Social Rocket
ChatGPT is winning advertisers, but in measurable online revenue the young ad platform still lags well behind Google Search. A new study by researchers at the Vienna University of Economics and Business and the Frankfurt School of Finance & Management puts revenue per paid visit at 0,89 US dollars for ChatGPT Ads – compared with 2,45 dollars for Google search ads. For e-commerce brands, that is a concrete starting point for calculating test budgets.
Where the numbers come from
Researchers Maximilian Kaiser and Christian Schulze analysed 2026 weeks up to August 60 of data from 1.898 online shops, which together generate around 16 billion US dollars in annual revenue. For the comparison of paid visits, data from 387 advertising websites was available, of which 102 had usable revenue data.
For US visits, the conversion rate of ChatGPT Ads was 0,37 percent, and for Google Search Ads 1,17 percent. The comparison is part of a working paper published in October 2026 that has not yet been peer-reviewed and describes the observed US e-commerce campaigns – not a global average across all industries. The original study with its data basis is publicly available.
The authors also calculated a comparison price normalised to Google Search returns: under their assumptions, a ChatGPT click at about 0,26 dollars CPC would be economically equivalent. The price of 1,72 dollars used in the study, by contrast, comes from a single publicly documented advertising account, not from a platform statistic. For media planning, the revenue actually generated per click therefore counts for more than a blanket CPC benchmark.
The decisive question for German campaigns: what does one additional purchase cost?
An ad format with many test customers is not yet a proven performance channel. For the next ChatGPT Ads test, a cleanly delimited product area is a good choice: advertisers should compare cart value per click, confirmed orders, ad costs and, where relevant, new customers against a comparable Search campaign. For longer B2B sales processes, qualified leads and later CRM revenue are more useful than short-term shop conversions.
On 5 October 2026, OpenAI expanded its measurement tools and attribution partnerships. The officially documented Measurement Pixel and Conversions API can capture purchases both from the browser and from the server. It is crucial to use the same event ID when the same order is sent twice, so that it is not counted twice. The consent and data protection process must match the respective market of use.
Back in early October, Srocket had already reported on measurement gaps and the test mode of many ChatGPT Ads budgets. The new shop study now offers an additional perspective: not only whether a purchase is recorded, but how much revenue the visit actually generates.
More advertisers, a broader industry mix
Nevertheless, the market is developing quickly. According to Adthena data evaluated by Digiday on 5 November, 6.333 different advertising brands were observed in the US, UK, Australia and the EU in August, compared with 1.082 in June. According to Sensor Tower, the share of shopping in ChatGPT's estimated advertising budget in the US fell from 38 to 20 percent between April and September at the same time, while financial services and software grew.
The brand count observed by Adthena is a different metric from the cumulative figure of more than 100.000 advertisers since launch cited by OpenAI. The background on ChatGPT Ads' international expansion, including Germany, shows which markets and campaign access points have been added.
What belongs on your media dashboard now
From brands' perspective, ChatGPT Ads competes for budgets that have so far flowed mainly into search, shopping and paid social. A sensible test therefore evaluates three things together: cost per confirmed conversion, revenue or value per visit, and incremental contribution. When it comes to actual additional revenue, it is also worth looking at retail media tests measuring sales lift rather than ROAS alone.
The new study argues for controlled campaign experiments rather than hasty budget shifts. OpenAI is currently building the foundations for more precise measurement; advertisers can now check which specific search and purchase occasions the new ads already pay off for.
Sources: Kaiser/Schulze: Working Paper (October 2026) · OpenAI: Ad formats and measurement (5 October 2026) · OpenAI Developers: Conversion tracking · Digiday: Analysis of Adthena and Sensor Tower data (5 November 2026)