Advertising on ChatGPT brings stores less revenue than Google Search
The study compares returns from ads in AI chat and in search engines using sales data from 102 online stores. The paper has not yet undergone scientific peer review, but it provides a reference point for companies considering shifting their budgets.
Maximilian Kaiser (Vienna University of Economics and Business) and Christian Schulze (Frankfurt School of Finance & Management) published a working version of the study “Can Advertising Fund Free AI?” in October 2026. It shows that, in U.S. traffic, a visit from a paid ChatGPT ad brought stores an average of $0.89 in revenue, while a visit from a paid Google Search ad brought $2.45. Conversion, meaning the share of visits ending in a purchase, was 0.37% and 1.17% respectively. The paper is awaiting peer review, and its findings were described, among others, by Search Engine Journal (opens in a new tab).
The authors analyzed Google Analytics data from June 29, 2025 to August 22, 2026. Ads in ChatGPT launched in the U.S. on February 9, 2026, and the European Union, where they only appeared on August 24, served as a comparison group. The data was provided free of charge by Grips Intelligence, where Kaiser is a scientific advisor.
Of the 87 advertisers in the U.S. buying both types of ads, 84 earned less per visit from a ChatGPT ad than from a search ad visit. 73 had no sales at all from those ads despite a total of 18,726 visits. One advertiser accounted for 66% of all purchases from ChatGPT ads.
The authors calculated that advertising in ChatGPT would deliver a return equal to search if the cost per click (CPC) were about $0.26, with an estimated range from $0.05 to $0.57. The largest publicly known advertising account paid $1.72 per click, based on 15 days of spending disclosed by the agency Opascope. OpenAI recommends starting a campaign with a maximum bid of $3–5.
Data calculated using other metrics is more favorable for OpenAI. On August 31, the company said that one online store achieved a threefold return on ad spend in 28 days, and according to DoubleVerify, the customer acquisition cost for Weight Watchers was 15.3% lower than its benchmark. At the same time, OpenAI acknowledges that it does not yet have general performance metrics for its ads. The study is observational: advertisers themselves decided to buy ads in ChatGPT, so the results describe their outcomes, but do not prove that the ads caused them. The data also covers only the last click before the visit.
Companies can calculate the profitability threshold using their own data: revenue per visit from ChatGPT should be divided by revenue per visit from search and multiplied by their own Google cost per click.