Adam Fry, who leads consumer product for ChatGPT at OpenAI, says Google sees about five times the search-style demand ChatGPT does, a ratio he put at roughly one to five. Search Engine Journal reported the comparison on Tuesday, drawing on a transcript that Big Technology published of remarks Fry made at an OpenAI-hosted dinner in New York on Monday evening.
The figure is an executive’s estimate, and the counting rule matters more than the ratio. Asked by Big Technology’s Alex Kantrowitz whether one conversation with a search is one search or several, Fry answered: “Every time you ask a question and go search the web, that’s a search.” A long chat that triggers three lookups therefore adds three to the tally.
Fry also described how usage has shifted. Early users leaned on ChatGPT for writing and coding, tasks nobody would have typed into a search box. During the last twelve months, per Fry, users added finance and health topics and queries on public figures, venues, and products that require fresh information. He framed the category as whatever a person might previously have sent to Google or Bing.
Several details are missing. According to Search Engine Journal, Fry did not say which Google number the comparison uses, what time period it covers, or which regions are included. Without a stated denominator, the one-fifth ratio cannot be reproduced, and teams should treat it as directional.
Outside estimates use different definitions and land in a different place. In February, Ahrefs put ChatGPT at about 12 percent of Google’s search volume. Its method began with a Harvard University and OpenAI study counting 2.5 billion ChatGPT prompts a day in July 2025, and assumed roughly 65 percent of them were the kind of query people once asked Google. It then set that against the more than 5 trillion annual searches Google announced in March 2025. Ahrefs classified prompts by intent, while Fry’s rule counts actual web lookups, so the two numbers describe different things.
Traffic tells a starker story. Ahrefs also publishes a referral benchmark built from sites that have installed its free analytics product. Its November 3 update covers October for 118,671 sites: Google supplied 24.74 percent of the traffic it logged, while ChatGPT supplied 0.36 percent. A product can field a large number of searches and still send few visitors, because an answer delivered inside the chat often ends the session without a click.
Google’s side of the ledger is reported as a rate, not a count. Alphabet’s third-quarter release credits AI investments for lifting overall query growth, and CEO Sundar Pichai said Google’s “year-over-year query growth rate is the highest it’s been in the last five years.” Neither that statement nor Fry’s estimate shows whether ChatGPT searches replace Google searches or add to total search activity. That distinction decides whether publishers face substitution or a new, mostly uncounted channel.
For search teams, the practical risk is a slide deck that quotes the one-fifth figure beside a referral chart and implies the two belong together. They do not. Query volume describes how often a platform goes to the web, and referral traffic describes who actually arrives at your site.
Anyone building a channel plan that includes ChatGPT should report both measures with their definitions attached, and pull the Ahrefs referral share monthly as an outside check. Hold off on reallocating budget based on OpenAI’s ratio until the company states which Google figure it used.
Search Engine Journal, Matt G. Southern, published November 3, 2026, based on a Big Technology transcript of Adam Fry’s remarks.
