OpenAI Launches Cost Per Click (CPC) Ads in ChatGPT

OpenAI has just taken a decisive step in the construction of its advertising model: after testing ads at CPM (cost per thousand impressions), the platform now activates ads at CPC (cost per click) inside ChatGPT. A strategic shift which transforms the chatbot into a real performance channel, and which places OpenAI in head-on competition with Google Search.

From CPM to CPC: why this change was inevitable

When OpenAI launched its first ads, the choice of CPM went without saying. This model, which allows advertisers to be billed for each thousand impressions served, is simpler to set up. It does not require click tracking infrastructure and allows you to quickly onboard advertisers focused on brand awareness, even on a platform that still has limited measurement capabilities.

But CPM has a ceiling. Performance-oriented advertisers, those who manage budgets based on concrete results, prefer to pay for real actions, not for displays. However, this segment represents the majority of online advertising spending. Keeping these advertisers on a print model indefinitely was not a viable option.

Price pressure also accelerated things. In just ten weeks, ChatGPT's CPM went from around $60 at launch to $25 in some cases, with a downward trend. When the price of impressions falls, the revenue generated by each of them automatically decreases. CPC offers OpenAI a lever for advertising growth that no longer relies on artificially maintaining CPMs.

According to screenshots of the Ads Manager interface, verified by Digiday, advertisers can nowset bids between $3 and $5 per click. This format coexists with the CPM model already in place, and is gradually deployed via a limited version of the advertising manager.

The move to CPC fundamentally changes the relationship between OpenAI and its advertisers. Rather than paying to be visible, they only pay an amount when a user actually interacts with their ad. This helps align ad spend with measurable results and calculate ROI comparable to other channels.

As Nicole Greene, vice president and analyst at Gartner, points out: this format will allow advertisers to directly compare their results on OpenAI to those of other major advertising platforms and to more easily justify a reallocation of their budget to ChatGPT.

The real challenge: proving the value of a click in a conversational context

Adopting the CPC means move into Google territoryand the Mountain View firm spent years perfecting this model. Its bidding system leverages accurate intent signals, quality scores, real-time bidding pressure, and retargeting data. Result: advertisers have tangible proof, quarter after quarter, that these clicks are worth this investment.

OpenAI will have to provide the same guarantees. And this is precisely where the task becomes complicated. Not all clicks are equal. On Meta, for example, the cost per click is three to five times lower than on Google Search, not because the inventory is necessarily lower quality, but because the intention behind those clicks is different. On social networks, users browse. On a search engine, they are looking for something specific. It's this gap in intention which justifies Google's pricing premium.

So the question is where ChatGPT will position itself in this spectrum. The argument for OpenAI is the conversational nature of the platform: the intention is gradually built over the course of the exchangesin conversations guided by prompts. This is a different signal of intent than traditional search, but potentially just as valuable.

An advertising infrastructure under rapid construction

To achieve its ambitions, OpenAI is going beyond new formats. The platform is building all the infrastructure necessary to a real advertising business.

The recruitment of a first manager of advertising measurement, a position currently open, is the most visible sign of this. This profile will have to build the measurement strategy for advertisers from scratch: define how the platform's reports work with attribution models, incrementality tests, media mix modeling and geographic experimentation. The hire will also work with third-party measurement providers and industry groups to facilitate advertisers' evaluation of ChatGPT against their other media buys.

The speed at which OpenAI structures itself is a particularly notable point. By comparison, Uber Ads didn't hire its first head of measurement, Edwin Wong, until 2025, about three years after launching its advertising business. For its part, Netflix began building its measurement team in 2023, almost a year after starting its ads. OpenAI is accelerating much earlier in its advertising development curve.

What this changes for advertisers

For marketers, enabling CPC in ChatGPT represents a concrete opportunity to gain early access to a channel with high potential for intentbefore the competition raises the stakes. The first to test the format have a first-mover advantage in understanding the performance mechanisms specific to the platform.

On a budgetary level, CPC makes it easier to integrate ChatGPT into existing media plans. Performance teams now have a familiar measurement framework to justify investments and manage campaigns. Claire Holubowskyj, senior analyst at Enders Analysis, sums up this issue well; OpenAI's CPC experimentation is largely motivated by the need to maintain advertiser demand and gain their trust, in a context where CPMs are already under pressure!