Ads in ChatGPT open a discovery channel, but results need testing
On August 11, 2026, OpenAI said it had started rolling out ads in ChatGPT in Brazil, as well as the United Kingdom, Mexico, Japan and South Korea. The country also gained self-serve access to Ads Manager, according to the company itself.
The novelty creates a discovery point where a person can explain budget, features and context over the course of a conversation. For marketing, the question becomes twofold: is the offer useful for that specific need, and can the company measure what happens after the exposure?
Contextual discovery isn't an organic answer
In a conversation, the need can take shape over several messages. The offer needs to make clear what problem it solves, who it makes sense for and in what situation it should be considered. Generic messages tend to waste the chance to connect.
According to OpenAI's information, ads are labeled and appear separate from ChatGPT's answers. So buying media doesn't mean appearing inside an organic answer or changing the content generated by the system.
Privacy also shapes the campaign design
OpenAI says advertisers don't receive users' conversations, history, memories or personal data. Performance is presented in aggregate, with metrics such as views and clicks. This limits targeting expectations: individual context shouldn't be treated as an asset available to the company.
In practice, the message, the landing page and the conversion need to carry much of the relevance. The campaign should explain the offer without depending on information the advertiser can't access.
The channel still needs to prove results
The launch and the materials consulted don't offer consolidated benchmarks by sector, format or advertiser. The novelty, therefore, isn't evidence of return. The test needs to answer whether the channel generates a qualified action at a cost that fits the offer.
A campaign can produce a view or a click without generating a contact, a sale or progress in the sales process. Measure the complete journey and compare the result with a reference period or audience whenever the design allows.
Start with a controlled experiment
Before investing, record the hypothesis, the audience the platform allows, the landing page and the event that represents commercial progress. That way, the test separates discovery, click and result, stages that shouldn't be confused.
Define an offer, a qualification criterion based on what the platform actually allows and a measurable conversion, plus a budget and a maximum cost per result. Record what happens after the click and set in advance the criterion for keeping, adjusting or stopping the campaign.
The decision criterion doesn't need to be sophisticated: compare the cost per qualified conversion with the offer's economic limit and track the progress rate after the click. The test may conclude that the channel isn't right for the company. That possibility is part of a responsible media decision.
Sources
OpenAI, 9 Feb 2026. https://openai.com/index/testing-ads-in-chatgpt/
OpenAI. https://help.openai.com/en/articles/20001047-ads-in-chatgpt
OpenAI. https://help.openai.com/en/articles/20001207-ads-in-chatgpt-the-basics
OpenAI. https://help.openai.com/en/articles/20001245-ads-manager-availability
OpenAI, 5 May 2026. https://openai.com/index/new-ways-to-buy-chatgpt-ads/