Even though ChatGPT ads are still in the experimental phase, they already reveal how advertising could change when people stop searching and start having conversations with AI.
Millions of people use ChatGPT to compare products, research purchases and ask for recommendations. It was only a matter of time before advertising followed them there.
OpenAI has started testing ads inside ChatGPT, and the first results have been a mix of excitement and frustration. Early advertisers ran into challenges such as limited inventory and basic reporting, but they also discovered people engage with ads differently mid-conversation than they do mid-search.
That single observation is why ChatGPT ads matter beyond their own platform. They're a preview of how AI-powered advertising could evolve everywhere.
ChatGPT Ads Are Still in Their Early Days
For anyone who feels late to ChatGPT advertising, you're not. The platform itself is only a few months old, and it's already changed shape several times.
Since its launch, OpenAI has steadily expanded the platform with:
- self-serve Ads Manager;
- CPC and conversion-optimized campaigns;
- conversion tracking with Automatic Advanced Matching;
- AppsFlyer and Adjust integrations for mobile attribution;
- average daily budgets with automatic budget pacing;
- geographic exclusions;
- refreshed product feed ads with pricing and star ratings.
In addition, it dropped the original $200,000 minimum commitment in May 2026, and CPA bidding went live in June, turning the pilot into a real performance channel.
In other words, advertisers are joining a platform that is taking shape before their very eyes.
That also explains why the first wave of advertisers had such mixed experiences. Some reported promising performance, while others struggled to spend even a fraction of their allocated budgets.
The reason for the latter came down to not enough ad opportunities to actually deliver against.
Why Early Advertisers Hit a Wall
Some of the brands that joined OpenAI's early pilot committed hundreds of thousands of dollars and then watched their campaigns sit almost idle.
| Expectation | Reality |
|---|---|
| Plenty of inventory | Limited ad opportunities |
| Campaigns spending as planned | Significant underdelivery |
| Detailed reporting | Basic measurement tools |
| A mature advertising platform | An evolving beta product |
The clearest example of this, according to Digiday, comes from an advertiser who spent only $2,500 out of a $250,000 commitment during the first four weeks of the pilot. Others reported similar experiences of facing weeks of near-zero delivery before things picked up.
The real issue was a mismatch in expectations. Marketers came in expecting Google- or Meta-level inventory, where delivery starts almost instantly. OpenAI, with well over half a billion weekly users, made that assumption reasonable. In reality, only a small slice of conversations were eligible to show ads at all, so available impressions were far scarcer than the user count suggested.
That's already changing. Advertisers report fill rates climbing steadily over recent months, and by April 2026, ads were appearing in roughly 26.5% of replies globally and 49% in the U.S., a sharp jump from the pilot's early trickle.
Why ChatGPT Can't Show Ads Everywhere
At first glance, the solution may seem obvious. In fact, you’re probably wondering, if advertisers want more impressions, why not just show more ads? Because that's one of the fastest ways to ruin the user experience.
ChatGPT isn't a feed or a results page. People go there for personal questions and conversations that can run several minutes long, which raises the bar for every ad shown. Before an ad appears, OpenAI reportedly checks whether the conversation is safe for advertising, whether the ad fits the topic and whether this is even the right moment to show one. Only after clearing all three does an ad enter the auction.
Each of those checks protects the experience and shrinks the inventory pool at the same time, which is exactly why advertisers hit the wall described above. Take a job-interview conversation for example. A course on resume writing fits naturally, whereas a protein powder ad doesn't.
OpenAI's own ad policy reflects a shrinking inventory, excluding ad placement near politics and mental-health topics and banning dating, alcohol, drugs and gambling as categories outright.
That restraint costs OpenAI inventory today. Long-term, it may be the thing that keeps users trusting AI recommendations enough for ads to work at all.
AI Conversations Create a Different Kind of Intent
This is where ChatGPT ads become genuinely interesting. Think about how people usually search for products.
A Google search might look like this:
best noise-cancelling headphones
A ChatGPT conversation is more likely to sound like this:
I work from coffee shops a few days a week, have a budget of around $300 and I need headphones with a great microphone for video calls. What would you recommend?
