The ad agency Jellyfish is in the process of introducing its first client to the platform’s pilot program. Jai Amin, the agency’s head of media activation, noted that although the initial rate remains at $60, he is observing average costs closer to $45, which can vary based on inventory composition.

Other advertising professionals report even lower figures. An anonymous executive mentioned that they’ve seen CPMs fluctuate between $35 and $25 in recent days when purchasing ads through Criteo. In some instances, prices have fallen significantly, with Ad Age citing rates as low as $15, although it remains uncertain if these rates were also processed through Criteo.

Typically, variations like these are attributed to auction mechanics: high demand tends to raise prices while low demand decreases them. However, ChatGPT currently lacks a live auction setup. A more significant indicator is the lowering of the entry fee; the minimum investment required to engage has decreased from $250,000 at launch to just $50,000. This adjustment opens the door for a broader array of advertisers, consequently broadening the pricing spectrum.

“It seems like everything is declining, gearing up for a broader auction accessibility to facilitate [a] global rollout,” noted Ashley Fletcher, the CMO of Adthena, whose clients have been involved in the pilot since it began on February 9.

Still, a CPM range of $25 to $45 is considered high for a platform that is largely viewed as experimental. This situation isn’t unprecedented; Netflix introduced its ad-supported tier with CPMs between $55 and $65 in 2022, which within a year dropped to around $20 to $30 as inventory levels increased.

However, Netflix had the luxury of time; it offered a clear value that users recognized and trusted. Viewers enjoyed high-quality content, and ads were simply the price of maintaining a more affordable subscription. The value proposition was straightforward, whereas ChatGPT’s allure partly lies in the belief that its responses are unbiased — suggesting that no payments are involved when it promotes a product or service. Introducing advertising complicates this perception, regardless of OpenAI’s assurances about separating paid promotions from organic responses. Until OpenAI demonstrates that advertisements within a conversational framework provide enough value to users leading to measurable results for advertisers, the discussions around CPMs remain somewhat theoretical.

“Google and Meta have a stronghold on industry measurement, and that’s the real advantage, not the user interface,” said Robert Webster, founder of the AI marketing consultancy TAU. “Measuring after a click is straightforward. However, post-view data at a $60 CPM is challenging. Until an independent party can verify the actual worth of a ChatGPT impression, advertisers are taking OpenAI’s estimates at face value. This is where the $102 billion forecast stands or falls.”

To illustrate how steep this uphill battle is, it’s worth looking at the broader market landscape.

According to Gupta Media, Facebook averages $4.82, Instagram $7.63, Google Display $10.33, and TikTok $3.02. These are established platforms with years of scaling, robust measurement systems, and demonstrated return on investment. A more indicative comparison might be LinkedIn, which currently averages $39.19 — a price point advertisers accept as high due to the context justifying the cost. OpenAI is putting forth a similar rationale, but it hasn’t had the time to validate it yet, although the company is acting swiftly to do so.

Models such as cost per action and cost per click are being developed alongside conversion tracking, which would provide advertisers with the attribution framework that the current pilot lacks. Should these be successfully implemented, the CPM discussion may become less critical, with advertisers focusing instead on whether the performance data stands up.

Few platforms have evolved this quickly. In nine weeks, OpenAI has transformed from having no advertising presence to creating a self-serve ads manager, establishing a partnership with Criteo, significantly reducing the entry barrier, and appointing a dedicated ads leader in David Dugan. This urgency reflects the competitive landscape, with OpenAI anticipating $2.5 billion in revenue this year while Google and Meta continue to advance. The projected advertising revenue of $102 billion for 2030 leaves little room for a gradual approach. If performance metrics are favorable, pricing will naturally adapt.

OpenAI did not respond to inquiries from Digiday.