ChatGPT Says It Has 20% of Google's Search Volume (2026)

What Did OpenAI Actually Say About ChatGPT Search?

The source is a conversation between Big Technology's Alex Kantrowitz and Adam Fry, ChatGPT's consumer product lead, at an OpenAI-hosted dinner in New York on Monday night. Kantrowitz asked what OpenAI found when it compared ChatGPT's search volume with Google's. Fry answered: “ChatGPT's volume of search-like queries is now roughly one-fifth the volume of queries of Google.”

He added a growth number in the same breath. According to the Big Technology write-up, ChatGPT went from 1 billion to 1.2 billion weekly active users in 60 days, which Fry described as “roughly 100 million users in 30 days, and another 100 million in another 30 days.” Big Technology also reported two other figures from the conversation: about 1.5 billion images created per week, and 250 million combined weekly users of voice mode. None of those numbers has been independently checked that we could find, and they were shared by the company that benefits from them being large.

Search Engine Roundtable's Barry Schwartz picked it up the next morning. His reaction is worth quoting because he is not an easy audience for a big claim about Google: he called the figure surprising, “especially since Google says their search volume is at an all-time high.” The SEO consultant Glenn Gabe, whose post Schwartz embedded, drew a comparison to Bing. Kantrowitz noted in the same thread that Bing has around 5% of search share.

The Verge added some context on how big that is. It pointed out that Microsoft CEO Satya Nadella said during Google's 2023 antitrust trial that Bing had about 3% of the search market after more than a decade of trying. One-fifth of Google's volume in roughly a year and a half, if it holds up, is a different order of growth. Fry himself called it “a pretty amazing statistic when you think about it, mainly because we started just about a year and a half ago.”

Notice the shape of what was and was not said. OpenAI gave a ratio. It did not give a count of searches, a count of Google's searches, a time window, or a method. Gizmodo did some arithmetic of its own, taking a widely repeated estimate that Google handles 16 billion searches per day and arriving at about 3.2 billion for ChatGPT. That is Gizmodo's back-of-envelope math, not an OpenAI number, and it depends on an outside estimate for Google that OpenAI never named.

What Counts as a Search Here?

This is the part to read slowly, because the whole claim turns on the definition. Kantrowitz asked it directly: if a person has 20 prompts back and forth in one conversation, does each one count? Fry's answer, as quoted by Search Engine Roundtable, was: “Every time you ask a question and go search the web, that's a search.”

The Verge reported a slightly tighter version from the press event. It says Fry described a search as any time a user inputs a query that results in a web search, as opposed to tasks such as defining a term or a simple how-to explanation. Gizmodo quoted Fry describing a search-like query as “basically anything that a user would have traditionally gone to a traditional search engine for, but is now deciding to go to ChatGPT for.”

Put those together and you get a counting rule that is reasonable on its face and loose in practice. A prompt counts when ChatGPT decides to go to the web. That decision belongs to ChatGPT, not the user, and Gizmodo made the obvious point that many prompts now lead the tool to check outside sources, so a large share of interactions can qualify as search-like. 9to5Google was blunter, saying there is “a lot of wiggle room” in the number, because the source of Google's volume was not revealed and the definition of a query is OpenAI's.

Fry also gave a hint about what kinds of questions are driving it. According to the Search Engine Roundtable quote, people have grown into topics such as looking up a person, a place or a thing that requires up-to-date information, and into finance and health questions where they want current, factual answers. “A place or a thing” is the part that matters to a business owner. A restaurant, a roofer and a dentist are all places and things.

Does This Mean Google Lost One Search in Five?

No, and it is worth being exact about why. A ratio of volumes tells you how big one pile is next to another. It does not tell you where the first pile came from. Some ChatGPT searches may replace a Google search. Some may be questions nobody would have typed into Google at all, such as a long, specific request that a person would have given up on. Some may come from the same person who also still uses Google ten times a day. The claim does not separate those, and neither can we.

