How do ChatGPT ads work?
ChatGPT ads are selected based primarily on the context of a user’s conversation in ChatGPT.
When someone enters a prompt, OpenAI’s ads system uses the conversation context to understand the topic someone is researching or asking about, runs a brand safety check, identifies ads that may be relevant based on the topic, landing page, headline, and ad copy, and then uses a relevance model to decide which ad is the best match.
According to OpenAI, at the ad group level, advertisers can provide “context hints” that describe the topics, conversations, or keywords where their product or service may be useful. These “hints” help guide matching, but they don’t work like exact-match search keywords, similar to the kind advertisers may be used to in paid search campaigns, and don’t guarantee that an ad will appear in a specific conversation. OpenAI still decides final ad delivery based on the context of the full conversation and how relevant the ad is to the user’s request.
Targeting is currently based on contextual relevance and the geographic areas marketers choose for their campaigns. Ads can be matched to conversations based on what someone is asking about, while geographic targeting helps control where those ads are eligible to appear.
For example, if someone asks ChatGPT about the best running vests for a marathon, they may see an ad from a running apparel brand or retailer with nearby stores or relevant product options. If multiple advertisers are eligible, OpenAI selects the ad most relevant to the chat to show first.
Currently, users may see a single ad unit below a ChatGPT response when there’s a relevant match, and that unit may include one or more items from the brand.
If ad personalization is turned on, OpenAI may also use past chats, memories, and previous ad interactions to make ads more relevant. Users can manage ad personalization and memory controls in settings, or use Temporary Chats to avoid using or updating memory.
ChatGPT advertising limitations and restrictions
Ads don’t influence the answers ChatGPT gives users. The ad system is separate from the chat model, which means advertisers can’t shape, rank, or alter ChatGPT’s responses. Advertisers also don’t receive users’ chats, chat history, memories, names, emails, precise location, IP address, or sensitive information.
OpenAI also applies placement and content policies designed to prevent ads from appearing in sensitive user contexts or brand-unsafe environments. As a result, early access to ChatGPT ads is generally better suited for industries with fewer regulatory restrictions, while some categories may be limited or unavailable based on OpenAI’s current platform guidelines and ad policies.
What are the creative best practices for ChatGPT ads?
ChatGPT currently supports one ad format. Each ad uses a square image asset with a 1:1 aspect ratio and 256×256 pixel resolution, along with a headline of up to 30 characters and body copy of up to 60 characters. Advertisers can also add UTMs to the target URL to support campaign tracking (more on that below).
Because the ad format is more compact compared to larger, highly visual channels like connected tv, display, or digital out-of-home, creative needs to be clear, specific, and closely aligned with the conversation where the ad may appear.
Here are a few ChatGPT advertising best practices to keep in mind:
- Align creative and messaging with the topic: ChatGPT ads are selected based on context, so the headline, description, image, and landing page should all connect clearly to the user’s prompt or query.
- Focus on user needs, not just keywords: Context hints help guide your ads to the types of conversations where they’re most relevant and may be eligible to appear. Instead of listing broad standalone keywords, describe the questions, needs, or situations someone might ask in ChatGPT. Keep each ad group focused on a single product category, theme, or use case to avoid matching against conversations that aren’t closely related to your ads.
- Keep the value proposition concise: With only 30 characters for the headline and 60 characters for the body copy, the ad needs to communicate the offer quickly. Avoid vague copy or messaging that requires too much explanation.
- Submit multiple headlines and copy variations: Rather than relying on a small number of variants, advertisers should turn major themes, category terms, and product use cases into multiple headline and description combinations to provide more relevant options to match against the topics and context behind different conversations.
- Double down on early traction: Once certain variants begin showing stronger performance, use those learnings to refine future creative and focus on the messages that are most closely aligned with user intent.
In doing so, advertisers can improve the chances that their ads feel relevant, useful, and aligned with the conversations where they appear.
How are ChatGPT ads measured?
Advertisers can currently access reporting either through ChatGPT Ads Manager if they buy directly through OpenAI, or with support from a partner like StackAdapt if they onboard through one. This includes core campaign metrics like impressions, clicks, click-through rate, average CPC, average CPM, amount spent, and conversions.
Advertisers can also add UTM parameters to their landing page URLs, allowing traffic from ChatGPT ads to be tracked in existing analytics tools. For conversion measurement, OpenAI supports pixel-based measurement and Conversions API, which helps marketers understand actions people take after engaging with an ad, such as a purchase or sign-up.
As measurement for ChatGPT ads continues to evolve, partners like StackAdapt can help marketers set up tracking, interpret results, and better understand how ChatGPT ads contribute to conversions alongside other channels, rather than measuring their impact in isolation.
What does advertising on ChatGPT mean for advertisers?
To succeed in digital advertising today, advertisers need to connect with customers not just where they’re spending time online, but in the moments that shape what they do next.
Increasingly, those moments are happening inside AI tools like ChatGPT, with EMARKETER reporting that AI chatbots are expected to help 63.3 million US shoppers in 2026.
ChatGPT ads give advertisers a way to reach people during their everyday moments where they’re researching, comparing options, and making decisions.
Given the early-stage nature of the format and the ongoing evolution of the platform, advertisers will need to experiment and identify how this fits into their broader media strategy.
StackAdapt helps advertisers navigate that learning curve with strategic guidance, creative support, and operational expertise, making it easier to test, learn, and scale confidently in an emerging channel.