sourced figures and clear statements are cited more often than vague wording. A verified statistic with a source is a strong signal for the AI.
Clear structure
Meaningful headings, short paragraphs and direct answers to real questions help the AI recognise and lift the right passage.
Original data
Original content such as your own analyses, comparisons or guides sets you apart and gives the answer machine a reason to name precisely you.
- Answer key questions directly in the first sentence of a section before you elaborate
- Back up statements with concrete figures and name the source immediately afterwards
- Structure content with clear headings and short, self-contained paragraphs
- Publish original data, comparisons or guides that exist nowhere else
- Keep core statements consistent across website, profiles and directories
- Refresh load-bearing articles regularly and show the date visibly
Content and Structure That AI Cites
Not every type of content gets cited equally often. What works well are comprehensive guides, clearly structured FAQ pages, glossaries, comparison articles and content with concrete statistics, that is formats that answer a question fully and comprehensibly. It is also striking that AI systems frequently refer to community content and to video platforms. Moving images therefore pay off twice, because a good explainer video strengthens both citability and conversion, as a well-placed video on the website shows. What matters is that every article covers a clear topic cleanly rather than merely touching on many.
Structure often decides whether the AI finds the right passage. A page that places direct answers under precise headings is easier to cite than a dense block of text without structure. Add to that machine readability: structured data, clean HTML and a fast, technically well-kept page help the systems capture content correctly. Anyone who combines this basis with well-considered content marketing as a strategy for small and medium-sized businesses builds, over time, a library of citable content. And because AI answers thrive on clear, user-oriented language, the same principles apply as for website copy that sells: benefit first, plain language, one thought per paragraph.
- Comprehensive guides that cover a topic fully and comprehensibly
- FAQ pages with real questions and concise, direct answers
- Comparisons and decision aids that weigh options objectively
- Glossaries and definitions that clarify terms unambiguously
- Articles with concrete, sourced figures instead of vague statements
- Explainer videos that bring complex points across vividly
No promises about AI rankings
Measuring and Steering AI Traffic
A channel you do not measure stays invisible, even as it grows. AI referrals can usually be recognised by the traffic source: when a user clicks from an AI answer onto your page, the assistant's domain appears in the origin data. Whoever captures these referrals cleanly sees in black and white how much traffic comes from AI, which pages get cited and how these visitors behave. It is important to proceed with data minimisation here; a website analysis without a cookie banner delivers the necessary metrics without burdening visitors with consent dialogs and without collecting unnecessary personal data.
Measurement turns into steering: if a certain page regularly receives visitors from AI, it is worth expanding precisely that topic and adding related questions. If important topics remain uncited, that is a hint to improve structure, facts or currency. This creates a cycle of publishing, measuring and refining. First effects often show quickly: according to analyses from GEO practice, many on-page measures already take effect within two to four weeks (Search Engine Land) after the next crawl.
| Feature | Classic organic search | AI search and assistants |
|---|---|---|
| Presentation | A list of several links | One answer with a few sources |
| Visibility | Position in the ranking | Cited as a source or not |
| Visitor intent | Mixed, often at the start | Usually sharpened and purchase-near |
| Conversion | Solid baseline | Higher in our experience |
| Lever | SEO and content | GEO, structure and original data |
| Measurement | Established | Captured via the traffic source |
Think of both channels together
AI search does not demand a completely new website but one that answers clear questions clearly, backs up its statements and is cleanly structured so that people and machines understand it alike. This is exactly the work we take on together with you: from search engine optimisation with an eye on AI search through citable content in content marketing to clear fixed prices for setup and support. That turns a trend everyone talks about into a channel that brings you qualified visitors.