# AI traffic converts better but is 3% of Google's: how to read the 2026 data

> Four October 2026 studies (Haynes, SEO Works, Adobe, Stradiji): AI sends 1-4% of organic visits but converts better. How to read the numbers in your own GA4 account.

URL: https://geosnap.ai/en/blog/traffico-ai-converte-di-piu-ma-vale-il-3-per-cento-di-google-come-leggere-i-dati-2026
Language: English
Version IT: https://geosnap.ai/blog/traffico-ai-converte-di-piu-ma-vale-il-3-per-cento-di-google-come-leggere-i-dati-2026
Author: Rinald Sefa, CMO Geosnap · Category: Research and data
Publisher: Geosnap (Maind Group S.r.l.), https://geosnap.ai

## In short

Four studies published in October 2026 measure traffic from ChatGPT and other AI assistants in GA4: it is worth between 0.62% and 3.9% of Google organic sessions, but it converts better (1.18 to 4.5 times depending on the study; 90% better in US retail according to Adobe). Total conversions still come almost entirely from Google (96.5%), reports undercount AI after 9 October, and the comparison has to be made site by site, against several baselines, with at least 100 AI sessions.

- ChatGPT sends 0.62% of Google organic sessions across 100 sites and 145 million sessions; on 95 UK sites organic is 39 times larger, with the AI share rising from 2.25% to 3.86% in a year.
- AI traffic converts better: 13.8% vs 3.02% (Haynes), 7.18% vs 6.08% (SEO Works), a 1.64x median (Stradiji); in US retail Adobe measures +90% conversion and +83% revenue per visit.
- The baseline matters: the same traffic is worth 1.18x organic but 0.67x paid search and 2.48x direct; on B2B product sites it converts at 0.62x.
- Since 9 October 2026 GA4 files about 46% of ChatGPT visits as direct: every later study undercounts the AI share.
- Compare the AI rate with the same site's organic rate, after at least 100 AI sessions, and measure presence in answers too, not only clicks.

Written by the Geosnap AI agent, reviewed and approved by Rinald Sefa.

Translated from the Italian original. [Read the original](https://geosnap.ai/blog/traffico-ai-converte-di-piu-ma-vale-il-3-per-cento-di-google-come-leggere-i-dati-2026)

In October 2026 four studies came out within days of each other, all measuring the same thing in Google Analytics 4: how many visits ChatGPT and the other AI assistants send, and how well those visits convert compared with Google organic search. The numbers agree on two points that look contradictory: AI traffic is a small fraction of organic traffic, and people who arrive from an AI answer convert better. Here are the figures, each study's limits and a way to read them in your own account without drawing the wrong conclusion.

## How many visits does AI really send compared with Google?

Between 1% and 4% of organic sessions, depending on the sample. The largest study is Marie Haynes', reported by [Search Engine Roundtable on 22 October 2026](https://www.seroundtable.com/ai-search-studies-42359.html): 12 months of GA4 data across 100 sites, more than 145 million sessions. ChatGPT sent 0.62% of the volume Google organic did. UK agency The SEO Works measured [95 GA4 properties from September 2025 to August 2026](https://www.seoworks.co.uk/downloads/ai-referral-traffic-benchmark/): 2,825,594 organic sessions against 72,095 from AI assistants, so organic delivered 39 times more. The share is rising, though: AI sessions were 2.25% of organic in September 2025 and 3.86% in August 2026. Across [35 sites, 22 of them selling in Turkey](https://www.stradiji.com/is-geo-a-distraction-ai-traffic-converts/), the median was 1.29% of organic sessions, rising to 3.6% for sites selling internationally.

| Study | Sample | AI traffic vs Google organic |
| --- | --- | --- |
| Marie Haynes (via Search Engine Roundtable) | 100 sites, 145 million sessions, 12 months | ChatGPT: 0.62% |
| The SEO Works | 95 UK sites, Sep 2025 - Aug 2026 | 2.6% (organic 39 times larger) |
| Stradiji | 35 sites, Oct 2025 - Sep 2026 | median 1.29%; 3.6% on international sites |
| Adobe (US retail) | over 1 trillion visits | share not disclosed; +71% year on year in September 2026 |

Adobe's figure is a different kind of number: it does not say how much AI weighs, it says how fast it grows. In its [29 October 2026 report](https://business.adobe.com/blog/ai-traffic-higher-revenue-per-visit-retail-holiday), AI traffic to US retail sites grew 71% compared with September 2025, and Adobe forecasts +159% on Thanksgiving and +95% on Black Friday. A small channel that doubles every year stays small for a while, but not forever.

