An OpenAI executive says ChatGPT's search-like queries are roughly one-fifth of Google's
OpenAI's ChatGPT consumer product lead, Adam Fry, said at a New York dinner hosted by OpenAI that ChatGPT's search-like query volume is about a fifth of Google's. According to the Big Technology interview published on November 3, 2026, it is an estimate offered in conversation, not a formal measurement report.
A ratio stated at a dinner: one-fifth of Google
Alex Kantrowitz of Big Technology published his conversation with Adam Fry on November 3, 2026. Both Big Technology and Search Engine Journal introduce Fry as ChatGPT's consumer product lead at OpenAI. The exchange was a brief fireside chat on Monday night at a dinner OpenAI hosted in New York City. Kantrowitz noted that OpenAI had compared ChatGPT's search activity with Google's and asked what it found. Fry said ChatGPT's search-like queries now come to "roughly one-fifth the volume of queries of Google."
Search Engine Journal presents the same statement as something said in an open discussion with Kantrowitz. That makes the ratio a claim by an executive about his own company's product, so it is best read as attributed, not as a measured fact. In the rest of the conversation Fry discussed how fast ChatGPT is growing, what personal agents cost to run and a plan to combine the Chat and Work modes; this article stays with the search and user figures.
How a search is counted, and what the user figure says
Search Engine Journal reports that OpenAI counts one search whenever a question makes ChatGPT go out to the web, and Fry described it in just those terms. In his account, early users mainly wanted help with writing and code, and over the past year topics needing current information were added: people, places and things, plus finance and health questions. For those, ChatGPT runs a web search. Fry framed the category as whatever a person would previously have taken to a conventional search engine.
In the same interview Fry said weekly active users went "from one billion to 1.2 in just the last two months." Big Technology summarizes that as 200 million added users in 60 days. Citing OpenAI, it also reports 250 million weekly active users of voice mode, counting dictation and conversation together. These too are OpenAI's own statements, relayed by the interviewer. According to Search Engine Journal, Fry did not say which Google figure he compared against, what period the comparison covers, or which regions it includes.
Different yardsticks give different answers, so definitions matter
Search Engine Journal sets the claim beside other estimates. An Ahrefs estimate from February put ChatGPT at about 12 percent of Google's volume, but it used a different definition. Its starting point was a July 2025 figure of 2.5 billion ChatGPT prompts a day, of which it treated roughly 65 percent as the kind of request people used to look up on Google, then compared that with the more than 5 trillion searches a year Google announced in March 2025. Fry's count covers only questions that go to the web, so the two methods are not measuring the same thing.
Traffic paints a separate picture. Ahrefs' October figures, drawn from 118,671 sites, show Google sending 24.74 percent of the traffic it tracked, against 0.36 percent from ChatGPT. That is a measure of visitors sent to websites, not of query volume. In Alphabet's third-quarter release, Sundar Pichai said Google's year-over-year query growth rate is "the highest it's been in the last five years." Search Engine Journal's conclusion is cautious: neither Fry's ratio nor Google's growth rate shows whether ChatGPT searches are replacing Google searches, and anyone citing the number should say which measure they mean. We covered Ahrefs data on the sources ChatGPT leans toward on October 12.
The sources say nothing about Türkiye, so this section is commentary
Neither source mentions Türkiye. There is no count of ChatGPT users there, no share of search in the country, no local institution and no data on Turkish-language queries. This section is therefore not a fact drawn from the sources but explicitly UNALSOFT commentary.
Our view is that because OpenAI defines the measure and has not said which regions it covers, a business in Türkiye cannot translate it directly into its own customers' behavior. The unknowns, stated plainly: which regions the figure covers, whether the share is higher or lower in Türkiye, how Turkish-language queries feed into this volume, and which absolute query counts Fry's ratio rests on. Before deciding anything, your own Google Analytics and Search Console data is a more dependable basis than this number.
The UNALSOFT view
We read this as a sign that AI search is a channel too big to ignore but not yet a replacement for classic search. Because the figure is an estimate given in conversation, it should not be the sole basis for a budget or strategy decision. The practical step is simple: invest in making your site's content readable and quotable for AI assistants, but judge the results against your own traffic and conversion data. That work falls within our agentic AI service. Since the sources carry no data about Türkiye, this article proposes no local conclusion.
Is your site readable by AI assistants?
A short conversation is enough to review your content and your measurement setup together.