Across U.S. traffic, each ChatGPT ad visit returned $0.89 in revenue, and each one from a Google paid search ad brought in $2.45. Maximilian Kaiser and Christian Schulze report the comparison in a working paper dated October 2026, still under review and not peer reviewed. Search Engine Journal reported the findings on Monday, November 2.
The gap matters because advertisers are already bidding. Any team weighing a ChatGPT budget now has an outside benchmark, even a provisional one, in a channel where, per OpenAI’s FAQ for advertisers, ChatGPT Ads “does not yet have performance benchmarks across advertisers, industries, or campaign types.”
The channel comparison
The authors followed site visits and purchases as OpenAI launched ads country by country. The paid-ad figures rest on 102 advertiser sites where sales were fully tracked. In the U.S., 0.37 percent of ChatGPT ad visits converted, against 1.17 percent of paid search visits. The gaps held when the authors compared both channels within the same sites and controlled for device type and week.
Eighty-seven advertisers ran ads on both ChatGPT and Google search for U.S. visitors, and 84 of them earned less per ChatGPT ad visit. Seventy-three of them recorded no sales from ChatGPT ads across 18,726 visits. The other 14 produced all 570 ChatGPT ad purchases.
That concentration is the study’s main caveat. A single advertiser drove 66 percent of every purchase from paid ChatGPT visits on the 102 sites. Removing it widens both gaps, so the headline figures flatter ChatGPT rather than punish it.
What a click is worth
Kaiser and Schulze define a “parity price” as the price per click where a ChatGPT click would return the same revenue per ad dollar as a paid search click. Using 16 accounts that drew at least 80 percent of search traffic from the United States, they started from a median paid search click cost of $0.72. Scaled by the revenue-per-visit difference, that yields roughly $0.26 per ChatGPT click.
The estimate is loose. One advertiser supplied 76 percent of the U.S. revenue from paid ChatGPT, so resampling the advertisers moved the figure anywhere from $0.05 to $0.57. The calculation also assumes similar profit margins in both channels and counts only the one tracked visit.
Against that sits $1.72 per click, a figure Opascope (a firm managing paid media for clients) reported after 15 days of spend on a single account. The parity estimate is under a sixth of that. For starting CPC campaigns, OpenAI’s basics page for ads recommends capping bids at $3 to $5.
Organic referrals
The authors also checked 1,898 sites for unpaid ChatGPT referrals, as a stand-in for whether ads weakened the product. Over the first 13 weeks, organic ChatGPT referrals fell an estimated 5.9 percent compared with organic search referrals on those sites and with the EU, where no ads ran. The range ran from a 12.6 percent decline to a 1.2 percent gain, so zero is included. After adjusting for faster ChatGPT adoption in North America, the estimate moved to a decline of 7.2 to 8.4 percent, with ranges that exclude zero. Conversion on organic visits that did arrive barely moved, shifting 0.02 percentage points from a 2.21 percent baseline.
The pre-launch check is shaky. Four rollout groups out of five failed it individually. The combined test also raised a flag in 49 of 100 simulations where no ads were shown.
Limits of the data
The numbers come from Google Analytics ecommerce tracking that Grips Intelligence, an ecommerce data company, supplied free. Kaiser advises Grips scientifically and once worked there. According to the paper, Grips played no part in the work and holds no commercial relationship with OpenAI. Attribution stops at the final click ahead of a visit, with no prompts or impressions, and the sample reflects advertisers who chose to buy, so it does not prove the ads caused the results.
The data window ends August 22. Since then OpenAI has extended ChatGPT Ads into 89 countries, and it reports over 100,000 advertisers have run campaigns. More favorable results exist but use other metrics: an ecommerce advertiser reached three times return on ad spend over 28 days, per an August 31 OpenAI post, and DoubleVerify put Weight Watchers’ attributed cost per acquisition 15.3 percent below benchmark.
Advertisers can test the same logic on their own accounts. Divide ChatGPT revenue per visit by paid search revenue per visit, multiply by your search CPC, and compare the result with your actual bids, after filtering ad-review crawler sessions that analytics tools may flag as paid. If your bids sit near the recommended $3 to $5 starting range, the paper suggests tracked returns will trail search unless your own data says otherwise.
Search Engine Journal, reporting on a working paper by Maximilian Kaiser and Christian Schulze, published November 2, 2026.
