OpenAI Says ChatGPT Search Volume Now Equals a Fifth of Google's

OpenAI says the volume of search queries run through ChatGPT now equals 20 percent of the volume run through Google. Adam Fry, the company's consumer product lead for ChatGPT, announced the figure at a press event on Monday, 2 November. The Verge reported the claim on 3 November.

OpenAI's definition of a search is narrow. It counts any instance in which a user enters a query that results in a web search, meaning ChatGPT goes out to the live web rather than answering from what it learned during training. Tasks handled without one, such as defining a term or giving a simple how-to explanation, fall outside the count.

The ratio is only as informative as its two inputs. The ChatGPT side follows OpenAI's own definition, and the Google side is a comparison figure that Fry supplied at the event. Those two counts may not measure the same thing, so the 20 percent number should be read as a company-supplied comparison, not as a like-for-like market share.

In May 2024, Reuters reported, citing sources familiar with the matter, that OpenAI planned to announce an AI-powered search product the following Monday. In July 2024, the Associated Press reported that OpenAI had tested a ChatGPT-powered search engine that would compete directly with Google Search. On 31 October 2024, Bloomberg reported that OpenAI had added search features to ChatGPT that let users look up timely information. OpenAI said on February 5, 2025, that ChatGPT search had become available to everyone in regions where ChatGPT is available, with no signup required.

Google's own response to the same shift was also reported in detail. Bloomberg reported in March 2025 that, shortly after OpenAI released ChatGPT, Google reassigned more than 1,000 engineers, about 20 percent of its search engineering team, to generative AI work.

The monetization question has run alongside the usage question. In reporting published in November 2024, Fry said there were "no plans" for advertising in ChatGPT. In April 2025, OpenAI announced updates to ChatGPT intended to produce better, more substantial shopping recommendations. Neither development settles how a high-volume ChatGPT search product would be paid for, and the two sit in different categories of commercial intent. A shopping query and a general informational query carry different expectations about results and placement.

The broader context here is that Google's search business has been built on advertising, so query share and revenue model are closely linked. A product that absorbs a fifth of comparable query volume without an ad model, if that remained the case, would be testing a different economic arrangement from the one that has shaped web search for two decades. Whether that arrangement holds is a question the announcement does not answer.

In my view, the most useful reading of the figure is probably as a measure of habit. Expert users have long turned to chat interfaces for synthesis, code, and quick factual lookups. A search-triggering query is the point at which a chat interface starts to behave like a search engine. A ratio of 20 percent says that a meaningful share of those moments now happen inside ChatGPT rather than on a results page. That is the part worth tracking over the next several quarters, because it will show whether the number holds up under independent measurement and whether the habit persists once the novelty fades.

For the people who build on either platform, the practical implication is that retrieval (fetching documents from the web), ranking, and citation behavior in conversational search are now a live competitive surface. Teams working on search quality, grounding (tying generated answers to the sources they were drawn from), and inference costs for web-backed answers have a larger consumer benchmark to measure against.

In my view, what this makes possible for users is a second large, conversational entry point to the web. Most people in this field have been expecting that change for some time.