TL;DR
Get everyday essentials delivered free with Prime
- Fast, free delivery on millions of items
- Prime Video, Amazon Music and more included
- Member-only deals all year
OpenAI consumer product executive Adam Fry said ChatGPT handles roughly one-fifth as many search-like queries as Google, but OpenAI has not disclosed the comparison data behind the figure. Independent models put the ratio lower, while traffic panels show ChatGPT sending far fewer visits to websites than Google. The measures count different things, so they do not directly confirm or disprove OpenAI’s claim.
OpenAI consumer product executive Adam Fry said on 2 November that ChatGPT handles roughly one-fifth of Google’s query volume when counting questions that trigger a web search. The statement, made at an OpenAI-hosted dinner in New York, is confirmed as something Fry said; OpenAI has not published the underlying calculation, and independent measures report lower ratios or far fewer visits reaching websites.
Fry made the claim during a fireside chat with journalist Alex Kantrowitz. Big Technology published the conversation on 3 November, and The Verge reported from the event. Fry described the measure as “search-like queries” and, when asked what counts, said: “Every time you ask a question and go search the web, that’s a search.” The Verge reported that simple requests such as defining a word or explaining a basic how-to do not count unless they involve a web search.
Fry did not identify which Google query total OpenAI used, the time period, the countries included or the data behind the comparison. The reports cited here say Google did not comment on the claim. That leaves the ratio as a statement by an OpenAI executive, rather than a figure that outside readers can reproduce from disclosed data.
Two independent estimates in the source material are lower, though both are models rather than direct counts. Ahrefs estimated ChatGPT at about 12 percent of Google’s volume in February 2026. seoClarity estimated about 7.5 percent on 28 October. Their methods and definitions are not identical to OpenAI’s stated measure, so the estimates are useful comparisons, not a definitive test of Fry’s number.
What the Numbers Say About ChatGPT Search
OpenAI says ChatGPT handles roughly one-fifth of Google’s search-like queries. The available comparisons use different measures, so the claim cannot yet be independently confirmed.
Three different measures
Each figure tracks a different kind of activity, so the ratios are not directly interchangeable.
Search-like questions
Adam Fry said ChatGPT questions that trigger a web search amount to roughly one-fifth of Google’s query volume.
Estimated search activity
Ahrefs and seoClarity offer modeled estimates based on their own methods and definitions.
Visits to websites
Traffic panels count website visits, which may not happen when a chatbot answers without a click.
Website reach remains far lower
Traffic panels show referrals within their measured networks, not every question asked or every website.
What businesses should track
Question volume and visits measure different kinds of discovery.
- Questions may be answered inside ChatGPT.
- Sources may be cited without a visit.
- Visits show people reaching a source.
- Track citations and traffic separately.
What is still missing
The disclosed reporting does not provide enough detail to reproduce the one-fifth comparison.
- Google query baseline and measurement period
- Countries included and comparable search definitions
- Published OpenAI data or an independent audit
The comparison matters because a question answered inside ChatGPT is not the same as a visit sent to a publisher, retailer or other website. For businesses, the two measures describe different kinds of reach: AI answers may surface information or cite a source, while a click can bring a person to the source itself. Treating query volume as a substitute for referral traffic would blur that distinction.
Ahrefs’ October traffic panel tracked 118,671 websites and found that Google sent about 69 times as much traffic as ChatGPT. On that panel’s measure, ChatGPT traffic was approximately 1.5 percent of Google’s. StatCounter’s monthly figures over October 2025 to October 2026 put ChatGPT’s share of clicks to websites in its network between 0.08 percent and 0.41 percent of Google’s, according to the source material. These panels cover their own networks, not the whole web, and neither measures all questions asked in ChatGPT.
The figures point to a practical distinction for organizations that depend on online discovery: monitor citations and visits separately. A system can receive many questions without sending a comparable number of people to the sites that provide information. The available evidence does not show that ChatGPT has replaced Google as a source of website traffic.
