OpenAI turns on cost-per-click ads inside ChatGPT

“As OpenAI looks to increase spending, this is going to help advertisers directly compare their results on Open AI to other major advertising platforms and manage their spend,” said Nicole Greene, vp, analyst at Gartner. “In a world where everything is changing due to AI tech and consumer behaviors, this consistent measurement will help advertisers justify reallocation of spend to OpenAI.

It was always a question of when, not if, OpenAI made this move. Yes, the company launched its ads on a CPM model, but that was more because it was a lighter lift. After all, it requires no click-tracking infrastructure and brand advertisers are easier to onboard on a new platform with limited measurement capabilities. But CPMs also have a ceiling. Performance advertisers account for the majority of online ad spend – and prefer to pay for clicks not impressions. Keeping them on an impression model indefinitely was never a serious option. That much was clear once OpenAI started testing its ads manager, with CPCs clearly labeled in an early build of it.

“OpenAI’s experimentation with CPC is driven largely by its need to maintain demand growth and build trust with advertisers, though declining CPMs and an expanding pilot are also factors,” said Claire Holubowskyj, senior research analyst at Enders Analysis. 

Put another way, advertising on ChatGPT is already getting cheaper. The rate advertisers pay to reach every thousand users has dropped from $60 at launch 10 weeks ago to as low as $25 in some cases now – and the trend is downward. When impression prices fall, so does the revenue each one generates. CPC gives OpenAI a way to grow ad revenue that doesn’t depend on holding CPMs up. 

The deeper challenge is what OpenAI is measuring itself against. CPC is Google’s home turf – one that Google has spent years building the presently most sophisticated version. Its bidding tech prices ever click based on intent signals, quality scores, auction pressure and retargeting data. CPCs and Google Search continue to grow quarter on quarter. Advertisers, in other words, keep getting evidence that they’re worth it. 

OpenAI will need to make the same case which is harder than it sounds. The tricky part is framing the value of a click since they’re not all equal. For instance, Meta’s CPCs run three to five times cheaper than Google Search, according to digital ad agency Adthena, not because the inventory is necessarily worse but because the intent behind them is different. Social users are browsing. Search users are looking for something specific. That gap in intent is why Google commands a premium, and where OpenAI lands within it will determine what its clicks are actually worth.

“LLMs are now starting to bridge that gap, as intent begins to build through the back-and-forth of prompt-driven conversations,” said Ashley Fletcher, CMO at Adthena, which has several clients testing ads on ChatGPT. 

The search for OpenAI’s first advertising marketing science leader is, needless to say, an urgent one. 

A job posting gives a sense of the scope. The person hired will own OpenAI’s advertiser measurement strategy from the ground up. Defining how its reporting connects to attribution models, incrementality testing, media mix modelling and geo experimentation. They’ll build the marketing science itself, starting as an individual contributor before scaling a team and operating model around them. 

A significant part of the role is external-facing, with the person leading measurement conversations with advertisers, showing up in executive briefings and translating statistical methodology – causal inference, attribution models, experimental design to name a few – into language that informs actual campaign decisions. The job also involves working with third-party measurement partners, clean-room providers and industry groups to reduce friction to advertisers trying to evaluate OpenAI against the rest of their media buys. 

Internally, they’ll partner with product and engineering to turn measurement methodology into durable tooling and reporting infrastructure. Privacy is, perhaps unsurprisingly given how much it’s been a focus up to now, is built into the brief from the start. The role calls for establishing aggregated measurement and conversion approaches in partnership with legal and privacy teams. 

The job vacancy is the latest sign that OpenAI is moving super fast. To compare, Uber ads didn’t hire Edwin Wong, its first head of measurement, until 2025, around three years after it had introduced an ad business. Even Netflix only started building out its measurement team in 2023 – about a year after its ad business launched.

OpenAI did not respond to Digiday’s request for comment.