How Much Traffic Do AI Assistants Actually Send, And Does It Convert Better?

AI Referral Traffic Benchmark| 12 months of data from 95 websites, September 2025 to August 2026.

Summary

• Organic search delivered 39 times more traffic than AI assistants. It remains the dominant discovery channel by a wide margin.
• For every 100 conversions organic search delivered, AI referrals delivered 3.2.
• AI visibility and organic performance move together: the websites gaining AI referral traffic were the websites whose organic performance was improving.
• AI referral traffic converted a fifth better than organic search, at 7.18% against 6.08%. However, B2B product businesses were the exception where AI traffic converted at 0.62 times the organic rate.
• AI traffic converted worse than organic on 23 of 63 websites.
• AI referral traffic converts worse against paid search (0.67x) and better against direct traffic (2.48x).

The full dataset is available to download at the end of this page. No email required.

What we measured

We analysed 95 websites, covering the 12 months from September 2025 to August 2026. All figures below use the 95 properties with 12 complete months, narrowing to 63 for conversion comparisons where a property had at least 100 AI referral sessions and 1,000 organic sessions in the period.

Organic search delivered 39 times more traffic than AI assistants

Across the period, the websites received 2,825,594 organic sessions and 72,095 AI referral sessions. Organic search delivered 39 times more traffic. Expressed the other way round, AI assistants sent 2.6% of the traffic that organic search sent.

Whatever has changed in the past year, organic search remains the dominant discovery channel by a very wide margin. There was an uplift in AI referral traffic from May 2026, which was when ChatGPT began linking brand names more prominently.

Many published figures express AI traffic as a share of total sessions. We use organic as the denominator because total sessions moves with paid activity, email campaigns and direct traffic, none of which have anything to do with AI visibility. For example, two websites with identical AI performance can report very different numbers if one of them runs a large paid campaign. Organic is the closer comparison in every respect: unpaid, discovery-driven, and responsive to the same underlying work.

Month	AI sessions	AI as % of organic	Organic delivered
September 2025	5,397	2.25%	45x more
October 2025	6,328	2.47%	41x more
November 2025	6,108	2.76%	36x more
December 2025	4,588	2.57%	39x more
January 2026	5,026	2.12%	47x more
February 2026	4,792	2.01%	50x more
March 2026	5,270	2.12%	47x more
April 2026	4,980	2.14%	47x more
May 2026	7,198	3.06%	33x more
June 2026	7,439	2.76%	36x more
July 2026	6,618	2.62%	38x more
August 2026	8,351	3.86%	26x more

AI traffic converts about a fifth better than organic

AI referral sessions converted at 7.18%. Organic sessions converted at 6.08%. That is a ratio of 1.18x pooled across all sessions, and a median of 1.21x when each website is measured separately and the results ranked.

Averages hide a lot here. Line up all 63 websites from worst to best, and the middle half sit anywhere between converting slightly worse than organic and converting two and a half times better. On 23 of the 63, AI traffic converted worse rather than better.

Five of those websites recorded no AI conversions at all, despite receiving 1,487 AI referral sessions between them across the year.

Our figure of 1.18x contrasts greatly with figures of 4.4x and 23x that have been circulated in other studies (1). Those come from single companies measuring their own websites, or from studies of publisher and news sites measuring newsletter sign-ups against a very low search baseline. Neither is comparable to 63 commercial websites measuring enquiries or sales.

The baseline changes the answer

The conversion multiple depends enormously on which channel you compare against, for example paid search converts better than AI referrals.

Compared against	AI conversion ratio
Paid search	0.67x
Organic search	1.18x
Total sessions	1.37x
Direct	2.48x

Anyone quoting a conversion multiple for AI traffic without stating the comparison channel has published a number that cannot be checked or reproduced.

We use organic because it is the closest substitute. Direct traffic is contaminated with repeat customers, untagged email and bookmark traffic, all of which convert well for reasons that have nothing to do with how the visitor discovered the business.

AI delivered 3.2 conversions for every 100 from organic search

The higher conversion rate for AI referrals is real, but it needs to be looked at in context of the volume of conversions it delivers.

Across the websites, AI referrals produced 5,001 conversions against organic search’s 155,605. For every 100 conversions organic search delivered, AI referrals delivered 3.2.

Holding conversion rates constant and scaling sessions, this is what different improvements would have been worth per website per month:

Scenario	Additional conversions per website per month
Double AI referral traffic	+6.6
Triple AI referral traffic	+13.2
Grow organic search by 5%	+10.3
Grow organic search by 10%	+20.6
Double organic search traffic	+205.8

A 5% improvement in organic search performance would deliver more additional conversions than doubling AI referral traffic. Growing organic search by 3.2% would deliver the same number of additional conversions as doubling AI referral traffic. 

This is a snapshot based on the data. AI referral volume rose relative to organic in May 2026, and the arithmetic changes if that trend continues. It also excludes Google AI Overviews entirely, which arrive as ordinary organic sessions and cannot be separated in GA4, so the measured AI contribution should be read as a floor.

B2C converted at 1.51x, B2B products at 0.62x

Segmenting conversions by customer type produces an interesting pattern – AI traffic converts worse than organic traffic for B2B product businesses.

Customer type	Websites	AI referrals as % of organic	Pooled ratio	Median ratio	AI worse in	Range if any one site is removed
B2C	19	2.37%	1.51x	1.21x	7 of 19	1.23x to 1.94x
Mixed	10	2.08%	1.46x	1.56x	1 of 10	1.29x to 2.86x
B2B Services	24	2.90%	1.21x	1.14x	9 of 24	0.99x to 1.95x
B2B Products	10	3.40%	0.62x	0.84x	6 of 10	0.59x to 0.71x

Notably, B2B product websites also received the highest proportion of AI referral traffic relative to organic, at 3.40%. They got more of it, but it performed less well.

