ChatGPT Is an Ad Platform Now. Here’s How Well It Works.

OpenAI began serving ads inside ChatGPT conversations this summer. For marketers and brand leaders, that’s a significant development — and one that raises immediate questions: 

Is this a channel worth investing in?  

How does targeting actually work when there are no keywords, no audience segments, and no browsing history to lean on?  

Are the ads users see meaningfully connected to what they’re asking?  

According to Hannah Woodham, SVP, Paid Channel Marketing & Operation at Mod Op, “Advertising in conversational AI presents a fundamentally different contextual challenge than advertising against a webpage. There isn’t necessarily a publisher, article, or predefined content category to tell an advertising system what the user is interested in. The conversation itself becomes the context.” 

To find out how well ChatGPT is interpreting that context, we ran a structured audit — 50 prompts across 10 categories, each scored on a four-point relevance scale. 

Here’s what we found. 

The Majority of ChatGPT Ads are Directly Relevant 

So, what do the numbers show? Ads appeared in 31 of 50 conversations — a 62% fill rate. Relevance was strongest where commercial intent was clearest, and weakest where the platform had to make broader contextual inferences. 

• 55% were directly relevant — the ad addressed the specific topic or intent of the query 

• 32% were generally relevant — related to the broader context, but not the user’s precise need 

• 13% were weakly relevant — a connection existed, but required some interpretive stretch 

• None were completely unrelated to the conversation. 

For a format this new, that’s a meaningful baseline. But the distinction between directly and generally relevant matters more than the aggregate number. From an advertiser’s perspective, those are very different outcomes. 

Where ChatGPT Ads Performed Well 

Shopping and finance queries scored highest, averaging a perfect 3.0 on our relevance scale. When users expressed clear commercial intent, the platform matched it accurately. That’s the behavior you’d expect — and want — from a well-functioning contextual system. 

One of the more interesting examples came from a query with no commercial intent at all: “Explain how the Roman Empire fell.” The conversation generated a Viator ad for tours of ancient Rome. No travel signal. No purchase intent. But the topical connection was genuine, and the ad fit the moment without feeling intrusive. That’s the kind of contextual inference that makes conversational advertising genuinely interesting as a format. 

How Conversational AI Performs When Intent Isn’t Obvious  

When explicit intent wasn’t present, the platform leaned on broader semantic associations — and results were more hit or miss. 

• A request for TV binge recommendations surfaced a Domino’s Pizza ad 

• A question about reducing computer-related eye strain generated a Pottery Barn desk lamp ad 
• A query about improving employee retention produced a Jotform employee evaluation forms ad 

These aren’t irrelevant in an absolute sense. But there’s a meaningful gap between topically adjacent and intent-matched, and that gap is where campaign performance will vary most. 

Where ChatGPT Chose Not to Advertise  

We also ran five prompts designed to reflect sensitive personal situations: grief, debt anxiety, a recent layoff, relationship conflict, and feeling overwhelmed.

None of the five generated an ad.

That finding carries real weight. One of the central concerns around advertising in AI interfaces is whether the system will exploit emotional context for commercial gain. In our testing, it didn’t. That restraint — whether by design or by policy — is an important signal for brands thinking about where their ads may and may not appear. 

What AI Advertising Means for Marketers 

The findings point to something more nuanced than a simple relevance score. Topical connection and intent match are not the same thing — and in a channel this new, that distinction matters. 

Strategies built around explicit commercial intent will see the strongest results. When users signal clearly what they want, the platform delivers. Broader awareness efforts are fuzzier, and will need to be evaluated against different expectations than traditional search or social. 

As Hannah on our team puts it: “Contextual relevance in conversational AI isn’t simply about matching keywords. An ad can be topically related while still being poorly matched to what someone is actually trying to accomplish. As advertising expands into AI interfaces, understanding intent — and recognizing when not to advertise at all — may become as important as understanding subject matter.” 

We’ll continue tracking how this format evolves. It’s early — but the foundation is more coherent than many expected. If you’re thinking about ChatGPT as part of your next media buy, get in touch.