OpenAI Says ChatGPT’s Search-Like Queries Are About One-Fifth of Google’s

OpenAI has offered a striking new comparison for the scale of AI-assisted discovery: ChatGPT’s "search-like" queries are approximately one-fifth of Google’s query volume.

The figure came from Adam Fry, OpenAI’s consumer product lead, during an open conversation in New York reported by Big Technology. As described in Big Technology’s interview transcript, OpenAI counts a search when a user asks ChatGPT a question that sends the product to the web.

That definition is important. The claim is a meaningful indication that web-connected AI interactions have become substantial, but it should not be read as evidence that ChatGPT sends one-fifth as much traffic to websites as Google. Query activity, web retrieval, citations, clicks, and referral visits are related measures, not interchangeable ones.

What OpenAI’s one-fifth comparison means

Fry’s statement compares ChatGPT search-like interactions with Google query volume. In practical terms, it covers cases where a ChatGPT question triggers a trip to the web, rather than every conversation a user has with the chatbot.

The comparison provides a newly documented benchmark from OpenAI for considering the relative scale of AI-assisted discovery. It also follows a broader shift in which users can ask conversational questions, receive synthesized answers, and sometimes continue to linked sources without conducting a conventional search-results-page journey.

However, crucial details behind the comparison were not disclosed. There is no stated Google baseline, timeframe, geographic scope, or detailed methodology for matching the two products’ activity. The figure therefore cannot establish a precise market share, a universal traffic ratio, or a forecast for how people will search in the future.

Search Engine Journal’s coverage of OpenAI’s statement also highlights the distinction between the search-like query count and referral-traffic data. That distinction should shape how publishers, marketers, and SEO teams interpret the headline.

Why query volume is not referral traffic

A traditional Google search and a ChatGPT web-connected question can both begin a research task, but they do not necessarily produce the same path to a website.

A Google results page presents a list of links and search features. An AI answer may summarize information directly, cite sources, recommend an option, or prompt the user to ask a follow-up question. A web retrieval event can occur without a source receiving a click. Conversely, a user can reach a site through a Google result without generating an identifiable AI interaction.

This means the one-fifth figure should not be used to infer any of the following:

• ChatGPT delivers one-fifth of Google’s website referrals.
• A site’s AI referral traffic should equal one-fifth of its Google organic traffic.
• AI answer visibility has the same commercial value as a conventional organic ranking.
• ChatGPT’s share applies consistently across countries, languages, industries, devices, or query types.

The comparison instead describes a type of user activity: questions that cause ChatGPT to search or retrieve from the web. That is strategically relevant, particularly for informational and research-led journeys, but it is not a direct traffic-share measurement.

How this compares with other available estimates

The reported OpenAI figure is not the only attempt to quantify ChatGPT’s relationship to Google, but different studies use different definitions. Ahrefs estimated in February 2026 that ChatGPT search volume was around 12% of Google’s. That estimate used a different definition of search, so it should not be treated as a contradiction or merged with OpenAI’s approximately 20% claim.

Google has reported more than 5 trillion searches per year, which illustrates the scale of the denominator implied by any comparison with Google. But without a confirmed, like-for-like baseline for Fry’s statement, multiplying that annual figure to estimate ChatGPT activity would create a false precision.

The useful conclusion is that both OpenAI’s figure and third-party estimates point to a meaningful level of AI-assisted discovery. They do not establish one settled ratio for every measurement purpose.

What this changes for search reporting and content strategy

Teams tracking organic visibility increasingly need to separate discovery metrics rather than compress them into one number. Google impressions, Google clicks, AI referral sessions, citations in AI responses, brand mentions, and web-search-like interactions each answer different questions.

For editorial and SEO planning, the practical response is not to reduce investment in traditional search. Google remains the much larger source of search activity and, according to the cited referral comparisons, of visits. Instead, organizations should account for the possibility that a user’s first encounter with a brand, product, or source occurs inside an AI-generated answer.

That calls for content that is both discoverable in conventional search and easy for retrieval systems to interpret. Clear page structure, accurate claims, accessible supporting evidence, identifiable entities, and useful first-party information can help content serve users regardless of whether they arrive through a search result, an AI citation, or a direct referral.

Reporting should also label metrics precisely. A dashboard that calls all AI activity “search traffic” can obscure whether the organization is measuring answer visibility, links shown, actual sessions, or conversions. The most useful comparisons are like for like: referral traffic against referral traffic, and query activity against query activity.

ChatGPT’s growing role in web-connected research makes it more important to understand where a site is cited, mentioned, and represented in AI-generated answers. Create a GEO Search Leads account to begin assessing your website’s GEO readiness and identifying optimization opportunities alongside your established search measurement.

Frequently Asked Questions

Did OpenAI say ChatGPT has one-fifth of Google’s traffic?

No. OpenAI’s reported statement concerned ChatGPT “search-like” queries, defined as questions that send ChatGPT to the web. It did not claim that ChatGPT sends one-fifth of Google’s website traffic or referrals.

What details are missing from OpenAI’s comparison?

The reported statement did not disclose the Google baseline, timeframe, geographic scope, or a detailed methodology for comparing ChatGPT’s web-connected questions with Google queries.

Why does the Ahrefs estimate differ from OpenAI’s figure?

Ahrefs’ February 2026 estimate of around 12% used a different definition of search. Figures based on different data sources and measurement methods should not be treated as directly interchangeable.

Should SEO teams shift their focus away from Google?

No. The available referral comparisons still show Google dominating visits. The practical approach is to retain strong traditional SEO while measuring and improving visibility in AI-assisted discovery separately.

Conclusion

OpenAI’s one-fifth comparison is an important directional signal, not a complete map of search market share or website traffic. It shows that web-connected AI questions have reached meaningful scale, while reinforcing the need for more careful measurement. Organizations should evaluate conventional search performance and AI discovery visibility together, but report each with definitions that reflect what is actually being measured.