ChatGPT's Ad Pilot Is Live, But Advertisers Can't Tell If It's Working

OpenAI is moving forward with advertising inside ChatGPT, and early adopters are not impressed.

The company's ad product, currently in a limited pilot, shares almost no meaningful performance data with advertisers, lacks automated buying tools, and offers minimal targeting capabilities, according to a Search Engine Land report published March 23. Two agency executives told The Information they couldn't prove the ads drove measurable business results for their clients.

That's a problem. Not a minor one.

For a company that builds some of the most sophisticated AI systems on the planet, OpenAI's ad infrastructure is stuck in a different decade. Deals happen over phone calls, emails, and spreadsheets. There's no self-serve platform. No programmatic buying. No attribution model worth mentioning. SEO consultant Glenn Gabe outlined the issues publicly on X, drawing from reporting by The Information.

And yet. OpenAI told advertisers it plans to expand ChatGPT ads to all U.S. users on free and low-cost tiers in the coming weeks. The audience is about to get much larger. The measurement tools haven't caught up.

Here's what early advertisers are working with, according to reporting from The Information and Search Engine Land:

No automated way to buy ad space. Every deal is negotiated manually through calls, emails, and spreadsheets. For performance marketers accustomed to real-time bidding and self-serve dashboards, this is a step backward by about 15 years.

No meaningful performance data. Advertisers can't see the metrics they need to evaluate whether their spend is generating returns. Click-through rates, conversion tracking, viewability scores: the standard toolkit doesn't exist yet in ChatGPT's ad environment.

Minimal targeting capabilities. The sophisticated audience segmentation that makes platforms like Google Ads and Meta's ad platform effective isn't available. Advertisers are essentially buying reach without precision.

No way to prove ROI. Two agency executives confirmed to The Information that they couldn't demonstrate measurable business results from ChatGPT ad placements. For agencies that need to justify spend to clients quarterly, this is a dealbreaker in any sustained budget allocation.

OpenAI has communicated to advertisers that performance may improve if they supply more variations of text and visual creative, per the Search Engine Land report. The company also indicated that ads will soon reach all U.S. free and low-cost tier users. But there's been no public timeline for when self-serve buying, automated optimization, or standard attribution will be available.

The irony is thick enough to cut with a spreadsheet. OpenAI builds some of the world's most advanced AI systems. Its ad reporting tools belong in 2011.

It's tempting to dismiss ChatGPT's ad product as fatally underbaked. But history offers some nuance. Google's early advertising efforts were manual and limited before AdWords scaled into a self-serve juggernaut. Meta's first ad products were crude compared to today's Advantage+ suite. Even TikTok's ad platform was widely criticized as immature during its early rollout before becoming a serious performance channel within three years.

The difference is that OpenAI is entering advertising in 2026, not 2002. Advertisers have spent two decades building expectations around measurable, automated, data-rich ad platforms. Asking them to accept a spreadsheet-and-phone-call buying process in 2026 requires either a very compelling audience or very patient budgets.

ChatGPT has the audience. Over 400 million weekly active users interact with the platform. The attention is real. The question is whether attention without attribution is worth the spend.

OpenAI isn't the only AI company exploring advertising. Perplexity launched sponsored questions and branded answers in late 2025, offering a model where ads are woven contextually into search-like responses. Early reports suggest Perplexity at least provides basic performance metrics including impression counts and click data. Google is embedding Gemini directly into its ad platform with full measurement integration from day one. Microsoft has been running ads in Copilot, leveraging Bing's existing ad infrastructure for targeting and reporting.

Each of these competitors either started with or quickly built measurement capabilities. OpenAI is the outlier in the AI advertising space. It chose to launch advertising before building the measurement layer. That sequence forces early advertisers into the position of paying for an experiment they can't evaluate, a proposition that works only if you believe the long-term strategic value of being early outweighs the short-term inability to prove returns.

