On the heels of introducing ads to ChatGPT in the United States, OpenAI is now preparing to launch mid-conversation ad recommendations in the U.K. and other countries.
The ChatGPT maker announced earlier this month that “strong interest from businesses looking to reach users in a more conversational, intent-driven environment” had encouraged it to expand the offering. The ads will arrive “in the coming weeks,” OpenAI said.
But how are British media buyers preparing for the rollout—and how will ChatGPT fit into their broader media buys?
“This is a pivotal moment in digital marketing ads in general,” said Jai Amin, chief solutions officer, media activation at Jellyfish. With one foot on both sides of the Atlantic, Jellyfish has been able to bring its experience of testing the ads in the United States to its U.K. clients.
Pilot program successes
Amin said the U.S. offer initially launched for select advertisers, with premium CPMs around $60 and pilot commitments of roughly $250,000 to $300,000. Since launch, the product has already matured from “passing sheets back and forth” to a more recognizable ad platform with CPC buying, conversion tracking pixels and daily campaign budgets, Amin said.
The early experience has been broadly positive, Amin added. In practice, CPMs have been closer to $30, and Jellyfish has seen early conversions and high-order values. Still, he stressed that the first wave required clients to “take a bit of a leap of faith,” treating the channel as brand awareness rather than something to be judged immediately against the usual performance metrics.
The number of those who have been willing to take that early leap of faith is limited, admitted Amin. At the moment, only a couple of Jellyfish advertisers are seriously committed to ChatGPT ads, but he expects that to rise to around 10 globally by the end of July.
In the U.K., he expects at least three or four advertisers to take part initially. For those doing so, the rough baseline is a commitment of at least $100,000 over three months—though he noted the campaigns were not necessarily spending quickly. Brands might end up spending only $50,000 to $60,000 over that period, due to things like OpenAI’s deliberately slow rollout of ads, inventory restricted to Free and Go-tier users, and a narrow slice of conversations that match commercial intent.
Caution shaped by unanswered questions
Not everyone is ready to envision ChatGPT ads as a core performance channel. Willis Annison, digital performance director at Bountiful Cow, said the format doesn’t map neatly onto existing media categories. At first glance, he said, it looks “kind of like search.” But because advertisers are giving contextual hints rather than bidding directly on specific keywords, it feels closer to programmatic. “It doesn’t feel super-focused on the bottom of the funnel like search would,” he said.
Despite that, Annison said Bountiful Cow is advising clients to fund early tests from experimental budgets rather than diverting spend from always-on activity. “We wouldn’t take it with the business-as-usual budget,” he said, adding that the channel was too new to be held to strict performance KPIs. Instead, he sees it as a mid-funnel play, useful for consumers who are using ChatGPT for research, comparisons and recommendations.
Annison said ChatGPT ads could work a little like paid search, supporting weak SEO performance by helping brands show up where they are not already being cited or referenced by LLMs. But for now, clients need to test whether the channel adds something incremental, or if it’s destined to be just another touchpoint on the path to conversion.
That caution is further shaped by unanswered questions over measurement, brand safety and user trust. Annison said the brand safety discussion could be “quite interesting” because advertisers don’t yet have the same level of control or verification they would expect from search, social or programmatic.
“Brand safety is always going to be a question, and you kind of have to take it at face value, that it will be brand-safe,” he said. There is also a consumer trust issue; people may become more careful about what they tell ChatGPT if they know advertising data is being passed back to brands, he said, echoing concerns raised by U.S. advertisers.
The bigger prize, though, may be less about conventional ads than about brands becoming part of the AI interface itself. Amin said the formats currently feel “not as different as you’d want,” with limited creative options. But he expects the experience will be about “brands being visible and interacting on LLMs, rather than having ads send people to brand sites,” he said. “I think that’s the ultimate goal.”