## Does AI traffic convert better? Yes, but it depends what you compare it with

All four studies say yes, with one important exception. In Haynes' data [ChatGPT visitors converted at 13.8% against 3.02% for Google organic](https://www.seroundtable.com/ai-search-studies-42359.html), more than 4.5 times. The SEO Works finds a smaller gap: [7.18% against 6.08%, a ratio of 1.18](https://www.seoworks.co.uk/downloads/ai-referral-traffic-benchmark/) (median 1.21), and on 23 of 63 sites AI converted worse than organic. On 10 sites with reliable tracking, Stradiji measures [a median of 1.64 times](https://www.stradiji.com/is-geo-a-distraction-ai-traffic-converts/), with AI converting better on 8 of the 10. Adobe, on retail, says [AI traffic converted 90% better than the rest in September 2026](https://business.adobe.com/blog/ai-traffic-higher-revenue-per-visit-retail-holiday), with 73% more time on site, 25% more pages per visit and 83% higher revenue per visit. A year earlier, in August 2025, it converted 9% worse.

The point that is easy to miss is the baseline. In the same SEO Works sample, [AI traffic converts at 1.18 times organic, 0.67 times paid search, 1.37 times the site average and 2.48 times direct traffic](https://www.seoworks.co.uk/downloads/ai-referral-traffic-benchmark/). The same conversions become "converts better" or "converts worse" depending on which channel you put next to them. And the sector matters: on B2B product sites AI converts at 0.62 times organic, on B2C sites at 1.51.

| Type of business (The SEO Works) | Sites | AI conversion rate vs organic |
| --- | --- | --- |
| B2C | 19 | 1.51x |
| Mixed | 10 | 1.46x |
| B2B services | 24 | 1.21x |
| B2B products | 10 | 0.62x |

## Why almost all conversions still come from Google

Because a high rate on a small volume produces few absolute numbers. In Haynes' sample [Google organic drove 96.5% of the conversions analysed](https://www.seroundtable.com/ai-search-studies-42359.html), 27 times more than all large language models combined. The SEO Works puts it even more concretely: [for every 100 conversions from organic, AI referrals delivered 3.2](https://www.seoworks.co.uk/downloads/ai-referral-traffic-benchmark/). Doubling AI traffic would add 6.6 conversions per site per month; growing organic by 5% would add 10.3. That is where Haynes' title "GEO is a distraction" comes from, and the Search Engine Roundtable headline about an investment that is not worth making.

The opposite reading, argued by Mike King of iPullRank, does not dispute the numbers but the category: [measuring AI search as a performance channel makes it look small](https://ipullrank.com/ai-search-is-a-branding-channel) because an AI answer works like earned media, not like a paid click. The Turkish study lands on the same working rule: ["measure it like a conversion channel, budget it like a brand channel"](https://www.stradiji.com/is-geo-a-distraction-ai-traffic-converts/). Both readings start from the same data; what changes is the question you ask of it.

## How much AI traffic is missing from the reports?

A share none of the studies can measure, and one that probably grew in October. The SEO Works states its own limits: [Google's AI Overviews land in organic and cannot be separated](https://www.seoworks.co.uk/downloads/ai-referral-traffic-benchmark/), visits from in-app browsers and with stripped referrers are not counted, and GA4 counts key events rather than people, so conversions are an upper bound. Since 9 October 2026 there has been a new problem: according to LLM Pulse, [ChatGPT sessions in GA4 fell by about 46%](https://llmpulse.ai/blog/chatgpt-referral-traffic-fell/) because many links now open without a UTM parameter and without a referrer, so GA4 files them as direct traffic. We covered it in [this article](https://geosnap.ai/en/blog/traffico-chatgpt-crollato-utm-visite-perse-o-etichette-perse): lost labels, not lost visits. Every study published after that date therefore measures an AI share lower than the real one.

Then there is visibility that produces no click at all. In BrightEdge data quoted by iPullRank, [ChatGPT mentions brands 3.2 times more often than it links to them](https://ipullrank.com/ai-search-is-a-branding-channel): a recommendation read but not clicked appears in no traffic report, yet it can lead to a branded Google search or a direct visit the next day. On this the studies disagree: Kevin Indig, across 632 tests and 12 sites, [found no reliable link between AI visibility and direct or branded traffic](https://ipullrank.com/ai-search-is-a-branding-channel). That is the link in the chain still to be proven.

