A new round of advertising data shows that traffic from ChatGPT ads converts into far less revenue than traffic from Google Search ads in the U.S. market. For small businesses weighing whether to shift ad dollars toward AI chat platforms, the numbers suggest that move would be premature.
What happened
An analysis of U.S. advertising performance compared revenue generated per visit from Google Search ads against revenue generated per visit from ads served inside ChatGPT. Google Search ad visits produced substantially more revenue per visit. More strikingly, a majority of advertisers running ads through ChatGPT reported no measurable sales from that traffic at all.
This matters because OpenAI has been building out advertising as a revenue stream for ChatGPT, following a familiar playbook: launch a free product, build a massive user base, then monetize through ads once usage is high enough to interest advertisers. ChatGPT has the users — hundreds of millions of weekly active people — but user volume and buyer intent are not the same thing.
Google Search ads work because people type queries when they are ready to act: they want a plumber, a flight, a mattress. That intent signal is what advertisers have paid for since the early 2000s. ChatGPT usage skews toward open-ended conversation, research, drafting, and brainstorming — activities that don't necessarily end in a purchase decision, even when a sponsored link or product mention is inserted into the response.
Why it matters
This fits a pattern seen whenever a new platform tries to bolt advertising onto a product that wasn't originally built around commercial intent. Social platforms went through a similar maturation curve — early ad products on Facebook, Instagram, and TikTok also underperformed search ads on a per-visit basis before each platform spent years refining targeting, ad formats, and measurement tools to close the gap.
The broader AI chatbot field is racing toward monetization right now. Perplexity, Google's own AI Overviews, and Microsoft's Copilot are all experimenting with ways to insert commercial content into conversational answers. Early performance data like this gives a sense of where that race currently stands: conversational AI has reach, but it hasn't yet proven it can replace the purchase-intent signal that makes search advertising valuable.
What this means for small businesses
If you've been approached about advertising inside ChatGPT, or you're curious about reallocating budget away from Google Ads, this data argues for caution rather than a wholesale shift. Google Search ad spend remains the more measurable, more proven channel for driving actual transactions, at least for now.
That doesn't mean ignore ChatGPT entirely. Being mentioned or recommended by ChatGPT in organic, unpaid responses — a discipline some are calling AI search optimization — is a separate and growing concern worth attention, distinct from paid ads inside the chat interface. Optimizing how your business appears in AI-generated answers is a longer-term play with different tactics than running ads.
For now, treat ChatGPT ad placements as a small, measured test rather than a budget migration. Track revenue per visit, not just click volume or impressions, since this data suggests clicks from AI platforms don't carry the same commercial weight as search clicks.
What to watch
Watch for OpenAI's response to performance data like this — expect new ad formats, better targeting options, or revised pricing models within the next two to three quarters, mirroring how social platforms iterated after their own weak early ad results. Also watch whether Google or Microsoft publish competing data on their own AI chat ad products, since each company has incentive to frame these comparisons favorably.
The bottom line
Current data shows Google Search ads generating meaningfully more revenue per visit than ChatGPT ads, with most ChatGPT ad campaigns producing no sales at all. Small businesses should keep core ad budgets in proven search channels while treating any ChatGPT ad spend as a small, closely measured experiment rather than a replacement strategy.