ChatGPT ads are now measurable against an established performance channel, and the first substantial outside-platform comparison gives advertisers a reason to be disciplined. In U.S. ecommerce traffic, paid ChatGPT visits generated $0.89 in revenue per session, compared with $2.45 for paid Google Search, according to a new working paper by Maximilian Kaiser and Christian Schulze. The findings are early, but they shift the discussion from where ads appear to what a paid visit is worth.
ChatGPT’s paid visits trail search on conversion and revenue
The October 2026 working paper, “Can Advertising Fund Free AI? Evidence from ChatGPT’s International Advertising Rollout”, analyzes first-party ecommerce data across 1,898 websites over 60 weeks. The full study covers nearly 10 billion sessions, while its paid-channel financial comparison uses a much smaller group: 102 advertiser sites with complete sales tracking.
For U.S. traffic, paid ChatGPT visits converted at 0.37%, versus 1.17% for paid Google Search. Among 87 advertisers that bought both channels, 84 earned less revenue per ChatGPT visit than per search visit. Seventy-three recorded no ChatGPT-ad-attributed purchase across 18,726 visits; the other 14 sites accounted for all 570 purchases in that comparison. Those figures suggest the average is not just being dragged down by a few weak tests. The shortfall was widespread in this sample, as Search Engine Journal’s review of the paper also notes.
The $0.26 figure is a test hypothesis, not a universal bid cap
The researchers estimate that a ChatGPT click would need to cost about $0.26 to deliver the same revenue per ad dollar as paid search. That is a modeled parity price, not a published market-wide rate or a guarantee of return. The calculation uses a $0.72 median search click cost from just 16 U.S.-dominant advertiser accounts and assumes similar profit margins across the two channels.
Results also vary sharply by advertiser. One site accounted for 76% of U.S. revenue from paid ChatGPT visits in the financial sample. That concentration means a standout case can make the channel’s aggregate performance look much stronger than a typical advertiser may experience. Businesses should compare their own qualified leads, sales, margin and repeat purchases rather than treating either the $0.89 average or $0.26 estimate as a forecast.
What the study does not prove about AI advertising
This is a version-one working paper under review, not a final industry benchmark. The paid-channel comparison is descriptive: advertisers chose whether to buy ChatGPT ads, so the study does not prove that the placement caused the observed difference. It uses last-click website attribution, which can miss later purchases or influence that happens outside the measured visit. Its findings are based on ecommerce sites and should not be transferred directly to every business model.
The paper also examines whether ads reduced unpaid ChatGPT referrals. The baseline estimate was a 5.9% decline in organic referrals during the first 13 weeks, but its confidence interval included no change. Adjustments for assumed faster ChatGPT adoption in North America produced larger estimated declines of 7.2% to 8.4%; the result therefore depends on those assumptions. Conversion among organic ChatGPT visitors did not measurably change. The researchers summarize the underlying tradeoff this way: “The central challenge for AI assistants is whether advertising can become valuable enough to finance the product without making the product less useful.”
How businesses should test ChatGPT ads now
For marketers, the practical conclusion is not to ignore a new channel or to scale it on novelty. Run a bounded test with a clear comparison against paid search and other acquisition sources. Keep landing pages, offers, geography and measurement windows as comparable as possible, and check that analytics distinguish paid ChatGPT visits from organic referrals and other traffic.
Set a maximum cost per qualified outcome from your own margins and conversion economics. Evaluate completed purchases, booked appointments or sales-qualified leads, not impressions or clicks alone. Allow enough time to observe the full buying cycle, but pause or revise campaigns when spend outpaces verified value. A channel can improve as its targeting, inventory and optimization mature; today’s evidence simply says that advertisers should demand proof before they fund that improvement at scale.
Real Internet Sales helps businesses evaluate emerging AI marketing channels, build reliable measurement, and turn digital strategy into accountable growth. Call 803-708-5514 or visit realinternetsales.com to discuss a practical plan.