OpenAI wants ChatGPT to be the place where people ask questions and, increasingly, where brands answer them. Early advertisers are finding the second half of that plan harder than it looks.
Results from the ChatGPT advertising pilot are all over the map. Some marketers say their acquisition costs beat search, while others say they are paying for clicks that never become a sale.
The recurring complaint is targeting. Ads are not consistently reaching the right users at the right point in a conversation, and that gap is showing up in the conversion numbers.
The numbers behind the mixed reviews
As of late 2026, ads appear on approximately 25.94% of commercial prompts. Those are queries with some kind of buying or shopping angle, which is the territory advertisers care about most.
Around 14% of ads have limited or no connection to the user’s query or the conversation thread.
Ad impressions also often reach users who show no clear purchase intent. Someone researching a topic out of curiosity is not the same as someone ready to buy, and the platform does not always tell the difference.
Click-through rates have landed between 0.5% and 2%. Cost-per-click ranges from under $3 to over $10, a spread wide enough that two advertisers could have very different experiences on the same platform.
One advertiser spent $477 over six weeks, collected 370 clicks, and recorded zero conversions.
Outcomes like that have pushed some marketers to treat ChatGPT ads as an experiment rather than a core channel. In practice, that usually means a small test budget, close monitoring, and little appetite for scaling until the data improves.
The case for patience
The picture is not uniformly grim. Some advertisers, particularly those working through partnerships, have posted competitive cost-per-acquisition figures.
For certain partners, CPA was reported to be 15.3% lower than paid search benchmarks.
OpenAI claims a 112% improvement in click-through rates over one tracking period. The company has also rolled out improved measurement tools and new creative formats in response to advertiser feedback.
The platform reached a claimed $1 billion annualized revenue run rate in under 200 days of launch.
Why conversational ads are a different animal
OpenAI launched the pilot in early 2026 with an ambitious goal: rethinking how ads get delivered inside conversation-based AI. That is a different problem from placing an ad next to a search result.
The 14% figure for loosely connected or unrelated ads is the clearest sign of that difficulty. When nearly one in seven ads misses the conversation, users learn to tune them out, and low click-through rates follow.
What this means for advertisers and OpenAI
There are a few signals worth watching. The first is whether the share of irrelevant or loosely connected ads falls from around 14%, since relevance sits upstream of nearly every other metric.
The second is whether OpenAI’s claimed improvement in click-through rates carries through to conversions. More clicks only matter if some of them turn into customers.
The third is whether results from partnership deals, where CPA has looked competitive, start to show up for ordinary self-serve advertisers. If the strong numbers stay concentrated among a handful of partners, the platform will look less like a new advertising channel and more like a set of custom arrangements.
Disclosure: This article was edited by Diego Almada Lopez. For more information on how we create and review content, see our Editorial Policy.