The Gap Between Hype and ROI: Why ChatGPT Advertising Is Struggling to Match Google Search Performance
Paid ChatGPT traffic generates substantially less revenue per visit than paid Google Search traffic, according to new research that casts a shadow over OpenAI’s aggressive push into the advertising sector. As the generative AI giant attempts to transform its massive user base into a profitable ad-supported engine, a working paper from the Frankfurt School of Finance & Management and the Vienna University of Economics and Business highlights a widening disconnect between advertiser adoption and actual return on investment.
The study, which analyzed 1,898 ecommerce sites over a 60-week period spanning from June 2025 to August 2026, provides a granular look at the economic reality of AI-driven commerce. During this window, researchers observed nearly 10 billion visits, creating a massive dataset that challenges the narrative that ChatGPT is poised to disrupt the search advertising duopoly held by Google and Meta.
The Financial Disparity in Conversion
The core finding of the research is a stark contrast in profitability. Paid visits to ecommerce sites originating from ChatGPT generated an average of 89 cents in revenue per session. In comparison, paid traffic originating from Google Search generated $2.45 per session. This discrepancy is further underscored by conversion rates: ChatGPT ads saw a conversion rate of just 0.37%, while Google Search maintained a significantly more robust 1.17%.
Perhaps most concerning for proponents of AI advertising is the finding that among the 102 advertisers tracked in the study, 82% recorded zero purchases directly attributable to their ChatGPT campaigns. When adjusted for the publicly reported costs of these ad placements, researchers concluded that Google Search delivers more than six times the revenue per advertising dollar spent. To reach a break-even point comparable to Google’s performance, the researchers estimate that the cost per click for a ChatGPT ad would need to drop to approximately 26 cents. Current market data suggests a much higher barrier, with some of the largest accounts in the study paying as much as $1.72 per click.
Chronology of the OpenAI Ad Rollout
OpenAI’s entry into the advertising market has been characterized by rapid, global expansion and high-stakes performance claims. Following a successful pilot, the company introduced ads in the United States before expanding to eight additional countries. Notably, the European Union remained an ad-free zone during the study period, providing researchers with a control group to measure the impact of these ads on organic traffic and site performance.
By late 2025, OpenAI claimed that its advertising unit had reached an annualized revenue run rate of $1 billion less than 200 days after launch. While this figure was touted as a major milestone, it remains difficult to verify. Because OpenAI is a privately held company, it is not required to submit public financial filings that would allow independent auditors or the public to confirm the authenticity of these claims. Furthermore, an "annualized run rate" is a projection based on short-term performance rather than realized revenue, leading industry analysts to view the milestone with cautious skepticism.
The Divergence of Organic and Paid AI Traffic
A significant nuance in the research is the distinction between paid ChatGPT traffic and organic AI referrals. Before the introduction of advertising, organic referrals from ChatGPT were highly effective, converting at a rate of 2.21%. This aligns with broader industry data, such as findings from Adobe Analytics, which noted that AI-referred traffic to U.S. retail sites in September generated 83% more revenue per visit than non-AI traffic, with users spending 73% more time on those sites.
However, the introduction of ads appears to have negatively impacted these organic flows. The study found that ChatGPT sent 6% to 8% fewer organic visitors to ecommerce sites after ads were launched. While the researchers noted that the reasons for this decline remain subject to further investigation, they hypothesize that the presence of sponsored content may be altering the user experience.
Currently, OpenAI’s ad design keeps sponsored placements separate from the conversational flow, often displaying a labeled "Sponsored" card below the AI’s recommendation. While this preserves the integrity of the chat, it creates a "walled garden" effect that may be stifling the organic discovery of brands that were previously woven naturally into the AI’s responses.
The Scaling Challenge and Revenue Projections
OpenAI’s long-term financial roadmap is ambitious, with internal projections assuming advertising revenue will reach $102 billion by 2030. This figure implies that the company expects to extract approximately $40 in annual revenue from every free user of the platform. Current data, however, suggests a significant shortfall: the researchers estimate that current ads generate only about $1 per free user annually.
Given that the cost to serve a free user is estimated at roughly $5, there is a fundamental math problem in the current business model. OpenAI cannot simply scale its way out of this deficit through increased clicks; it must fundamentally increase the value of each click. The current advertiser base is growing, with firms like Adthena tracking an increase from 1,082 brands in June to over 6,000 by August, and New Street Research estimating upwards of 60,000 advertisers by October. Yet, as seen in the research, this growth in participation has not translated to a growth in effective, profitable outcomes.
Industry Perspectives on Attribution and Measurement
The disparity between the rapid growth in advertiser volume and the lack of proven return on investment is often attributed to "teething problems." Farhad Divecha, group CEO at the digital agency Accuracast, has noted that discrepancies in measurement are common when new ad systems are integrated into existing advertiser tech stacks. Attribution, or the process of tracking a user’s journey from an ad click to a purchase, is currently "messed up" in the ChatGPT ecosystem, according to industry insiders.
Furthermore, the composition of the advertiser base is shifting. While retail previously dominated, its share of U.S. ChatGPT ad spending dropped from 38% in April to 20% in September. Meanwhile, sectors such as software, financial services, and travel have seen significant increases in spending. This suggests that while retail is struggling to find success, other industries are still experimenting with the platform, potentially hoping to capture a first-mover advantage in a medium that is still in its infancy.
Broader Implications for the AI Ecosystem
The findings from the Frankfurt and Vienna researchers present a critical dilemma for OpenAI. The company faces a "product-revenue trade-off." To make the ad inventory more valuable—thereby increasing the return for advertisers and, by extension, the revenue for OpenAI—the company may feel pressured to place advertisements more directly into the conversational output of the model. However, doing so risks degrading the user experience that drives the high-value organic referrals that currently benefit the ecommerce ecosystem.
If the AI-generated responses become cluttered with direct, intrusive advertising, users may abandon the platform, leading to a decline in the very traffic that advertisers are seeking to capture. Conversely, if OpenAI maintains its current, less-intrusive ad format, it may struggle to prove the value proposition required to justify its massive, multi-billion-dollar valuation.
As the paper remains under peer review, these findings serve as early evidence rather than a settled verdict. However, they underscore a vital lesson for the technology sector: high-volume engagement does not automatically equate to high-value commerce. For the foreseeable future, the "AI gold rush" in advertising remains more speculative than structural, with both OpenAI and its growing roster of advertisers left to bridge the gap between user intent and actual conversion. The coming months will likely be defined by whether OpenAI can refine its ad-targeting technology or whether the platform will continue to serve as a high-visibility, low-conversion channel for brands.
