Three independent analyses published within about a week reached the same conclusion: AI assistants deliver a sliver of the visits and sales that Google organic search does. Search Engine Roundtable’s Barry Schwartz gathered them on October 22, and the pattern is hard to wave away even though the methods differ.
Marie Haynes examined 12 months of GA4 data from 100 websites of varying sizes, covering more than 145 million sessions. In her dataset, ChatGPT sessions equaled 0.62% of the volume Google organic produced. ChatGPT visitors did convert better, at 13.8% against 3.02% for Google organic, which is more than 4.5 times the rate. Even so, Google drove 96.5% of conversions across all channels she analyzed, and 27 times more than every other LLM combined, according to Haynes.
SEO Works measured the gap differently. It found organic search delivered 39 times the traffic of AI assistants. For every 100 conversions from organic, AI referrals produced 3.2. The firm also reported that AI referral traffic converts worse than paid search (0.67 times the rate) and better than direct traffic (2.48 times).
Kevin Indig asked a separate question: does AI visibility show up as extra direct or branded traffic that referral reports miss? Across 12 sites in different sectors and sizes, he said he could not detect it. In his words, AI-driven direct traffic to longtail pages was “not detectable yet,” and he could not establish a dependable connection from how often a brand shows up in AI answers to its branded-search or direct visits. He allowed that a firm whose AI footprint is expanding fast could see an effect.
The common thread is scale. A higher conversion rate on a tiny base of visits still yields a tiny total. A channel converting at 4.5 times the rate of Google organic would need far more than 0.62% of Google’s volume before it changed a revenue line. Search Engine Land ran a same-day headline stating that ChatGPT sends less than 1% of Google’s traffic, which fits the Haynes figure, though the full text was not available for this report.
The studies do have limits. Haynes and Indig each describe samples of 100 and 12 sites, and the SEO Works summary does not describe its sample or attribution rules. Referral analytics undercount AI influence when users read an answer, then search the brand later or type the URL. Indig tried to test for that leakage and came up empty, but a null result on 12 sites is not proof of absence. None of the three can see whether being named in an answer shapes a purchase decision that happens elsewhere.
Here is the position that follows. Treat GEO as a small, cheap hedge, not a budget line that competes with organic search. The evidence supports spending that overlaps with good SEO anyway, such as clear pages, strong entities, and citable facts. It does not support dedicated retainers, new headcount, or tooling spend justified by promised traffic. Vendors selling AI visibility can point to conversion rates, but the volume behind those rates is what these studies show to be small.
The more useful discipline is measurement. Most teams can pull GA4 channel data in an afternoon and compare AI referral sessions with Google organic, as Haynes did, then check conversion share rather than visit share. If your numbers resemble these three studies, hold AI-specific spending flat through the next two quarters and redirect any increase to the organic pages that produce the conversions. If your numbers look very different, that gap is worth more than any industry benchmark.
Based on reporting by Barry Schwartz at Search Engine Roundtable, published October 22, 2026, which summarized studies from Marie Haynes, SEO Works, and Kevin Indig.
