OpenAI Says ChatGPT’s Search-Like Query Volume Is About One-Fifth of Google’s
OpenAI says ChatGPT’s web-search-like query volume is roughly 20% of Google’s traditional search volume, a ratio that puts the AI product’s discovery activity in clearer context while also exposing a major measurement problem for marketers.
The figure comes from OpenAI product lead Adam Fry, who told Big Technology that ChatGPT’s search-like queries are about one-fifth of Google’s volume. However, the two numbers do not describe identical user actions. OpenAI counts a search whenever a person prompts ChatGPT to search the web, while conventional Google search-volume measures generally refer to queries submitted to a search results page.
That distinction matters. Query activity can signal demand for information, but it does not by itself show whether users clicked through to publishers, ecommerce sites or business websites. For SEO and GEO planning, AI-assisted discovery should be measured alongside, rather than substituted for, traditional search visibility and referral data.
What OpenAI said about ChatGPT search activity
In Big Technology’s interview with Adam Fry, Fry described ChatGPT’s web-search-like query volume as approximately 20% of Google’s traditional search volume. In practical terms, that is the basis for the widely reported framing that Google processes about five times as much search activity.
The statement is a useful first-hand indication of how OpenAI is thinking about search usage inside ChatGPT. It is not, however, an independently audited market-share figure. The interview does not fully disclose the geographic scope, the detailed counting methodology or how OpenAI separates web-search prompts from other ChatGPT interactions.
ChatGPT has also been cited as handling roughly 2.5 billion prompts per day, or around 18 billion prompts per month. But prompts and searches are not synonymous. Many prompts ask for writing, analysis, coding or other assistance without requiring a web search. Conversely, a single research task may involve multiple conversational turns and web queries.
Why a direct comparison with Google needs caution
The key issue is definitional. A Google query generally begins a search-results-page session. A ChatGPT search-like query, under the definition Fry described, begins when a user asks the product to search the web. Both can reflect information-seeking behavior, but they occur in different interfaces and may lead to very different downstream actions.
| Measure | What it represents | Important limitation |
|---|---|---|
| ChatGPT web-search-like queries | Instances where a user prompts ChatGPT to search the web, under OpenAI’s stated counting approach. | Does not necessarily represent a conventional SERP search or a visit to an external website. |
| Google traditional search volume | Queries submitted through Google Search. | Volume alone does not reveal clicks, rankings, conversions or value by query. |
| Website referral traffic | Visits that arrive at a website from a referring platform. | It measures outbound traffic, not all discovery or influence that happened within an answer interface. |
The distinction becomes especially important when discussions jump from query volume to claims about website traffic. Search Engine Journal’s recap of the interview notes that AI-assisted query counts and traditional referral traffic can differ substantially.
Supporting analysis from Ahrefs illustrates the gap. Its representative-sample analysis reported Google at 24.74% of referral traffic and ChatGPT at 0.36%, and characterized Google as driving 190 times more traffic than ChatGPT in that analysis. Ahrefs also published a separate estimate that ChatGPT had 12% of Google’s search volume, underscoring that results can vary with definitions, datasets and counting methods. Read the methodology and findings in Ahrefs’ analysis of ChatGPT search volume.
These figures should not be combined into a single market-share claim. OpenAI’s 20% statement concerns search-like activity, whereas Ahrefs’ referral figures concern outbound visits in a particular sample. A platform can answer a question, cite a brand or influence a decision without producing a click. Equally, high query activity does not guarantee meaningful traffic to a site.
What this means for SEO and GEO measurement
The most practical takeaway is to make the metric explicit before using it in planning or reporting. Teams comparing Google Search with ChatGPT should state whether they mean query volume, prompts, searches initiated, citations or referrals. Each answers a different question.
For example:
- Traditional SEO reporting should continue to distinguish impressions, rankings, clicks and conversions in Google Search.
- AI visibility reporting can track whether a brand, product, source or page appears in relevant AI-generated answers, but that is not the same as receiving referral traffic.
- Discovery strategy should consider both pathways. A page may need to earn rankings and clicks in Google while also being understandable, well-supported and citable in AI-assisted experiences.
This does not make traditional SEO less relevant. Google’s much larger query volume and referral contribution in the cited data reinforce the continuing importance of crawlability, indexation, helpful content and search-result performance. It does mean that reporting frameworks built only around blue-link clicks may miss some brand exposure occurring inside answer interfaces.
Marketers should also avoid presenting ChatGPT’s reported one-fifth ratio as proof that AI search delivers one-fifth of Google’s website traffic, advertising value or commercial intent. None of those conclusions follow directly from the interview. The ratio is best understood as a directional indication that AI-assisted web discovery is significant, with comparability limits that need to remain visible.
If you need a clearer way to assess where your site may be represented in AI-assisted discovery, create a GEO Search Leads account. It can help you assess website GEO readiness and identify optimization opportunities around content accessibility, entities and potential AI visibility. Use that assessment alongside Search Console, analytics and referral data so each channel is evaluated with the metric that actually fits it.
Frequently Asked Questions
Did OpenAI say ChatGPT has 20% of Google’s search volume?
OpenAI product lead Adam Fry said ChatGPT’s web-search-like queries are roughly 20% of Google’s traditional search volume. The statement reflects OpenAI’s own counting approach, not an independently audited comparison of identical metrics.
Why are ChatGPT prompts not the same as Google searches?
ChatGPT prompts can cover many tasks, including writing, coding and analysis. OpenAI’s search-like measure counts instances where users prompt ChatGPT to search the web, while Google search volume generally refers to queries submitted to Google Search.
Does one-fifth of Google’s query volume mean ChatGPT sends one-fifth as much website traffic?
No. Query volume and referral traffic measure different outcomes. Ahrefs’ cited representative-sample analysis found a much wider gap in referrals, with Google accounting for 24.74% and ChatGPT for 0.36%.
How should marketers compare AI discovery with organic search?
Use separate measures for separate outcomes. Track Google impressions, clicks and conversions for SEO; assess citations, brand mentions and answer inclusion for AI visibility; and review referral traffic independently where it is available.
Conclusion
OpenAI’s one-fifth estimate is a meaningful signal about the scale of ChatGPT’s web-assisted activity, not a like-for-like replacement for Google search-volume or traffic data. Search teams should treat AI discovery as an additional measurement layer and define every comparison carefully, especially when translating visibility into expected visits or business results.