AI Search Traffic Benchmarks 2026: How Much It Sends and How Well It Converts

AI search traffic benchmarks for 2026 show a small channel with a real conversion edge in retail. ChatGPT sent 0.32% of all website traffic in May 2026 (SE Ranking), all AI platforms together sent about 1.08% (Conductor), and Adobe found AI-referred retail visits converted 54% better than other traffic. Outside retail, the premium is unproven.

That paragraph is the whole story if you are short on time. The rest of this post is the evidence behind it, with every figure traced to a named source and a stated measurement window.

We pulled these numbers because the claims circulating about AI search are wildly inconsistent. One camp says AI is about to replace Google. Another says AI referrals are a rounding error. Both are partly right, and the gap between them is almost entirely explained by what each source measured, who it measured, and when.

Where sources disagree, we give the range instead of picking a winner. Where a figure traces to a single company's internal data, we say so and leave it out of our synthesis.

NAMEX 2026 AI Search Traffic Benchmarks

The table below covers how much traffic AI platforms send, who sends it, and how Google's own AI summaries change click behaviour. Every row carries a number, a source and a period.

How Big Is AI Referral Traffic? The NAMEX AI Referral Share Range

Small. That is the first thing the data says, and it holds across both large samples. SE Ranking, working from Google Analytics data on 101,574 consenting websites, put ChatGPT at 0.32% of total traffic in May 2026. Conductor, analysing 3.3 billion sessions across 13,770 domains, put every AI platform combined at 1.08%.

Those two figures measure different things over different periods, so they are not a contradiction. One is ChatGPT alone in 2026. The other is all AI platforms through September 2025. Put together they bracket the realistic answer for a typical site.

The NAMEX AI Referral Share Range: 0.3% to 1.1% of sessions, with a stated midpoint of 0.7%. It is built from the SE Ranking figure (0.32%) and the Conductor figure (1.08%). The midpoint is the simple average of those two values, not a measured number. Information Technology sites run well above the range at 2.80%, and Utilities and Communication Services sit below it at 0.35% and 0.25%.

Growth is where the sources genuinely disagree. SE Ranking found ChatGPT's share barely moved from January to April 2026, up only 1.5%, then jumped 36.7% in a single month to reach May's 0.32%. Similarweb's January 2026 data showed AI platform visits up 28.6% year-over-year while referral traffic to external sites stayed flat. Adobe, looking only at US retail, measured a 138% year-over-year rise in May 2026.

The reconciliation is plausible but unproven. People are using AI assistants much more, but most of those sessions end inside the assistant. Retail is the exception, because shoppers click through to buy. If your site is not a retail site, plan for the flat end of that range.

Two more details matter for planning. ChatGPT supplies 87.4% of AI referral traffic in Conductor's data, so this channel behaves like a single-platform channel. And SE Ranking found 60% of AI-referred visits land on the homepage, against 17% for organic search. AI sends people to your brand, not to your deep pages.

NAMEX 2026 AI vs Google Organic Quality Benchmarks

Volume is small, so quality is the argument for caring. This table sets AI-referred visitors against their comparison group on conversion and engagement. Read the sample column before the result column, because the comparison groups differ.

The engagement rows pull in opposite directions, and that is worth being direct about. Ahrefs' 2025 data across all site types shows AI visitors viewing fewer pages and bouncing more than search visitors. Adobe's 2026 retail data shows the reverse, with more time on site and more pages per visit. The differences are site type and year, and we cannot separate the two with the data available.

One figure we deliberately excluded: a 23x conversion advantage that circulates widely. It traces to Ahrefs' own internal data, not to a published, reproducible sample, so it cannot sit next to the others in a table.

The NAMEX AI Referral Conversion Premium

Three sources give us a usable conversion comparison: Adobe (1.54x, May 2026, US retail), Similarweb (7% against 5%, or 1.40x, transactional sites) and Opollo (14.2% against 2.8%, or 5.07x, IT services firms).

The NAMEX AI Referral Conversion Premium: 1.4x to 5.1x, with a planning figure of 1.5x for retail and transactional sites. The 1.5x comes from the midpoint of the two retail and transactional sources, 1.40x and 1.54x, which average 1.47x. The 5.1x upper bound is a single vendor sample of 312 B2B firms and should be treated as a ceiling you have not earned until your own data shows it.

Why plan against 1.5x rather than 5x? Because the premium is new and unstable. Adobe's own data shows AI-referred retail visits converting 38% worse than other traffic in March 2025, then 42% better a year later. A gap that swings by 80 points in twelve months is a moving target, not a constant.

Definitions also differ. Adobe measures retail purchases. Opollo counts qualified enquiries such as booked consultations and demo requests, and it excluded firms with fewer than 10 AI sessions a month and outliers above 35% conversion. Similarweb measures conversion on transactional sites. None of these is the same event, which is another reason to hold the synthesis loosely.

Not sure whether the AI Assistant traffic on your own site is converting? Our free audit breaks out AI referrals from organic and direct traffic so you can see the real numbers before you change any budget.

What These Benchmarks Mean for GEO and SEO Budgets

At 0.3% to 1.1% of sessions, AI referrals are not yet a reason to move budget on their own. They are a reason to measure, and to make small, cheap changes to how your content is written.

