2026-07-04
How Much Traffic Does Each AI Engine Actually Send? (2026 Data)
AI engines send only about 1% of your traffic today — and ChatGPT alone accounts for roughly 60% of that. But each AI visitor is worth several times an average one, and the channel is doubling year over year.
The short version
- AI referral traffic is tiny — roughly 1% of all website visits[1] — while organic search still drives 33–53% depending on the study[1][10]. Don't panic about the size yet.
- ChatGPT dominates the referrals. Across cross-source B2B data in early 2026, ChatGPT sent around 62.6% of AI referral traffic, Claude ~18.5%, Gemini ~10.6%, Perplexity ~7.3% and Copilot ~4%[3].
- Usage and referrals are two different rankings. By visit-share among chatbots themselves, ChatGPT sits near 53.9%, Gemini ~27.9% and Claude ~9.2%[5] — Gemini is used far more than it refers.
- It's growing at a violent pace. Adobe tracked AI-referred retail traffic up more than 1,300% since October 2024[7], and +393% year over year in Q1 2026[6].
- And it converts. AI referral traffic converts roughly 4.4x higher than organic[8] — yet most of it hides as "direct" or "(not set)" in your analytics[9], so you're probably undercounting it.
How big is AI traffic, really?
Small. Right now, AI engines account for roughly 1.08% of all website visits[1], while organic search still sends the lion's share of traffic — over 33% of visits in the same analysis[1] and as high as ~53% in broader channel studies where it remains the single dominant source[10]. So if you're judging AI purely on volume today, it barely registers next to Google.
But that flat 1% hides two things. First, the growth curve is close to vertical — the same channel was a rounding error 18 months ago. Second, and more important, an AI-referred visitor is not an average visitor: this traffic converts far above the site average[8]. So "1% of visits" massively understates its share of revenue. Size is the wrong metric here.
Which AI engines send the most traffic?
This is the question that actually matters for your acquisition mix: of the visits AI sends to websites, which engine is the referrer? Across cross-source B2B data from March–April 2026, one engine runs away with it. ChatGPT sent around 62.6% of all AI referral traffic, with Claude a distant second at ~18.5%, Gemini ~10.6%, Perplexity ~7.3% and Microsoft Copilot around 4%[3]. Treat these as directional — the exact split shifts by dataset, audience and month — but the shape is consistent everywhere: ChatGPT is the firehose, and everyone else is a stream.
Read that carefully: this is referral traffic share, not popularity. It answers "when a person lands on a website from an AI tool, which tool did they come from?" — nothing more. The next section is a completely different number, and mixing them up is the most common mistake in every AI-traffic report you'll read this year.
Which AI tools people actually use
Now flip the lens from referrals to usage. By share of visits to the chatbots themselves — measured across the seven biggest generative-AI chatbots in May 2026 — ChatGPT held about 53.9%, Gemini roughly 27.9% and Claude around 9.2%[5][4]. In that same window ChatGPT slipped below the 65% mark it once held while Gemini pushed past 20%[5] — Google's assistant is gaining real ground on raw usage.
Here's why this matters and why you can't collapse the two charts into one. Usage share (this section) is not the same as referral share (the last one). Gemini is used almost three times as much as Claude[5], yet in the referral data Claude appears to send more traffic to sites than Gemini does[3] — and since those two figures come from different datasets, read it as a directional signal, not a precise ranking. Why? Because how often a tool links out to the open web — and how often users click those links — varies enormously by engine and by task. A tool can be wildly popular and still be a poor referrer if it answers everything inside the chat. So: use the usage chart to understand where mindshare is heading, and the referral chart to plan where your website traffic will actually come from. They are different jobs.
How fast is AI traffic growing?
The flat 1% only makes sense once you see the slope behind it. Adobe's retail data shows AI-referred traffic up more than 1,324% since it began tracking in October 2024[7], growing 138% year over year as of May 2026[7]. Zoom into Q1 2026 and it grew 393% year over year, after peaking at an eye-watering +1,151% in December 2025[6]. On the engine side, ChatGPT's web visits rose about 84% between September 2024 and March 2026, while Gemini grew roughly 9x and Claude around 770% over comparable windows[5].
The baseline
Adobe starts tracking AI-referred retail traffic. This is the "zero" every later percentage is measured against[6].
The surge
AI referrals climb steadily as chatbots add web browsing and citations, and users start clicking through to sources[6].
The peak
AI-referred retail traffic spikes +1,151% in December, then holds a +393% year-over-year pace across Q1 2026[6].
Roughly doubling, year on year
Momentum continues — AI-referred retail traffic has roughly doubled year over year as of May 2026[7]. Still ~1% of the total, but not for much longer at this rate.
Compounding at these rates, a channel that's 1% today does not stay 1% for long. The brands treating it as noise are the ones who'll be caught flat-footed when it becomes 5–10% of revenue.