Products Worth Considering
EcoFlow Delta 3 Max Portable Power Station – 2048 Wh LiFePO4 Battery, 2400 W Output (3400 W X‑Boost)
DJI Osmo Mobile 8 3-Axis Phone Gimbal Stabilizer
Samsung Galaxy Tab A7 Lite 8.7 32GB Wi‑Fi Android Tablet – Long‑Lasting Battery, Compact Slim Design, Sturdy Metal Frame (US Version)
As an affiliate, we earn on qualifying purchases.
How the Measures Differ
The figures compared here use at least two different yardsticks. OpenAI’s claim concerns questions in ChatGPT that lead to a web lookup, compared with a Google query total that Fry did not name. Ahrefs and seoClarity offer modeled estimates of ChatGPT prompts that resemble search activity. Those estimates depend on their own definitions and methods, which the supplied material does not fully detail.
Traffic panels count another outcome: visits referred to websites. A ChatGPT response can answer a question without a user clicking a source, while a search engine query may also end without a click. As a result, a lower ratio in a traffic panel does not by itself refute a higher ratio in question volume. Questions, modeled search intent and website visits are not interchangeable units.
Fry also said ChatGPT had grown from one billion to 1.2 billion weekly users in two months. That is a separate statement in the same conversation. The source material does not provide a method or supporting data for that user figure, and it does not establish how weekly users relate to the search-like query estimate.
Google’s recent results offer additional context, rather than a direct comparison. Alphabet said third-quarter query growth was its fastest in five years, and Google Search revenue rose 12 percent to $63.35 billion. Those company-reported figures indicate continued growth in Google Search, but they do not independently verify or invalidate OpenAI’s query ratio.
The Missing Comparison Details
The main unresolved issue is how OpenAI calculated the one-fifth ratio. The source material does not identify the Google total, the dates covered, the countries included or whether the comparison uses the same definition of a search on both services. Without those details, outside researchers cannot reproduce the result or make a direct like-for-like comparison.
The independent estimates also have limits. Ahrefs and seoClarity modeled activity; they did not report a complete count of all ChatGPT web searches. Their percentages differ, and the material does not provide enough methodological detail to explain the gap fully. The traffic panels are narrower still: Ahrefs counted visits to a specified group of websites, while StatCounter covered clicks within its network. Neither panel measures every website or every search.
The available reporting also does not include a response from Google, a published OpenAI dataset or an independent audit of the company’s figure. It remains unclear whether OpenAI will release more detail, and whether its definition can be aligned with the measures used by outside analysts.
Whether OpenAI Publishes Its Method
The next useful development would be disclosure from OpenAI of the Google baseline, measurement period, geographic coverage and counting method behind Fry’s statement. Those details would let analysts assess whether the comparison is reproducible and how much it differs from external models.
Until then, readers and businesses can treat the one-fifth figure as an attributed OpenAI claim, and the Ahrefs and seoClarity figures as estimates based on their own models. Traffic panels provide a separate view of referrals within their measured sites. New data using clear definitions could show whether these gaps reflect different measures, different periods or a change in how people use ChatGPT and Google.
Key Questions
Did OpenAI confirm that ChatGPT has one-fifth of Google’s search volume?
Adam Fry said ChatGPT’s search-like query volume was roughly one-fifth of Google’s. The source material does not include the underlying data or an independently verified calculation, so the figure is confirmed as his statement, not as an independently established fact.
What does OpenAI count as a search?
Fry said a question counts when the user asks it and ChatGPT searches the web. The Verge reported that requests such as a word definition or simple how-to explanation do not count unless they involve a web search.
Why do website traffic figures look much smaller?
Traffic panels count visits sent to websites, not all questions asked in a chatbot. A user may receive an answer without clicking a source. The panels also cover particular website networks, so their results do not represent the entire web.
What did independent estimates find?
Ahrefs estimated about 12 percent of Google’s volume in February 2026, while seoClarity estimated about 7.5 percent on 28 October. Both are models, and their definitions may differ from OpenAI’s.
Source: Thorsten Meyer AI
