OpenAI has put a number on something marketers have been guessing at for two years: how much of ChatGPT's traffic actually behaves like a search engine. The company says the queries it sends out to the web add up to roughly a fifth of Google's search volume — meaning Google still handles about five times as many searches.
The comparison counts a specific slice of ChatGPT activity: questions that trigger a live web lookup rather than being answered from the model's own training data. That's an important distinction, because a large share of ChatGPT conversations never touch the open web at all — they're drafting, coding, summarizing, or brainstorming tasks that don't generate the kind of traffic a website owner could capture or measure.
Third-party traffic data tells a messier story than OpenAI's internal figures. Analytics firms that track referral traffic to websites have shown ChatGPT sending a small but fast-growing trickle of visits compared to Google, with wide variation depending on the measurement method and which bot activity gets counted. The gap between OpenAI's self-reported numbers and outside measurement reflects a broader problem: there's no agreed-upon standard yet for what counts as a "search" when an AI chatbot is involved, which makes any single statistic easy to cite and hard to verify independently.
This isn't the first time a company has tried to frame its product's scale against Google's. Search market-share comparisons have been a recurring talking point since ChatGPT's search features launched, with each side's numbers tending to favor whoever is publishing them. What's notable here is that OpenAI itself is now the one supplying a direct ratio, rather than letting outside analysts estimate it.
The pattern fits a wider trend of AI companies racing to position their tools as alternatives to traditional search, even as usage data suggests a more complicated reality: coexistence rather than replacement, at least for now. Google has responded by layering AI-generated summaries directly into its own results pages, blurring the line between "search" and "AI answer" rather than ceding ground.
For small businesses, the practical issue isn't which company wins the numbers debate — it's where potential customers are actually forming opinions about products and services before they buy. If a meaningful and growing slice of people are asking ChatGPT for recommendations instead of typing into Google, then how a business's information appears in AI-generated answers starts to matter alongside traditional search rankings.
That means revisiting the basics that make a business easy for any system, human or AI, to understand: accurate business listings, clear product descriptions, up-to-date hours and pricing, and a website structured in plain language rather than buried in jargon or flash. AI tools tend to pull from and synthesize existing web content, so a business with thin or outdated information online risks being invisible or misrepresented in an AI-generated answer, even if it ranks fine on Google.
The trade-off is effort versus uncertainty. Optimizing specifically for AI chatbot visibility is still a new and poorly measured discipline, with no equivalent yet to the decades of established search engine optimization practice. Businesses investing heavily in it now are, to some degree, betting on a channel whose size and durability are still being argued over by the companies that built it.
Watch for two things in the coming months: whether independent analytics firms converge on a consistent way to measure AI-driven web traffic, and whether Google's own AI summary features start pulling click-through traffic away from websites in ways that show up in site analytics. Both will say more about the real shift in customer behavior than any single company's self-reported ratio.
The bottom line: the specific 5-to-1 figure matters less than the underlying trend it points to — AI chatbots are handling a growing, if still modest and hard-to-measure, share of the queries that used to go straight to a search engine.