With only ten websites in the B2B Products group, the obvious risk is that one unusual website is producing the whole result. So we removed each website in turn and recalculated using the remaining nine, ten times over. The worst case still came out at 0.71x, comfortably below parity. No single website is responsible for the finding.

Six of the ten converted worse on AI traffic than on organic, so this is not one outlier dragging down an otherwise healthy average.

The gap narrowed in May 2026

Between September 2025 and April 2026, organic search delivered 44 times more traffic than AI assistants. From May 2026 onwards it delivered 33 times more. In percentage terms, AI referrals moved from 2.29% of organic sessions to 3.04%.

The rise appeared across every customer type: B2C +39%, B2B Services +36%, B2B Products +18%, Mixed +18%. A separate study by SERanking also noted the increase at 38.7% for UK sites (2).

Around 7 May 2026, ChatGPT began making brand names clickable inline within its answers rather than confining links to citations. We treat the connection between that change and the traffic rise as a working hypothesis rather than an established cause.

Organic growth and AI growth move together

The websites gaining AI referral traffic were the websites whose organic performance was improving.

Across 61 websites, the correlation between AI referral growth and organic growth was +0.557. Controlling for growth in the rest of the website’s traffic, so that the effect of a website simply getting bigger is removed, it was +0.534 (p < 0.0001, meaning the odds of this being a fluke are less than one in ten thousand). The relationship holds in absolute levels as well as growth rates: +0.730 raw, and +0.594 controlling for website size.

The most plausible reading is that the same underlying work, being a clearly defined business with consistent information and genuine authority, drives both. What the data does establish is that the two are not independent – to become stronger in AI, SEO visibility plays a large part.

The test that failed

To check that the control above has any force, we applied it to a second hypothesis.

The “billboard” theory holds that AI presence works like advertising: people see a brand in an AI answer, do not click, and arrive later by typing the address. If that were happening, websites gaining AI visibility should also gain direct traffic.

Test	Raw correlation	Controlled for rest-of-site growth
AI growth vs organic growth	+0.557	+0.534 (p < 0.0001)
AI growth vs direct growth	+0.234	-0.230 (p = 0.075)

Under the same control that leaves the organic relationship almost untouched, the billboard association does not merely disappear. It turns slightly negative, though not enough to be statistically meaningful. Websites that grew, grew everywhere; once that is accounted for, there is no evidence that AI growth drives direct traffic. We also found no relationship between AI session growth and direct conversion growth (+0.083, p = 0.55).

What the 12 months show

AI referral traffic is real but small. Organic search delivered 39 times more traffic than AI, and of the traffic these websites received from the two channels combined, 97.5% came from organic search. That gap is closing, but it is closing from a long way back.

AI visibility and organic performance move together. The websites gaining AI traffic were the same ones whose organic search was improving, and the relationship holds even after allowing for sites that were simply growing overall.

AI traffic converts better than organic, but by far less than existing figures in circulation suggest. Our 1.18x figure sits well below the 3x, 4.4x and 23x reported elsewhere. 

Conversion is not improved in all cases. On 37% of websites, AI traffic converted worse than organic, and for B2B product businesses it converted worse consistently.

Methodology

We collated data to track how much website traffic is now arriving from AI assistants (ChatGPT, Claude, Gemini, Perplexity, Copilot and similar) as a referral source, alongside traditional organic search, paid search, and direct traffic, sessions and conversions, broken down monthly, per GA4 property. 

Google’s own “AI Assistant” traffic category only exists from June 2026 onward, so it can’t show a proper trend on its own. To get a full 12-month view, we identified AI referral traffic by matching the actual referring domain (e.g. chatgpt.com, claude.ai) using the regex query Session source matches partial regex ‘bard|chatgpt|openai|claude|copilot|gemini|perplexity|grok’.

Sample

Stage	Websites
GA4 properties	95
Meeting volume thresholds	66
Of those, with recorded organic conversions, used for conversion figures	63

Eligibility for conversion comparison required at least 100 AI referral sessions and 1,000 organic sessions across the period.

How figures are calculated

Every percentage is derived once, from summed counts. We sum sessions across all websites in a group, sum conversions across the same group, then divide. We do not average the per-website rates, because a website with 200,000 sessions and one with 3,000 would then carry equal weight.

Where pooled and median figures differ materially we report both, because a divergence between them indicates that one large website is influencing the group.

Limitations

GA4 counts key events, not people. One person submitting a form twice registers as two conversions. Every conversion figure here is an upper bound on the number of real enquiries. The AI versus organic ratio is unaffected, because the same definition applies to both channels within each website.

Google AI Overviews are not included. Those clicks arrive as ordinary organic sessions and cannot be separated. The influence of AI Overviews on discovery is impossible to isolate in GA4.

Some AI referrals may not be counted. In-app browsers, stripped referrers, and the common pattern of reading an AI answer and then searching for the brand by name would not be tracked using our regex query.

Four months is a short post-May window. Any statement about the trend since May 2026 is provisional.

Download the data

The full aggregated dataset is available below. Website identities have been removed and replaced with an index number and customer type. No email address is required.

Download here

Footnotes

(1) https://ahrefs.com/blog/ai-search-traffic-conversions-ahrefs/ | https://www.semrush.com/blog/ai-search-seo-traffic-study/ | https://clarity.microsoft.com/blog/ai-traffic-converts-at-3x-the-rate-of-other-channels-study/ 

(2) https://seranking.com/blog/chatgpt-referral-traffic-may-2026/