Gizmodo noted that Google owns over 90% of the global search engine market share, if AI tools are not counted. Google's own position, per the Search Engine Roundtable post, is that its search volume is at an all-time high. Both can be true alongside OpenAI's claim, because they measure different things. If ChatGPT adds a new kind of query rather than taking an old one, you would expect to see exactly this: Google steady or up, ChatGPT large and growing.

There is also a business context to keep in mind. Gizmodo observed that this is the kind of thing a company would want to say publicly ahead of a planned initial public offering. The Verge reported that OpenAI's IPO is potentially slated for early next year. We are not saying the number is wrong. We are saying that a figure announced by the seller, in the run-up to a stock offering, belongs in the category of claims that deserve a second source before anyone rebuilds a budget around it.

And the plumbing matters. The Verge notes that OpenAI likely uses Microsoft Bing for a significant part of its searches, along with its own search index, though there is still some speculation that it routes some queries through Google. That is the outlet's own reporting and it is not confirmed by OpenAI. If it is right, even a ChatGPT answer about your business may be built partly on the same web index your Bing and website footprint already feed. We looked at a version of this question in our post on where ChatGPT local results appear to come from, and the short version is the same: the tool answers from places on the web, and the places it reads are not always the ones you control.

What Do Local Customers Say They Do With AI?

A headline about total query volume does not tell a plumber anything. The more useful data is what buyers say they do when they need a local business. The best recent source we know of is BrightLocal's Local Consumer Review Survey 2026, which surveyed a representative panel of 1,002 US adults through SurveyMonkey. BrightLocal then published a separate mini-report on the AI findings, cutting some of the data down to the 455 respondents who had used an AI tool for local business recommendations in the past 12 months.

The top-line finding is large. The share of consumers who used AI to find local business recommendations rose from 6% in 2025 to 45% in 2026. In the survey, AI became the third most used tool for local business recommendations, behind only Google and Facebook, and ahead of Yelp and TripAdvisor. Google reviews are still widely used, but the share who used them slipped from 83% to 71%. BrightLocal links that drop to AI tools shifting consumer behavior, with the caveat that it describes this as something many people are speculating about, not something it proved.

On which tools, ChatGPT is the clear frontrunner. BrightLocal reports that 31% of consumers had used it for business recommendations in the last 12 months, followed by Google's AI Mode at 23% and then Gemini. Microsoft Copilot and Claude trail, but they are being tried at a far higher rate than a year ago. That fits OpenAI's claim without proving it: a survey of 1,002 people about their own past year and a company's count of its own queries are different instruments pointed at the same shift.

Age matters here, and BrightLocal is specific about it. Adults aged 30 to 44 lead, with 64% having asked AI tools for a business recommendation in the past year. People over 60 are the most cautious, at 24%. If your customers skew older, your numbers will look different from the national picture. If you sell to homeowners who are in their thirties and forties, such as the people who hire lawn care, remodelers or a dentist, they sit near the top of that range. That last part is our reading, and it is a prompt to check your own customer mix and not a measured finding.

We should be careful about a limit on this data. It is self-reported, national, and from a survey taken earlier in 2026, so it describes what people say they did. It does not describe how often it happened, what they asked, or whether the answer changed who they called. We have seen the 45% figure repeated in many places this year. It is real survey data, but it is a single survey, and a single survey is a thin base for a budget. We cover how different AI tools answer the same local question in our post on why ChatGPT and Gemini pick different local businesses.

Do Customers Trust What ChatGPT Tells Them?

Mostly the people who use it do, with a habit of checking. BrightLocal reports that among active AI users, nearly two-thirds, 63%, trust AI tools' recommendations and only 10% express distrust. Among people who do not use AI, 53% say they do not trust its business recommendations. Trust follows use, which is what you would expect from any new tool, and it means the group of people who rely on AI is more comfortable with it than the headline skepticism suggests.

