OpenAI built an AI that answers before you click. Its ad model depends on the click.

OpenAI is running a bet on advertising inside a product that makes advertising's core premise optional.

The company launched cost-per-click ads inside ChatGPT this month, adopting Google's pricing model for a product where the answer arrives before you click anything. The structural tension is not subtle: advertisers are buying clicks in a conversation where the value was supposed to be the answer, not the link. OpenAI charged $60 per thousand impressions when it launched its ad product in February. Within ten weeks, the effective rate had fallen to as low as $25, The Next Web reported, as click-through rates came in at 0.91 percent, roughly seven times below Google Search benchmarks, according to AdWeek.

In traditional search, you click a link and get an answer. In ChatGPT, the answer arrives before the click. So what exactly are advertisers buying?

OpenAI hired Shivakumar Venkataraman, a 21-year Google veteran who led Google's search ads business, as vice president in June 2024, The Next Web reported. He knows the playbook. The question is whether the playbook works inside a product that was designed to make the click optional. Sam Altman spent two years building a public position against advertising. He called it a momentary industry in 2024 and, at Harvard, described it as a last resort. The reversal is complete: OpenAI has cut the minimum ad spend from $250,000 to $50,000, according to The Next Web, opening the product to a wider range of advertisers who can now test whether a click inside a conversational answer converts at rates that justify the bid.

OpenAI projects $2.5 billion in ad revenue this year, growing to $100 billion by 2030, according to documents reviewed by Reuters. The projections face a credibility problem: click-through rates at 0.91 percent are running roughly seven times below Google Search benchmarks, and OpenAI has already cut its minimum ad spend threshold by 80 percent to attract more advertisers. Whether the product can demonstrate conversion attribution that justifies the CPC pricing is the open question.

Advertisers cannot see individual user conversations, chat history, names, email addresses, or IP addresses. They receive only aggregated performance data, The Next Web reported. This privacy structure limits the targeting precision that makes Google's ad machine work, and it limits what OpenAI itself can tell advertisers about whether the product is delivering ROI.

The counterargument, and it is a real one, is that a click inside a high-intent conversational query is worth more than a click from a web search. Someone asking ChatGPT which laptop to buy is closer to a purchase decision than someone scrolling social media. If that premise holds, the 0.91 percent click-through rate may be a misleading metric and the $25 cost per thousand impressions may be underpriced.

OpenAI is betting it holds. The company has committed to an advertising model its own founder called a last resort, built by a former Google ads executive, inside a product designed to eliminate the need for links. The CPM collapse suggests the market is not yet convinced. The revenue targets suggest OpenAI cannot afford to wait for it to figure itself out.