Since OpenAI first began testing advertisements within its flagship product in February, the digital marketing landscape has been forced to grapple with a new, unpredictable variable: the AI-driven auction. Six months into the rollout of ChatGPT Ads, the platform has expanded into a self-service powerhouse spanning 52 countries. However, for media buyers and performance marketers, the transition from "experimental pilot" to "global ad platform" has raised more questions than it has answered.
While advertisers now have the ability to buy, optimize, and measure campaigns with relative ease, they are doing so in an information vacuum. Without industry benchmarks, competitive auction insights, or transparency into how the AI matches intent to creative, the platform remains a high-stakes arena where "success" is often defined by the individual advertiser’s bottom line rather than industry standards.
A Chronology of Rapid Evolution
The evolution of ChatGPT Ads has been nothing short of aggressive. What began as a closed-beta test in February—initially targeting the U.S., Canada, Australia, and New Zealand—quickly outgrew its infancy. By May, OpenAI signaled its intent to scale by expanding into the U.K., Japan, South Korea, Brazil, and Mexico.
By August 31, the company achieved a major milestone: self-service access to the Ads Manager for advertisers in 52 countries. This rapid expansion was accompanied by shifts in the platform’s bidding mechanics. In August, OpenAI pivoted toward a "Maximize results" default bidding strategy for new ad groups. This move signaled a broader shift in the company’s philosophy, pushing advertisers away from manual control and toward an automated, black-box approach to inventory acquisition.
The Data Dilemma: Why Benchmarks Remain Elusive
The most significant hurdle for early adopters is the lack of context. In established ecosystems like Google Ads or Meta, a cost-per-click (CPC) of $10 is easily interpreted through the lens of industry averages, impression share, and competitive bidding density. In ChatGPT Ads, a $10 CPC exists in isolation.
Some advertisers are reporting aggressive, low-cost acquisition, with CPCs dipping below $3. Others, however, are finding themselves paying $10 to $13 per click for traffic that yields little to no conversion. The fundamental problem is that OpenAI has not yet published performance benchmarks across industries, campaign types, or business sectors.
Without an "Auction Insights" equivalent or metrics like "Impression Share," advertisers are flying blind. They know what they paid, but they have no way of knowing why they paid it. Is the price a reflection of high market competition, or is it a result of the AI’s internal relevance-weighting algorithm deciding the ad was a poor fit for the specific conversation? For the modern performance marketer, this lack of diagnostic visibility is a significant barrier to scaling.
The Mechanics of the "Invisible" Auction
To understand why performance varies so wildly, one must look at how the ads are served. OpenAI utilizes a relevance-weighted, second-price auction. Unlike traditional search engines that rely heavily on keyword matching, ChatGPT Ads are triggered by the context and intent of a user’s conversation.
Advertisers provide "context hints"—descriptions of products, needs, or scenarios—that guide the AI. Crucially, these are not keywords. OpenAI makes it clear that it does not guarantee delivery against specific terms or audiences. This decoupling of "intent" from "input" means that two advertisers in the same industry might receive vastly different results based on the AI’s interpretation of a conversation.
When a CPC spikes, the current reporting suite provides no mechanism to distinguish between rising market competition and a decline in the AI’s perceived relevance of the ad to the user. This leaves marketers unable to optimize their creative effectively, as they are essentially guessing which "context hints" the algorithm favors.
Real-World Performance: Lessons from the Frontlines
Early data from companies that have moved beyond small-scale testing provides a glimpse into the volatility of the platform.
The Hostinger Experience
Hostinger, a major player in the web hosting space, stands as one of the most transparent testers to date, having invested nearly $70,000 into the platform. Huseyin Ograk, head of PPC at Hostinger, initially noted that CPCs were competitive with Google Search. However, as the spend scaled, the narrative became more nuanced.
Ograk reported that while CPMs hovered above $65, the real struggle lay in click-through rates (CTR). His team discovered that broader, generic messaging failed to gain traction, while highly specific, use-case-driven ads performed significantly better. The primary takeaway from the Hostinger experiment is that ChatGPT users are highly sensitive to context; they aren’t just looking for a service—they are looking for a solution to the specific query currently active in their chat.