It's the same category, but the context is completely different.
By the time an AI assistant suggests a product, the user has already shared their budget, priorities, concerns and what they're trying to achieve. That's far richer context than a short search query can provide.
That richer context also makes targeting more nuanced. Unlike search keywords, conversations often reveal the specific pain points someone is trying to resolve. Two users looking at the same product category may have completely different intent. For advertisers, understanding conversational intent means looking beyond keywords and considering where the user is in their decision-making journey.
In other words, AI understands what people are looking for and why.
Despite a much smaller audience, some ChatGPT campaigns delivered efficiency close to Google's non-brand search. Fewer people saw the ads, but the ones who did were much further along in the sales funnel.
Of course, that doesn't mean AI advertising will replace search anytime soon. Search is still unmatched when it comes to scale.
However, AI conversations allow advertisers the chance to reach users while they're actively thinking through a decision rather than simply looking for information.
That also raises the bar for relevance. In search, matching the query is often enough to earn a click. In conversational AI, users expect responses to reflect the context they've already shared. Ads that fail to acknowledge that context may feel interruptive rather than helpful.
What Advertisers Should Do Today
Waiting for the platform to mature is the wrong instinct here. Competition is still low, and the brands testing now are the ones who'll know how to spend well once inventory catches up. Best Buy, Lowe's and VistaPrint are already among the early advertisers running campaigns on OpenAI's self-serve Ads Manager, which officially launched on July 22, 2026.
A few priorities are already becoming clear.
Start Small and Treat It as a Learning Channel
The first generation of AI advertising is about gathering insights rather than maximizing ROAS. Test different industries, audiences and messaging. Pay attention to where AI conversations naturally fit into your customer journey.
Rather than trying to maximize reach immediately, focus on situations where your product or service genuinely helps solve the user's problem. Early success is likely to come from matching meaningful customer moments instead of simply targeting broad categories.
Rethink Your Creative
People use ChatGPT to find answers, making ads a secondary part of the experience. That changes what good advertising looks like. Content that explains, compares or helps solve a problem is likely to feel far more natural than aggressive promotional copy.
The closer your ad feels to a useful recommendation, the more likely users are to engage with it.
Continue the Conversation After the Click
The landing page matters just as much as the ad itself. Someone clicking after a detailed AI conversation arrives with a specific problem already in mind. Sending that visitor to a generic product page or broad lead form can create friction if the page doesn't address the question that prompted the click.
Advertisers should think of landing pages as an extension of the conversation. The closer the page aligns with the user's original intent, the more likely it is to convert.
Measure What Matters
As AI platforms mature, success may be measured differently.
Alongside CTR and conversions, advertisers will likely pay more attention to signals like engagement quality, conversation context and how AI influences purchase decisions throughout the journey.
Early advertisers have also highlighted the importance of validating platform-reported performance against independent measurement. Website analytics, CRM data and actual sales outcomes provide a more complete picture, particularly while reporting capabilities continue to evolve.
Stay Flexible
Perhaps the most important advice is also the simplest.
Features and ad formats will change, so be open and adaptable in this evolving landscape. What works six months from now may look very different from what works today.
ChatGPT Ads Is Becoming a Serious Performance Channel
The latest platform updates show that OpenAI is no longer focused solely on introducing advertising to ChatGPT. It is building a mature performance marketing platform. Features such as conversion optimization, advanced attribution, automated budget pacing and mobile measurement are standard capabilities on Google Ads and Meta Ads. Their arrival in ChatGPT Ads signals that OpenAI is rapidly closing the functionality gap while maintaining its unique conversational advertising experience.
The Conversation Has Only Just Started
ChatGPT ads offer an early glimpse of where digital advertising may be headed. OpenAI reportedly expects about $2.5 billion in ad revenue in 2026, while Google has said Gemini has no ads and no current plans for them. This makes OpenAI the main company testing this model today.
For advertisers, understanding how people talk to AI may soon matter as much as understanding how they search. The brands that succeed won't necessarily be those with the biggest budgets, but those that know when their message genuinely adds value to the conversation.
