ChatGPT now handles roughly 20% of Google's search-like query volume — but the websites of the world received just 0.36% of their visitors from it in October. Adam Fry, who leads ChatGPT's consumer product at OpenAI, said at a company-hosted dinner in New York on November 2: "ChatGPT's volume of search-like queries is now roughly one-fifth the volume of queries of Google." By OpenAI's definition, that covers "basically anything that a user would have traditionally gone to a traditional search engine for."

It is a striking number. It is also one the company declined to ground in specifics. OpenAI did not say which Google figure it compared against, over what period, or in which countries.

Does big query volume mean big referral traffic?

No — and the gap is the story. Ahrefs tracks referral traffic across 118,671 websites; in October, Google sent 24.74% of their visitors while ChatGPT sent 0.36%. Both figures can be true simultaneously, because a question answered inside ChatGPT never has to send anyone to a website at all.

Money is moving regardless. Media buyers told Digiday this month that clients are questioning 10 to 30% of their Google search budgets as AI answers occupy more of the results page. Whether that reallocation is warranted is a separate question.

What does this change for B2B vendors?

Less than the headlines suggest, according to 6sense — a vendor of account-based marketing (marketing targeted at specific companies rather than individuals) and buyer intent data, and therefore a party with its own stake in the answer. Its 2026 B2B Buyer Experience Report surveyed buyers behind more than 4,600 recent B2B purchase journeys across North America, Europe, the UK and Asia-Pacific.

The findings cut against the "buyers have moved to AI" narrative:

• 81% of buyers used an AI model somewhere in their evaluation — down from 94% a year earlier, not up.
• Only 26% used AI to find new vendors.
• Before ever speaking to a salesperson, buyers credited their own existing knowledge first (34%) and internal company discussions second (32%); AI tools ranked at 21%.
• Buyers made first contact with a seller roughly 59% of the way through a buying cycle averaging just over 10 months.
• By that point, 91% had already ranked their shortlist — and the pre-favored vendor won 82% of the time.

That last figure deserves attention. The battle appears to be largely decided before any conversation, AI-assisted or otherwise.

The data does not necessarily contradict G2's widely shared finding from earlier this year that 51% of B2B software buyers now start their research in an AI chatbot. A buyer can open ChatGPT to learn what a category of software does, or to summarize three vendors' websites, and still end up choosing between names they already knew.

Why AI keeps surfacing the same names?

Because it reflects reputation rather than building it. Ahrefs analyzed 10,000 US brands across 260 million AI prompts that did not name the brand, and found that how often a brand appeared in AI answers tracked closely with how often it was mentioned elsewhere online.

Ahrefs concedes it cannot prove the mentions cause the visibility, and that brand size explains part of the pattern. For business leaders, that caveat is arguably the point: AI tends to name the companies people already talk about. When B2B leaders are asked where their best deals came from, the answers remain referrals, conference conversations, or someone on the buying team who used the product at a previous job — and the 6sense data suggests that pattern has shifted far less than the AI search coverage implies.

What did OpenAI itself launch?

On November 10, OpenAI opened a beta called Businesses in ChatGPT, letting companies claim a profile in front of what it says are 1.2 billion weekly users. Its own FAQ states plainly that "claiming does not guarantee higher placement or sales" and that ChatGPT "continues to make independent recommendations and comparisons." A vendor-endorsed tool that promises no advantage is, at minimum, honest about its limits.

The practical steps the reporting recommends are correspondingly modest:

• Protect what builds the shortlist. When 6sense asked buyers how they keep up with a category between purchases, vendor websites, industry events, review sites and industry publications led the list; AI models ranked last of 11 channels at 15%. Any budget moved into AI visibility should be asked which of those channels it comes out of.
• Claim the business in ChatGPT this month. Any eligible business with a website, in a country where ChatGPT is offered, can verify its profile through OpenAI's Business Hub, which also shows mention frequency, clicks and conversation themes — the first AI visibility figures from OpenAI rather than a third-party estimate. Business Agents (chat between customers and companies inside ChatGPT) remain limited to US ecommerce and online service businesses; claiming is not.
• Keep the rest of the verification clean. State plainly what you sell, who it is for, and roughly what it costs. Monthly, run your ten most important buying questions through ChatGPT, Gemini and Google, and fix anything wrong at its source.
• Ask customers where the shortlist position came from — including the ones who chose someone else. Within months, that tells you more than any AI visibility dashboard can.

ChatGPT handling a fifth of Google's searches is a real shift, and it will change how buyers research. But for most B2B purchases, the research AI assists happens after the shortlist exists. The mistake is treating AI as the front door and defunding the things that put a company on the list in the first place.