## How to read these numbers in your own account

Compare your site with itself, not with the study averages. The samples are small and not random (agency clients, sites that already invest in SEO) and [45 of Haynes' 100 sites did not track a single key event](https://www.stradiji.com/is-geo-a-distraction-ai-traffic-converts/): averages say little about your case. Stradiji's method is the right one for a single site: [compare the AI conversion rate with the same site's organic rate](https://www.stradiji.com/is-geo-a-distraction-ai-traffic-converts/), because conversion definitions (a form, a call, a purchase) differ from one business to the next. In practice:

1. Check that you have at least one active key event in GA4 and a segment for AI sources, as explained in [this guide](https://geosnap.ai/en/blog/traffico-da-chatgpt-come-misurarlo-google-analytics-4). Without them, no comparison is possible.
2. Wait for at least 100 AI sessions before calculating a rate: it is the minimum threshold every study used.
3. Look at the AI rate against three baselines (organic, paid, direct), not just one, to see where it really sits.
4. If AI converts worse than organic, check the gap between the promise in the answer and the landing page: a furniture store recommended for "48-hour delivery" whose product page shows no delivery times loses exactly the visitor who was closest to buying.
5. Measure presence in answers too, not only clicks: the share of answers in which you appear on ChatGPT, Gemini and Google AI Mode is the upstream figure that traffic reports cannot see. It is what Geosnap measures for every monitored prompt.

Adobe adds a pointer about content. Its "citation readability" index gives [the average US retail homepage a score of 63%](https://business.adobe.com/blog/ai-traffic-higher-revenue-per-visit-retail-holiday), with electronics customer-service and category pages at 78-79% and general-merchandise homepages at 40%. The content AI picks up best is the kind that explains: guides, FAQs, specifications, return policies. Not a new formula, but it is the first time a dataset of a trillion visits ties it directly to higher revenue per visit.

## So is GEO a distraction or not?

None of the studies says to stop doing SEO, and none says AI visibility is worthless: they say AI traffic today is small, growing and converts well. The most useful figure for a decision is perhaps the least quoted one: in The SEO Works data, [growth in AI traffic and growth in organic are correlated (+0.557 across 61 sites)](https://www.seoworks.co.uk/downloads/ai-referral-traffic-benchmark/). The sites gaining visits from ChatGPT are the same ones growing on Google, because clear content, authoritative sources and a crawlable site serve both. Choosing between the two is a false choice; the work is the same, what changes is what you measure to know whether it is working.

## Sources

- [AI Referral Traffic 2026: How Much, And Does It Convert?](https://www.seoworks.co.uk/downloads/ai-referral-traffic-benchmark/) (The SEO Works, 2026-10-14)
- [AI Traffic Drives 83% Higher Revenue Per Visit](https://business.adobe.com/blog/ai-traffic-higher-revenue-per-visit-retail-holiday) (Adobe (Adobe Digital Insights), 2026-10-29)
- [3 New Studies Suggest AI Visibility & Search Not Worth Investing In](https://www.seroundtable.com/ai-search-studies-42359.html) (Search Engine Roundtable, 2026-10-22)
- [Is GEO a Distraction? 35 Sites Say AI Traffic Is Tiny, and It Buys More](https://www.stradiji.com/is-geo-a-distraction-ai-traffic-converts/) (Stradiji, 2026-10-26)
- [AI Search Is Not a Distraction, It's a Branding Channel](https://ipullrank.com/ai-search-is-a-branding-channel) (iPullRank, 2026-10-22)
- [ChatGPT Referral Traffic Fell by Almost Half on October 9, 2026. Here's What We Know](https://llmpulse.ai/blog/chatgpt-referral-traffic-fell/) (LLM Pulse, 2026-10-21)

## Frequently asked questions

### How much traffic does ChatGPT send compared with Google?

In the October 2026 studies, between 0.62% (100 sites, 145 million sessions) and 3.86% (95 UK sites, August 2026) of Google organic sessions. The share grows every month and in US retail AI traffic rose 71% in a year, but for now it remains a small fraction of organic.

### Does ChatGPT traffic convert better than organic?

On average yes: 1.18 times across 95 UK sites, 1.64 times across 10 sites with reliable tracking, more than 4.5 times in Marie Haynes' data, 90% better in US retail according to Adobe. But on 23 of 63 sites it converted worse, and on B2B product sites the ratio drops to 0.62. Check it in your own account rather than inferring it from averages.

### Why are AI conversions so few if the rate is high?

Because a high rate applied to a small volume produces few absolute numbers. Across 95 sites, for every 100 organic conversions AI delivered 3.2; across 100 sites Google organic drove 96.5% of conversions. That is why some analysts call GEO a distraction, while others treat it as a brand channel to be measured with metrics other than clicks.

### How do I measure whether AI traffic converts on my site?

Turn on at least one key event in GA4, build a segment for AI sources and wait for at least 100 AI sessions. Then compare the AI conversion rate with the same site's organic, paid and direct rates. Keep in mind that since 9 October 2026 many ChatGPT visits arrive without a referrer and end up in direct traffic, so the real share is higher than what you see.

### Is AI visibility worth investing in if it sends so little traffic?

The studies do not say to stop doing SEO, nor that AI visibility is worthless: they say the channel is small, growing and converts well. Growth in AI traffic and growth in organic are correlated (+0.557 across 61 sites), because clear content and a crawlable site serve both. The choice between SEO and GEO is a false one.