The more useful finding for an owner is what those users do next. BrightLocal found that 88% of AI users check either whether a review is legitimate (51%) or look at the source (37%), and only 12% do not check sources. It also found that 97% of AI users sometimes double-check AI recommendations against real reviews, and 42% always check reviews on the native review platforms. BrightLocal's own reading is that AI has become the first step in local business research, not the final authority.

That reading matters because it changes what you are optimizing for. If ChatGPT were the last step, the goal would be to get named. Since it is often the first step, the goal is to get named and then hold up when the customer checks. A person who gets your business from ChatGPT and then finds three reviews, a profile with last year's hours and a website that disagrees with both has just been given a reason to call someone else. We wrote about that second step in our post on why many customers do not trust AI's business picks, and this survey is a good reminder of how consistent the pattern is.

BrightLocal also asked about AI-written summaries of reviews. It reports that 50% of consumers trust AI platforms to accurately summarize reviews from real people, rising to 71% among active AI users, and that 23% are happy to rely on a summary alone while 59% check the review profile for more detail. If a summary is how a customer first meets your business, the themes in your reviews are doing the work of an ad. Our guide to how reviews affect SEO and AI visibility goes through what to ask customers to mention.

Is Being Visible Only on Google Still Enough?

BrightLocal's CEO and co-founder, Myles Anderson, made a pointed argument in the mini-report, and it is worth reading as an argument and not as a measured result. He wrote that businesses operating with a “Google-only” mindset are at high risk of missing customers and revenue, and he cited a 12% drop in consumers relying solely on Google for reviews, with platforms such as TikTok, Apple Maps and ChatGPT seeing double-digit growth as business discovery platforms.

His warning is the sharpest line in the report: “If your reputation only exists on Google, you are effectively invisible to the millions of people using ChatGPT to find local services.” He frames the fix as a “Reviews Everywhere” strategy, meaning authority built across Google plus the sites where your customers and AI models spend their time. He also made a careful concession: it is not that consumers are abandoning Google, and usage of Google AI Mode and Gemini is surging. The journey from idea to purchase has simply become more fragmented.

BrightLocal sells local AI visibility software, so read the claim about Google's reviews being out of reach of other AI tools as the company's position, and verify it by testing your own business. We cannot confirm from our own reading how every AI tool treats Google reviews. What we can say is that the mini-report's own description of how AI tools work supports the same direction. It says AI responses may draw on reviews, local directories, business websites, social media and other third-party sources, depending on the tool and the prompt, and that the reasoning is not always clear.

So, is Google-only enough? Our answer is no, but not in the dramatic way the headline invites. Google is still where the largest share of local searching happens, and Google Business Profile still deserves your first hour of work each month. The shift is that a second set of places now matters for the first impression: your own website, a few directories that are accurate, and reviews that exist outside a single platform. Our post on making your website the source of truth for local AI search covers the website half of that.

What Should You Check This Month?

Nothing here requires new software or a new budget. It is a one-hour exercise that you can repeat each month, and it gives you your own local data instead of a national estimate. Items marked as ours are our suggestions, not findings from the sources above.

One note on measurement. Google's Search Console reports on Google's own AI features, and we explain what that report can and cannot tell you in our post on Search Console's AI performance reports. We found no equivalent report from OpenAI for local businesses. That is why the manual check and the intake question matter: they are the only way for a small business to see ChatGPT's side of the picture today.

On advertising, The Verge reported that Fry said ads have not hurt the product or its usage, and that “most people do not seem to mind ads in the product.” We have written separately about the ChatGPT ads geo-targeting beta for Fort Wayne small businesses. Nothing in this announcement gives a price or a local result, so it is not a reason to move budget this month.

How Does This Look for Different Local Businesses?

We do not have local measurements for Northeast Indiana, and no source we read breaks the numbers down by trade. What follows is our own application of the national findings, so use it to decide what to test first.