Tiny, but mighty
Now the payoff — and it shows up two different ways, measured against two different baselines. Compared to organic traffic, AI referrals convert roughly 4.4x higher[8]; visitors arrive having already researched inside the assistant, so they land closer to a decision. Compared to the average site visitor, Adobe's retail data shows AI-referred visitors with a 54% higher conversion rate, 23% more pages viewed and 53% longer time on site[7]. Different comparison groups, same direction: this traffic punches far above its weight. One SaaS analysis makes the point bluntly — AI platforms drove 12.1% of signups from just 0.5% of traffic[2], a 23x higher conversion rate.
| Metric | AI-referred visitor | Standard visitor |
|---|---|---|
| Conversion rate | +54%[7] | baseline |
| Pages per visit | +23%[7] | baseline |
| Time on site | +53%[7] | baseline |
| Overall conversion vs. organic | ~4.4x[8] | 1x |
AI-referred visitors vs. standard visitors, retail data, 2026. Source: Adobe via Digital Commerce 360[7] & Emarketed[8].
Put the two facts side by side. This is your smallest traffic source by volume and, per visit, one of your most valuable. That's exactly the profile you want to be catching early — and exactly the profile analytics tools are worst at seeing.
The catch: most of it is invisible in your analytics
Here's the twist that makes every number above an undercount. A large share of AI traffic doesn't arrive with a clean referrer, so it gets bucketed as "direct" or dumped into "(not set)" in your reports. In an analysis of 181.6 million GA4 sessions, Workshop Digital found that 22% of ChatGPT sessions and 32% of Perplexity sessions landed in the "(not set)" bucket in GA4[9] — invisible as AI traffic unless you go looking.
The consequence is predictable: brands look at a dashboard that says "AI = almost nothing," conclude it doesn't matter, and underinvest in the highest-intent visitors they have. If a third of your Perplexity traffic is misattributed, you're not measuring a 1% channel — you're measuring a fraction of it and calling that the whole. The real number is bigger than your dashboard admits.
What to do about it
Two jobs, and they're separate. First, make sure AI engines can read and cite you at all — because you can't get referral traffic from an answer you're not in. That means AI crawlers can access your site, your key facts are in structured data, and your pages are machine-readable; you can check all three in a couple of minutes with our free GEO checker. Second, make sure your analytics can actually see AI traffic when it arrives — otherwise you'll keep undercounting your best channel. Confirm your tags fire correctly with the tracking checker, and use the UTM builder to keep the sources you can control clean, so AI-referred visits don't silently collapse into "direct."
Do those two things and you flip from guessing to knowing: you become citable, and you can prove the traffic and revenue it drives. Everything else — content, KPIs, budget — follows from having that visibility.
The bottom line
AI engines send about 1% of your traffic today, and ChatGPT sends most of that 1%[1][3]. But that number is small, growing triple digits year over year[6], and made of visitors worth several times an average one[8]. Judge it by volume and you'll ignore it; judge it by value and growth and you'll get in early. And whatever you do, don't trust the raw dashboard — a big slice of this traffic is hiding in "(not set)"[9], which means the opportunity is even larger than it looks.
FAQ
Sources
- Analyze — AI traffic research (citing Conductor 2026 benchmarks): AI referral ≈1.08% of visits; organic search over 33% of visits, 2026. tryanalyze.ai
- Ahrefs — AI chatbot traffic: AI search visitors were 0.5% of total visitors but drove 12.1% of signups (a 23x higher conversion rate), 2026. ahrefs.com
- Goodie — AI search traffic report 2026 (referral share by engine, cross-source B2B). higoodie.com
- DigitalApplied — AI platform traffic share 2026 (reporting Similarweb visit-share: ChatGPT ~52.7%, Gemini ~27.3%, Claude ~8.9%), corroborates the usage ranking. digitalapplied.com
- Similarweb — generative-AI stats: chatbot visit-share (May 2026) and web-visit growth by engine, 2026. similarweb.com
- Adobe (via Search Engine Journal) — 2026 Q2 AI traffic report: +393% YoY in Q1 2026, peak +1,151% YoY in December 2025. searchenginejournal.com
- Adobe (via Digital Commerce 360) — AI-referred retail traffic up +1,324% since Oct 2024 and +138% YoY in May 2026; AI-visitor lift: +54% conversion, +53% time on site, +23% pages per visit. digitalcommerce360.com
- Emarketed — AI referral traffic converts ≈4.4x higher than organic, 2026. emarketed.com
- Workshop Digital — The AI Referral Gap: 22% of ChatGPT and 32% of Perplexity sessions land in "(not set)" (181.6M GA4 sessions, 22 clients, full-year 2025). workshopdigital.com
- BrightEdge — Organic channel share expands to ~53% of all trackable website traffic (largest digital channel), 2026. brightedge.com