Context matters for understanding why OpenAI is pushing ads despite the product's immaturity. The company reportedly faces significant revenue pressure as it works toward profitability. ChatGPT's subscription revenue, while substantial, may not be growing fast enough to offset the enormous compute costs of running the platform. Advertising represents a potentially massive revenue stream, and every month of delay is a month of foregone income.

That pressure may explain the decision to expand ads to all free-tier users before the measurement infrastructure is ready. OpenAI may be betting that advertiser demand for ChatGPT's audience is strong enough to sustain spending even without ROI proof, at least in the short term. It's a risky bet.

Brand awareness budgets have always been harder to measure. For advertisers with upper-funnel goals, being present in ChatGPT conversations could have value that's difficult to quantify but strategically important. Think of it like early podcast advertising: limited targeting, limited measurement, but high engagement and a first-mover advantage in a new medium.

There's also the relationship play. Getting in early with OpenAI's ad team means having a seat at the table as the platform evolves. Advertisers who help shape the product through feedback and testing could benefit disproportionately when self-serve tools and measurement eventually arrive. Several major brands reportedly view their current ChatGPT ad spend as a relationship investment rather than a performance channel.

The sheer scale of ChatGPT's user base provides a basic argument. With 400 million weekly active users engaging in text-based conversations, the platform offers a unique advertising context. Users are actively thinking, researching, and making decisions. That cognitive engagement is qualitatively different from passive scrolling on social media or skimming search results.

For anyone managing a performance budget with ROI accountability, there's simply no case for ChatGPT ads today. You can't optimize what you can't measure. You can't prove returns you can't track. And you can't scale a channel that requires manual deal negotiation for every placement.

The opportunity cost matters too. Every dollar spent on unmeasurable ChatGPT ads is a dollar not spent on channels where attribution exists and optimization is possible. For a demand gen director managing a quarterly pipeline target, that trade-off doesn't pencil out.

One genuinely surprising detail: OpenAI reportedly hasn't even built basic frequency capping into the ad product yet. Advertisers can't control how many times the same user sees their ad in a single session. That's a feature most ad servers have offered since the early 2000s. Its absence suggests the ad product was built quickly, without input from experienced ad operations professionals.

The broader reaction from the advertising industry frames AI advertising as a trust experiment. MarTech.org described the situation as one where "the real risk isn't relevance, it's whether advertisers can verify that their placements are performing in environments they don't fully understand."

That framing applies beyond ChatGPT. As AI platforms broadly introduce advertising, the measurement challenge becomes an industry-wide question. But OpenAI's position is uniquely exposed because it's the largest AI platform by user base and the one with the most advertiser interest, yet the least developed measurement infrastructure.

The next 90 days will determine whether ChatGPT ads become a real channel or remain an expensive experiment. Three signals to monitor closely.

First, watch for self-serve tooling. If OpenAI launches a self-serve buying platform by mid-2026, it signals serious intent to scale. If deals still require phone calls in Q3, advertisers will lose patience and budgets will shift elsewhere. The absence of a self-serve platform is the single biggest barrier to advertiser adoption at scale.

Second, track measurement partnerships. OpenAI could shortcut its measurement gap by partnering with third-party attribution providers like DoubleVerify, Integral Ad Science, or AppsFlyer. Any such announcement would materially change the value proposition for performance marketers. It would also signal that OpenAI recognizes the problem rather than dismissing it.

Third, watch competitor reactions. If Google and Meta begin offering "AI placement" options that mirror ChatGPT's conversational ad format but with their existing measurement infrastructure, OpenAI's window of differentiation narrows quickly. Google's NewFront 2026 announcements suggest the company is already thinking about AI-native advertising. Meta hasn't announced conversational ad placements yet, but the technical capability exists.

The platform has the audience. It doesn't have the infrastructure. Until it does, treat ChatGPT ads as experimental budget only. Not a performance channel. Not a scalable investment. A placeholder for what could eventually become significant, if OpenAI builds what advertisers actually need.