A big number, offered at dinner

The setting matters. This was not a filing, a blog post or a data release. It was a fireside chat at a dinner OpenAI hosted for journalists in New York on Monday night, and the number arrived the way numbers at dinners do: rounded, quotable and unsupported by anything a reader can check afterwards. "ChatGPT’s volume of search-like queries is now roughly one-fifth the volume of queries of Google," Adam Fry said.

Fry runs consumer product for ChatGPT, and he used to work at Google, so he knows what the comparison invites. When Kantrowitz raised Bing’s roughly 5% search-market share and asked whether the comparison implied ChatGPT was several times larger than Bing, Fry did not provide a direct comparison. Instead, he cited ChatGPT’s growth in weekly active users. ChatGPT has gone from one billion to 1.2 billion weekly active users in two months, roughly 100 million new users every 30 days, on OpenAI’s account.

Everything in this story turns on a definition, and OpenAI supplied its own. Google’s comparison is also difficult to reproduce. A conventional Google search query is counted as a search, while Fry’s definition counts a ChatGPT interaction whenever a question triggers a web search. A single ChatGPT conversation can therefore generate multiple search-like queries. Without matching definitions on both sides, the ratio is not an apples-to-apples comparison.

Those two units are not the same size, and adding them into a ratio produces a number that is precise in form and soft in content. OpenAI has not published how many search-like queries it counted, over what period, in which countries, or where it obtained Google’s total. Google does not publish a daily query count either; the most recent anchor is its disclosure of more than five trillion searches in a year, from 2025. Any 20% claim therefore has a numerator only OpenAI can see and a denominator nobody has published.

The estimates that can be checked land lower

Two outside attempts exist, and both are explicit about being models rather than measurements. seoClarity’s October 2026 update, built from OpenAI’s own Signals v3.0 message data, classifies about 27.5% of ChatGPT messages as search-equivalent — its "seeking information" category plus an assumed quarter of "practical guidance" — and arrives at roughly 1.24 billion search-like messages a day. Against a modeled 16.44 billion Google queries a day, that is about 7.5%. The Google side of that model itself assumes 20% cumulative growth on the five-trillion figure, an assumption seoClarity labels as its own.

Ahrefs, working in February 2026, put ChatGPT at about 12% of Google’s search volume. The estimates point in the same broad direction: ChatGPT’s search-related activity is substantially smaller than Google’s. But they are not directly comparable measurements. seoClarity’s model puts the ratio at 7.5%, Ahrefs estimated about 12% in February, and OpenAI now claims roughly 20% using a definition and methodology it has not disclosed. The gap between those figures is not necessarily growth. It may reflect different definitions, measurement periods and assumptions.

Queries are not visitors

For anyone who runs a website, there is another comparison worth separating from search volume: referral traffic. Ahrefs’ February 2026 study, as reported by PPC Land and Steven Digital, examined 76,000 websites and found Google accounting for nearly 40% of their traffic, compared with 0.21% from ChatGPT. Within that sample, Google generated roughly 190 times more website referrals than ChatGPT. That is a different measurement from search volume, but it illustrates why a search-like query count should not be confused with the traffic publishers actually receive.

That gap is structural, not a rounding error. A search engine’s product is a list of doors to other sites. A chatbot’s product is an answer, and the answer usually ends the trip. A query counted inside ChatGPT can inform a purchase, settle an argument or kill a news story’s traffic without a single outbound click. Publishers and marketers who budget against the 20% figure as if it were 20% of Google’s distribution will misread their own analytics, which will keep showing Google as the source that actually arrives.

What would settle it

Three things, none of them difficult for a company of OpenAI’s size: the count itself, the period it covers, and the method, including where the Google denominator came from. Until those appear, the 20% belongs in the category of dinner numbers — directionally interesting, because ChatGPT search plainly is large and growing, and unusable for any decision that needs a denominator. The user-growth figure Fry cited alongside it is a separate claim: OpenAI says weekly active users increased from 1 billion to 1.2 billion over two months. That figure does not resolve the search-volume question, and it has not been accompanied by a detailed measurement methodology either.

Sources

  1. [1] Big Technology, Alex Kantrowitz, "OpenAI: ChatGPT Search Is About 20% of Google’s, 250 Million People Speak With ChatGPT, and Chat Is Merging With Work", November 3, 2026 (primary source — the Adam Fry interview; the free portion was read first-hand at Gate 1: the one-fifth quote, the Bing exchange and Fry’s answer citing weekly-user growth, 1 billion to 1.2 billion weekly users in two months, 250 million combined voice users and 1.5 billion images a week)Read source
  2. [2] 9to5Google, "ChatGPT hits 20% of Google Search volume, OpenAI says", November 3, 2026 (corroborating account of the same event, read first-hand at Gate 1 — including Fry’s definition quote: "Every time you ask a question and go search the web, that’s a search")Read source
  3. [3] seoClarity, "ChatGPT vs. Google in 2026: Updated Search Volume Estimate", October 28, 2026 (read first-hand in full at Gate 1 — the modeled 7.5% estimate: about 1.24 billion search-like messages a day against a modeled 16.44 billion Google queries; the 27.5% search-equivalent share built from OpenAI Signals v3.0 (Seeking Information plus an assumed 25% of Practical Guidance); the assumed 20% cumulative growth on Google’s disclosed more than five trillion annual searches; seoClarity labels the figure a modeled comparison, not a measured count)Read source
  4. [4] Ahrefs, Patrick Stox, "ChatGPT Has 12% of Google’s Search Volume but Google Sends 190x More Traffic to Websites", February 11, 2026 (the study itself read first-hand at Gate 1 — ChatGPT at about 12% of Google’s search volume for queries people traditionally searched on Google; Google at nearly 40% of website traffic against ChatGPT’s 0.21%; Google sending 190x more traffic. The study’s 76,000-website sample is as documented in PPC Land’s account of the study)Read source