Adthena, for example, whose 400-strong client roster includes a number of household names, such as The Home Depot, Dell and Inspire Brands, has noticed the caliber of brands OpenAI has targeted. 

“The only clients we’ve heard that have been approached are the biggest of the big,” Adthena’s CEO Phillip Thume said. He added that OpenAI has reached out to “multiple” clients, with more being approached everyday. “We don’t really have small clients, but even our medium-big sized clients have not been approached.”

Those talks have largely centered on OpenAI asking marketers to commit a share of their total search budgets, according to two ad execs familiar with the meetings. In some of those instances, talks started around a $200,000 minimum commitment. In others, agencies said the number could be lower or higher depending on how negotiations shake out.

Whatever those figures net out to, the dollars will fund a test period in the U.S. lasting several weeks, beginning on February 6, according to ad execs who have spoken to OpenAI. During that period, marketers will buy sponsored placements at the bottom of answers on the free and Go tiers of ChatGPT at $60 per 1000 views. That rate is comparable to targeted streaming and premium TV inventory like NFL games, media buyers say, and well above Meta’s typical CPMs, which often come in under $20. 

Marketers might assume those prices come with a deep bench of performance data. They would be mistaken. Advertisers will receive view and click-through metrics, according to documents OpenAI has shared with advertisers that Digiday has reviewed. Beyond that, key details about the test remain murky; where ads appear, the context surrounding them, which prompts trigger them, what tracking information might be available or what future metrics OpenAI themselves will provide, or any indication of long-term vision – questions marketers have pushed back to their agencies after raiding them with OpenAI. 

For some marketers, that lack of clarity is enough to prompt a pass, said one ad exec whose clients have opted out. For others, the appeal of learning the platform ahead of rivals – and the cachet of being early on the industry’s newest stage – outweighs those concerns, particularly with the trial kicking over this coming Super Bowl weekend, when query volume could surge. 

A sign of things to come

Looking ahead, it’s clear that marketers are approaching these tests with measured expectations. They do not expect to walk away with a complete playbook for advertising on ChatGPT. But they will get a feel for the environment – and, for now, that’s enough. 

Longer term, it will not be. Because while the idea of buying their way into ChatGPT conversations is magnetic, the mechanics behind that process are far less so, especially the way the ads are packaged and priced.

Here’s why: a CPM model typically lends itself to brand advertisers and sponsorships, rather than the cost-per-click (CPC) model traditionally used for search – something which has created hesitancy among some advertisers because there’s no guarantee of what return they’ll get.

With a CPM, advertisers pay for impressions rather than actions, meaning visibility is guaranteed but engagement and conversions are not. And that’s what makes returns harder to measure or predict. Ironically, it’s the same metric that Perplexity launched its ad business with back in November 2024. At the time, Digiday reported that search advertisers were less than keen because it isn’t the standard method by which search ads are usually bought, and allows them to model performance.

Adthena’s Thune made the point that by viewing it as a CPM, advertisers are effectively “taking a step back from the conversion metric they care about.”

What they’re not stepping back from is Google. While OpenAI is targeting search-linked dollars to fuel its test, few marketers see that coming out of Google’s budgets. If anything, many are trying to spend more, not less, with the incumbent. To them, OpenAI’s early grab for search-linked dollars looks more like a way to anchor expectations than a direct raid on the category. Whether that framing holds will depend on how quickly OpenAI builds the infrastructure of an ads business that, for now, doesn’t even have a formal ads chief. 

Paid social media expert Shamsul Chowdhury noted that OpenAI is looking to tap into the mid-funnel, an area that has always been a bit elusive with Google.

“They know that if they can win the mid-funnel, it plants the seeds to compete with Google in the lower funnel where Google is king,” he said. 

Narratives move markets

Unsurprisingly, OpenAI is being very cautious about how it introduces ads. Look at how deliberate it has tried to frame them from the outset, clearly aware of the risks that come with monetizing interactions inside a consumer-facing product. Advertisers are seeing that caution up close in the guidelines governing the test phase, which underscores just how careful the company is trying to balance commercial ambition with user trust. 

OpenAI is tightly scripting how partners talk about the ChatGPT ad test, positioning it as a user-experiment rather than a formal product rollout. Messaging, it said, should emphasize learning, gradual scaling and user benefit, while steering clear of language that suggests a broad “ad launch”. Partners are instructed to stress that answers are not influenced, ads are clearly labelled and conversations remain private. They cannot discuss pricing, performance data, deal terms or internal results, and are expected to route sensitive questions back to OpenAI. Materials require pre-approval and communications are time to OpenAI’s signal that the test is live. 

As eMarketer’s principal analyst, AI in marketing and commerce Nate Elliott pointed out, OpenAI, like all AI platforms, is terrified of scaring away users. 

“We forecast there’ll be 40 million new genAI users in the U.S. alone between now and 2029, and it’s vitally important to OpenAI that they attract the lion’s share of those new users, but they know people don’t love advertising,” he said. “It explains why they would encourage their ad buyers to talk so carefully about how the trial is working. They desperately need ad revenue, but equally they need to avoid the perception that ChatGPT is lousy with ads.” 

Careful or not, OpenAI cannot afford to move too slowly. Competition for ad dollars is already fierce among established platforms, let alone new ones. That pressure is only intensifying as more companies look to advertising to help fund the soaring costs of building and running AI systems. 

Executive editor of video and audio Tim Peterson, senior media buying editor Michael Burgi and senior reporter Sam Bradley contributed reporting to this article.

Editor’s note: An earlier version of this article misstated the minimum ad spend commitment OpenAI was seeking. The figure has been corrected.