• Trades such as roofing, masonry, HVAC and plumbing: People often ask for a name when something is broken and urgent, which is the kind of up-to-date place lookup Fry described. Test “who fixes a leaking roof near me” style questions and see who is named. A service-area business with a thin website has less for an AI tool to read than one with a clear page per service and town.
• Lawn care and landscaping: Demand is seasonal, and homeowners in the 30 to 44 range, who BrightLocal found most likely to ask AI for recommendations, are a large part of this market. Check that your seasonal services and service area are stated plainly on your site, not only on a social profile.
• Dental and medical practices: Fry specifically named health questions as a category where people want up-to-date, factual answers. For a practice, that makes accuracy the priority: insurance accepted, new-patient availability, hours and the providers on staff. We would also be cautious about anything an AI tool says about a clinic, and check it monthly.
• Restaurants: Hours, menus and ordering links change often and are easy to get wrong across platforms. A diner whose hours differ between Google, a delivery app and the website gives an AI summary three answers to pick from.
• Retail shops: The simple questions, such as what you carry, where you are and when you are open, are the ones that AI tools answer from public pages. If your site lists brands and categories in readable text, there is more to work with than a page of images.

The common thread is dull and useful: consistent facts, written plainly, in more than one place. None of it is a trick, and none of it depends on whether OpenAI's one-fifth figure turns out to be exactly right.

What Do We Still Not Know?

A fair amount, and the honest list is longer than the confident one.

• We cannot verify the one-fifth figure. It comes from one OpenAI executive, in one conversation, with no stated method and no named source for Google's volume.
• We do not know how many of ChatGPT's search-like queries are about local businesses. Fry named people, places, things, finance and health, but gave no breakdown.
• We do not know how many Northeast Indiana customers use ChatGPT to find a business. The BrightLocal data is national and self-reported.
• We could not read the full text of the Big Technology interview past its opening, so a few details may sit in parts we did not see. We relied on the portions we read and on the other four outlets quoting the same conversation.
• We do not know whether ChatGPT usage of this kind changes who a customer finally calls. BrightLocal's data shows people verify, but it does not show conversion.
• We found no study showing that any specific step in our checklist raises how often ChatGPT names a business. The steps are consistent with BrightLocal's guidance, but they are not proven fixes.

The cheap thing to do is also the sensible one: run the one-hour check, fix what it shows, and ask customers how they found you. In three months you will know more about your own market than any national number can tell you.

Sources & Further Reading

• 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”: bigtechnology.com (published November 4, 2026, read November 5, 2026): the primary interview with Adam Fry, source of the one-fifth quote, the 1 billion to 1.2 billion weekly users figure, and the image and voice numbers. Only the opening portion was readable to us.
• 9to5Google (Ben Schoon), “ChatGPT has hit roughly 20% of Google Search volume, OpenAI says”: 9to5google.com (November 3, 2026, read November 5, 2026): the “wiggle room” comment and the note that Google's volume source was not revealed.
• The Verge, “OpenAI says ChatGPT is catching up to Google in search”: theverge.com (read November 5, 2026): the definition of a search, the Bing 2023 trial comparison, the note on Bing and OpenAI's own index, the ads comments and the possible IPO timing.
• Gizmodo, “ChatGPT Claims to Have Achieved 20% of Google's Search Volume”: gizmodo.com (November 4, 2026, read November 5, 2026): the “search-like queries” definition, the 16 billion searches per day arithmetic and the over-90% market share note.
• Search Engine Roundtable (Barry Schwartz), “OpenAI Says ChatGPT Is At 20% Of Google Search Volume”: seroundtable.com (November 4, 2026, read November 5, 2026): Fry's answers on what counts as a search, the Glenn Gabe comparison to Bing and Kantrowitz's note that Bing has around 5% share.
• BrightLocal, “Nearly Half of Consumers are Asking AI for Business Recommendations” (Local Consumer Review Survey 2026 mini-report): brightlocal.com (read November 5, 2026): the 45% and 6% figures, ChatGPT at 31% and AI Mode at 23%, trust and fact-checking numbers, age breakdown, and Myles Anderson's comments. Methodology: 1,002 US adults via SurveyMonkey, with some data limited to 455 AI users.