The B2B Mismatch
The experience of Floyd Blaikie offers a cautionary tale regarding traffic quality. Her B2B team spent roughly $7,000 CAD and achieved a $9.29 CPC. While these numbers might look acceptable on paper for certain B2B sectors, a deeper audit revealed a startling reality. Using visitor deanonymization tools to track the 336 clicks the campaign generated, her team identified 146 organizations. Only five of those organizations matched their Ideal Customer Profile (ICP). This suggests that while ChatGPT can generate "traffic," it may currently lack the precision targeting capabilities required by B2B enterprises.
Geography and Variance
Synter’s recent analysis highlights another layer of complexity: geographic disparity. Their test found that CPCs fluctuated wildly by region—from $5.10 in the U.K. to $22.89 in New Zealand. Such extreme variances confirm that ChatGPT Ads are currently an immature market, where supply and demand dynamics are highly localized and potentially volatile.
The "Recommendation" Trap: Decoding the $3-$5 CPC
A recurring point of confusion for new advertisers is OpenAI’s recommended starting maximum CPC of $3 to $5. It is critical to note that this is a bid recommendation, not a market benchmark.
There is a palpable risk in the marketing community that this figure will be misconstrued as the "standard" price for a click. By normalizing this number, marketers may feel pressured to set bids within this range, regardless of their specific industry or the competitive landscape. As OpenAI shifts toward automated bidding strategies—like the "Maximize results" default—these manual bid recommendations may become less relevant, yet the psychological anchoring effect remains.
The Audience Paradox
Perhaps the most overlooked factor in the ChatGPT Ads equation is the audience composition. OpenAI does not serve ads to its entire user base. Ad inventory is currently limited to Free and Go users. The platform’s most lucrative, high-intent segments—including Pro, Business, Enterprise, and Edu accounts—are entirely ad-free. Furthermore, users under 18 are excluded.
For luxury brands or high-ticket service providers, this creates a potential "audience paradox." If your target customer is an enterprise decision-maker or a high-net-worth individual, they are statistically more likely to be on a paid, ad-free subscription tier. This doesn’t mean the platform won’t work for these brands, but it does mean that marketers should approach the platform with a clear understanding that they are not reaching the "entire" ChatGPT audience.
Strategic Recommendations for Advertisers
Despite the lack of benchmarks and the black-box nature of the auction, the opportunity to reach users in a highly engaged conversational environment is too significant to ignore. To approach ChatGPT Ads effectively, marketers should adopt the following strategies:
- Define Success Externally: Do not rely on the Ads Manager for your primary success metrics. Implement robust cross-channel attribution models. Use tools like Triple Whale or visitor identification software to verify if the traffic is actually converting into revenue or high-value leads.
- Prioritize Specificity: The AI rewards context. Instead of broad campaigns, create highly granular ad groups that map to specific, narrow use cases. The more specific your "context hints" are, the more likely the AI is to match your ad with the right conversation.
- Test for "Quality," Not Just "Volume": With current reporting limitations, a high CTR might mask low-quality traffic. Focus your test on downstream actions—sign-ups, purchases, or qualified leads—rather than surface-level metrics like CPC.
- Adopt a "Test-and-Learn" Budget: Given the unpredictability of the auction, treat your initial spend as a R&D investment rather than a performance budget. Use the first $5,000 to $10,000 to identify which geographic markets and context signals produce the most qualified traffic.
- Monitor Platform Updates Closely: OpenAI is actively developing the Ads Manager. As they introduce more reporting views and competitive insights, your strategy will need to evolve.
The first six months of ChatGPT Ads have proven that while the platform is potent, it is not a "set-it-and-forget-it" channel. Success in this new frontier requires a sophisticated approach to data, a healthy skepticism of platform-provided metrics, and an unwavering focus on one’s own business outcomes. As the platform matures, "good" performance will continue to be a moving target—defined not by what others are paying, but by the tangible value the platform drives for your unique business goals.
