OpenAI ad business faces ROI test as ChatGPT seeks permanent media budgets
After a rapid push that drew more than 100,000 advertisers into testing ChatGPT Ads, OpenAI now faces pressure to prove the product can deliver consistent business results. Early evidence points to uneven performance across brands, even as some marketers shift spending from trial budgets into broader growth plans.
Highlights
- OpenAI's ChatGPT Ads achieved a $1 billion annualized revenue run rate in about nine months but face scrutiny over long-term ROI versus Google.
- A working paper found ChatGPT Ads generated $0.89 revenue per U.S. session, significantly trailing $2.45 per session for Google search ads.
- Adthena reported 52% of 30 enterprise brands struggled to recover ChatGPT ad spend, while 26% matched or outperformed Google, with measurement and scale as key hurdles.
Performance gap shapes next phase
As reported by Business Insider, OpenAI has scaled ChatGPT Ads to an annualized revenue run rate of $1 billion within about nine months of launch, but marketers and researchers are still questioning whether the platform can justify a long-term place in media plans.A working paper under review by e-commerce researchers Maximilian Kaiser and Christian Schulze found that across 102 advertiser sites, ChatGPT Ads generated $0.89 in revenue per U.S. session, versus $2.45 for Google search ads. The researchers said the size of that gap makes them cautious about assuming performance will improve naturally over time.
Adthena also shared results from 30 enterprise brands covering July through September, saying 52% struggled to recover their ChatGPT ad spending, while 26% matched or outperformed Google ads. OpenAI has pointed to stronger examples as well, including an October case study with Evernote that found trial-to-paid conversion rates ran about 10% to 20% above the app's earlier paid-acquisition average for the year.
Audience limits and measurement remain hurdles
Marketers say the platform's next challenge is scale, as ChatGPT's ad business remains constrained by the size of its free user base, the limited pool of ad-relevant conversations, and the number of ads users are willing to accept before leaving the product.Qualcomm CMO Don McGuire said audience targeting is still difficult unless campaigns are extremely broad, which raises questions about return on investment. Kaiser and Schulze said one possible way to improve performance would be to place ads more naturally inside answers rather than beside conversations, though that appears unlikely because OpenAI's ad principles say ads will not influence ChatGPT's responses and that user experience takes priority over revenue.
Measurement is also a sticking point for advertisers. Adthena CMO Ashley Fletcher said OpenAI's newer tools, including a measurement pixel and conversion bidding, should help brands better assess cost per acquisition and cost per click, but cleaner data may also confirm that ChatGPT works more as an upper-funnel nudge than as a quick-conversion channel.
Some advertisers are nevertheless finding early traction. Natural Intelligence, which operates comparison sites such as Top10.com and BestMoney.com, has moved ChatGPT Ads from a test budget into a growth budget, according to head of strategy Jonathan Vardi. He said performance improves when marketers build ads around specific customer pain points and feed accurate post-click conversion data back into the platform so its systems can learn which users become paying customers.
In earlier coverage, we examined a KPMG survey on AI agents showing efficiency gains at big U.S. firms, even as transformation lags.