The first change is measurement. Set up AI Assistant as its own channel in GA4 and track its conversions separately for 90 days. Our guide to average conversion rate by industry in 2026 gives you the baseline to compare against, and our breakdown of attribution models compared explains why last-click will understate any channel that sends people to research first and convert later.

The second change is structure. AI assistants quote self-contained answers, so each section should open with a direct two-sentence answer that makes sense when lifted out of context. If you are new to the topic, start with what generative engine optimization is. Because 60% of AI-referred visits land on the homepage, the homepage needs to say clearly who you serve and what you do.

The third point is about not overcorrecting. Google AI summaries cut clicks on traditional results from 15% of visits to 8% in the Pew data, which is a real headwind. But you recover from that with better titles and answer blocks on the pages you already rank, which is search work, not a new channel. Our search team handles exactly that. For the wider picture on where budgets are moving, see digital ad spend by channel in 2026.

If you are weighing AI search against paid channels, the cost side matters too. Our Q2 2026 CPC and CPM benchmarks show what a comparable visit costs through paid media, and our targeting and measurement services cover how we connect those costs to outcomes.

What NAMEX's Own Site Shows

We checked our own GA4 property for the 28 days from 3 to 30 September 2026 (ET). The site recorded 670 sessions. AI Assistant sent 16 of them, or 2.4%, and 7 of those 16 were engaged sessions (44%). Organic Search sent 102 sessions, with 50 engaged (49%).

Direct traffic was 527 sessions with only 69 engaged, a 13% engagement rate that looks like crawler noise. Because that inflates the total, AI Assistant's share of real human traffic is probably higher than 2.4%.

Sixteen sessions is a tiny sample. It is consistent with the 0.3% to 1.1% range being a floor for a B2B marketing site, and it says nothing reliable about conversion. We report it because a benchmark post should show its owner's numbers, not only other people's.

Methodology

This analysis draws on seven sources retrieved on 1 October 2026. Traffic share comes from SE Ranking's ChatGPT referral traffic analysis, published 9 July 2026, covering 101,574 websites from January 2025 to May 2026 using Google Analytics data from consenting sites, and from Conductor's 2026 AEO and GEO Benchmarks Report, covering 13,770 domains and 3.3 billion sessions from May to September 2025. Retail growth and conversion figures come from Adobe Analytics, whose Q1 2026 AI traffic report was published 16 April 2026 and whose May 2026 figures were reported by Digital Commerce 360 in June 2026, both covering more than 1 trillion visits to US retail sites.

Transactional-site conversion and session figures come from Similarweb's 2026 Generative AI Brand Visibility Report, using US data from January 2026. IT services conversion figures come from Opollo's 2026 AI Search Benchmark Report, published 22 February 2026, covering 312 firms in North America, Australia and the UK from January 2025 to January 2026. Engagement figures across all site types come from Ahrefs' AI traffic quality study, published 24 June 2025, covering 81,947 websites in May and June 2025. Click behaviour on Google AI summaries comes from Pew Research Center, published 22 July 2025, covering 900 US adults and 68,879 searches in March 2025.

Our own site figures come from GA4 for the 28 days to 30 September 2026 (ET). The NAMEX AI Referral Share Range and the NAMEX AI Referral Conversion Premium are our own syntheses of the sources above; the midpoints are stated band midpoints, not measured values. Source definitions of conversion differ and are noted where they matter.

Cite this data

Cite this data: North American Media Experts, "AI Search Traffic Benchmarks 2026: How Much It Sends and How Well It Converts" (2026). https://www.namediaexperts.com/blog-posts/ai-search-traffic-benchmarks-2026-share-conversion-click-data

Frequently asked questions

How much website traffic comes from AI search in 2026?

Less than most headlines imply. SE Ranking measured ChatGPT at 0.32% of total website traffic across 101,574 sites in May 2026, and Conductor measured all AI platforms together at 1.08% across 13,770 domains for May to September 2025. IT and technology sites sit higher, at 2.80% in the Conductor data.

Does AI search traffic convert better than Google organic?

For retail and transactional sites, yes, by a moderate margin. Adobe measured a 54% conversion advantage for AI-referred retail visits in May 2026, and Similarweb measured 7% against 5% for ChatGPT versus Google on transactional sites. The 5x premium seen in one B2B services study rests on a single vendor sample and should be treated as unproven.

Do Google AI Overviews reduce clicks to websites?

Yes. In Pew Research Center's March 2025 browsing study of 900 US adults, users clicked a traditional result link in 8% of visits when an AI summary appeared, against 15% when it did not. Only 1% of visits included a click on a link inside the AI summary itself.

What share of AI referral traffic comes from ChatGPT?

Conductor found that 87.4% of AI referral traffic across ten industries came from ChatGPT in its May to September 2025 data. That concentration is why most AI referral benchmarks are effectively ChatGPT benchmarks, and why a drop in ChatGPT referrals moves the whole channel.

Should we shift ad or SEO budget toward AI search in 2026?

Not on traffic volume alone. At roughly 0.3% to 1.1% of sessions, AI referrals are too small to justify a budget shift by themselves. The case rests on conversion quality, so start by measuring AI Assistant sessions and conversions separately in GA4 for 90 days before reallocating spend.

Where to take this next

If you want a second opinion on whether AI search is worth chasing for your business, book an intro call with Ryan and we will go through your numbers together. If you would rather see the data first, request a free audit and we will send back a written read on where your traffic and spend are actually going. Questions before either: get in touch.