"ChatGPT traffic converts nine times better" does not hold up
Half the industry repeats it. Traced back to its origin, the figure does not exist.
The published evidence puts conversion from ChatGPT traffic between 13% worse and 31% better than organic traffic — roughly at par. The largest peer-reviewed study — 973 sites and 164.8 million transactions — is the one finding 13% worse. In services and B2B, the only study with Spanish data (GA4 of 25 websites, October 2026) finds a median of ×1.87 measured on key events, not sales; among the sites that convert most, the difference can no longer be told apart from chance.
- Sample
- Review of the published evidence. Largest study: 973 sites and 164.8 million transactions. Since October 2026, also a Spanish study of 39 websites
- Period
- Studies published between 2025 and 2026
- Method
- Desk research. No measurement of our own: we trace the original sources of a figure widely repeated in the industry.
- Language
- Sources in English and Spanish
Where the "nine times" comes from
Two different things get cited, and they collided by accident:
- An agency case study whose own title says "1 site": a single client. What it calls "conversion" is a Google Analytics key event, which can be a PDF download. It compares 10,969 AI sessions with roughly 14 million organic ones.
- A business-press headline that popularised the multiplier and does not even cite that case study: its number came from an A/B test of an on-site search engine inside online stores. Different metric, different context.
Neither source measures what the sentence claims.
What the serious evidence says
- Kaiser and Schulze, Marketing Science (INFORMS), 2026. Peer-reviewed. 973 sites, 50,251 transactions from ChatGPT against 164.8 million traditional transactions: ChatGPT traffic converts around 13% worse than organic.
- Adobe, across more than a trillion visits: for most of 2025, AI traffic converted worse than organic.
- Visibility Labs, 94 sites: +31%, the most favourable published comparison.
In e-commerce, the honest range is −13% to +31%. That is parity within margin of error, not a multiplier.
Update, 14 October 2026: a Spanish study in services and B2B
iSocialWeb has published the first study with Spanish data: GA4 for 39 of its clients' websites, from October 2025 to September 2026, with the aggregated data published openly. On the 25 sites with reliable key events, AI traffic converts better than Google organic on 18, with a median of ×1.87. On the 5 online stores, ×1.15.
- It does not contradict the above: it measures something else. Its conversion is a key event (a form, a call), not a sale, and most of the sites sell services. The ×1.15 for online stores falls inside our range.
- 18 out of 25 is hard to explain by chance: if both channels converted the same, it would come out like that less than once in 20. But on the 16 sites with at least 30 conversions from AI, AI wins on 11, and that happens by chance once in five. This is our own calculation on their published figures; the study does not publish intervals.
- It confirms what matters. AI brings 0.4% of visits and around 0.5% of conversions. Whatever it converts, the volume is small today.
That is why the headline stands: "nine times" still has no source. What has changed is the product. The engine now computes that ratio for each client from their own GA4 (12 months, AI against Google organic, with its confidence interval) and uses it instead of a third-party study. If there is not enough data, it says so.
The argument that does hold up
The value of AI visibility is not in its conversion rate: it is in the volume and the trajectory of the channel.
Today, traffic from AI search engines is a tiny fraction of the total and is growing at triple-digit rates. What you buy by positioning now is not a present return but a position in a channel that is still cheap to occupy. That argument does not need an inflated figure, and it does not collapse when somebody checks it.
Why we publish this against our own interest
"Nine times better" is the most convenient sales argument a tool like this could have. We used it: it fed our own ROI calculator until we traced it. The reason for retiring it is practical before it is ethical: a figure that a peer-reviewed paper refutes is a time bomb inside a commercial proposal, and all it takes is the client asking the AI itself.
What this finding does NOT prove
- It does not say AI traffic is not worth it. It says its value is not in a superior conversion rate.
- The studies cited measure different periods, sectors and definitions of conversion: they are not directly comparable, which is why we publish a range and not an average.
- This is not our own measurement. It is a review of published evidence, and as such it ages: if a larger study appears, this page changes.
- AI traffic attribution is systematically undercounted in analytics — many assistants send no referrer — which affects every study cited, in an unknown direction.
- The Spanish study comes from one agency portfolio and measures key events: we do not know how many of those forms and calls became sales.
Frequently asked questions
Does traffic arriving from ChatGPT convert better?
In e-commerce, the published evidence ranges from 13% worse to 31% better than organic traffic, which is roughly the same. The largest peer-reviewed study (973 sites, 164.8 million transactions) finds it converts 13% worse.
Where did the "converts nine times better" figure come from?
From the collision of two sources measuring something else: a single-site case study whose "conversion" was a Google Analytics event, and a press headline whose multiplier came from an A/B test of an on-site search engine in online stores.
What about services and B2B?
The only study with Spanish data (iSocialWeb, 25 websites, October 2026) finds a median of ×1.87 measuring key events, not sales. Among the sites with the most conversions the difference can no longer be told apart from chance, so it is a signal, not a figure for a budget.
So why invest in AI search visibility at all?
For volume and trajectory, not for conversion rate. The channel is a tiny fraction of traffic today and growing at triple digits: positioning now is cheap, and it will be less so later.
Home · Methodology · Plans & pricing · X · Product Hunt