Does AI Assistant Traffic Convert Better? What the Published Data Shows

Last checked: October 27, 2026. Every figure below links to where it was published.

You will find the claim everywhere: visitors sent by ChatGPT, Perplexity or Gemini "convert better than any other channel". Some posts put a number on it, often a conversion rate of 30% or more, usually without a link. This article looks at what has actually been published, what it measures, and what it does not.

The one regular data series: Adobe, US retail

The most cited numbers come from Adobe Analytics, which reports on visits to US retail websites that use its product.

Early 2025, AI traffic converted worse. In March 2025 Adobe reported that traffic from generative AI sources to US retail sites had grown 1,200% between July 2024 and February 2025. Those visitors stayed 8% longer, viewed 12% more pages and bounced 23% less than other visitors, but they were 9% less likely to convert. Adobe noted the gap had narrowed from 43% in July 2024.

By mid-2026, the gap had reversed. For July 2026, Adobe's figures as reported by Digital Commerce 360 (August 19, 2026) show AI-referred visits converting at a rate 60% higher than non-AI traffic, the eleventh month in a row of AI traffic converting better, with 53% more revenue per visit. AI-referred traffic to US retail sites was up 62% year over year.

So the honest summary is: in one large retail panel, AI assistant visits went from converting noticeably worse than other traffic to converting noticeably better within about eighteen months.

What those numbers do not say

They are relative, not absolute. "60% higher than non-AI traffic" means that if other visitors convert at 2%, AI visitors convert at about 3.2%. It does not mean 60% of AI visitors buy. A claim of a "30 to 40% conversion rate" for AI traffic does not come from this data, and we did not find a published study that supports it.

They are US retail. Software trials, lead generation, publishing and B2B sales have different funnels. A retail result tells you little about whether a SaaS signup from Perplexity will turn into a paying customer.

The baseline is a blend. "Non-AI traffic" mixes paid search, email, social, direct and organic. A channel can beat that average while still converting below your best channel.

One vendor, one panel. The figures cover sites that use Adobe Analytics, and the July 2026 numbers reached us through press coverage rather than a public report we could read in full.

Why AI visits might convert better

Adobe's figures say what happened, not why. The usual explanation is plausible but unproven: an assistant visitor described their need in a conversation, received a short list, and clicked a recommendation, so part of the comparison shopping happened before the visit. That would also explain why engagement was higher even in early 2025, when conversion was not.

It is a hypothesis to test on your own data, not a law.

How to check it on your own site

The only conversion rate that matters for your decisions is yours.

1. Separate the channel. GA4 has had a native AI Assistant channel since May 2026 (Google's definition); most privacy-first tools now classify AI assistants too. Check that ChatGPT, Perplexity, Claude and Gemini each appear under their own source ↗.
2. Compare like with like. Put the AI channel next to organic search and email, on the same goal and the same period, rather than against "all traffic".
3. Go past the first conversion. A signup is not revenue. Tie the visits to payments, and for subscriptions to renewals, before deciding the channel is worth the effort ↗.
4. Mind the volume. On most sites AI referrals are still a small number of visits, so a conversion rate computed on a few dozen visits moves a lot from one month to the next.

FAQ

Does ChatGPT traffic convert better than Google search? For US retail sites in Adobe's panel, AI-referred visits converted better than non-AI traffic as a whole through 2026, after converting worse in early 2025. Whether that holds for your site and against your organic search specifically is something only your own data can tell you.

Is it true that AI traffic converts at 30 to 40%? We found no published source for that figure. The data that exists compares AI traffic with other traffic (for example, 60% higher in Adobe's July 2026 retail figures), which is a relative difference, not a conversion rate.

Why does some ChatGPT traffic show up as Direct? Clicks from mobile apps can arrive without a referrer, and people who read an answer often visit later by typing your name. Referrer-based counts of AI traffic are therefore a floor.