ChatGPT Ads Now Sold by 'Prompt Relevance'

StackAdapt, OpenAI's official ad partner, has begun selling ChatGPT ad placements based on 'prompt relevance'—meaning the ads shown to users depend on the specific question they ask. A leaked deck, first reported by Adweek on April 20, 2026, reveals the mechanics behind this shift, marking the first time an AI chatbot has monetized user intent at the prompt level.

StackAdapt, OpenAI's ad partner, is selling ChatGPT ad placements based on 'prompt relevance'—ads that match the user's question context.
A leaked deck from Adweek reveals the targeting mechanics, including keyword triggers and brand safety filters.
This move turns ChatGPT into an ad platform, challenging Google's search ad dominance while risking user trust.

What Does 'Prompt Relevance' Actually Mean for Ad Targeting?

According to Adweek, the leaked StackAdapt deck describes 'prompt relevance' as a system where ads are served based on the semantic context of a user's query in ChatGPT. For example, if a user asks 'What are the best hiking boots?', an ad for REI or Merrell might appear in the response. This is not random—StackAdapt uses keyword triggers and brand safety filters to match ads to prompts. The deck includes examples of how advertisers can bid on specific prompt categories, such as 'travel planning' or 'home improvement,' to reach users at the moment of intent.

This is a fundamental departure from traditional search ads. In Google, ads appear alongside results; in ChatGPT, ads are woven into the conversational flow. StackAdapt's system reportedly uses a real-time auction where advertisers bid on prompt relevance scores, not just keywords. The deck claims a 20% higher click-through rate (CTR) compared to standard display ads, though this figure is from StackAdapt's internal tests and not independently verified.

Why Did OpenAI Choose StackAdapt Over Google or Microsoft?

ChatGPT Ads Now Sold by Prompt Relevance

OpenAI's partnership with StackAdapt, announced in late 2025, was initially seen as a minor ad experiment. The leaked deck changes that perception. StackAdapt is a demand-side platform (DSP) specializing in programmatic advertising, not a search giant. According to StackAdapt's own website, it processes over 1 trillion ad requests monthly across display, video, and connected TV. But why not partner with Google or Microsoft, both of whom have massive ad infrastructure?

The answer likely lies in control. Google's ad stack is deeply integrated with its search monopoly—handing OpenAI access would be a strategic blunder. Microsoft, despite its investment in OpenAI, has its own ad business through Bing and LinkedIn. StackAdapt offers a neutral, third-party platform that gives OpenAI flexibility. The leaked deck reveals that OpenAI retains full control over ad frequency, placement, and brand safety rules, with StackAdapt serving as the technical pipe. This allows OpenAI to test ad monetization without ceding data or decision-making to a rival.

Who Actually Benefits From This Deal?

Stakeholder	Gain	Risk
OpenAI	New revenue stream from free-tier users; data on user intent without selling raw conversations	User backlash if ads degrade experience; regulatory scrutiny on prompt-based targeting
StackAdapt	Exclusive access to ChatGPT's user base; proof-of-concept for AI-native ad platform	Dependence on OpenAI's API and policy changes; competition from Google's AI ads
Advertisers	Access to high-intent conversational contexts; potential for higher CTRs	Brand safety concerns if ads appear in sensitive prompts; lack of standard measurement
Google	None—this directly threatens search ad monopoly	Loss of ad revenue from users migrating to ChatGPT; pressure to accelerate AI ad products
Users	Free access to ChatGPT with relevant ads instead of generic banners	Privacy concerns over prompt analysis; risk of manipulative ad placements
Verdict	Openai and StackAdapt win short-term; Google loses. Users face a trade-off between free access and ad intrusion.

Is This a Threat to Google's Search Ad Dominance?

Yes, but not immediately. Google's search ad business generated over $200 billion in 2025, dwarfing any potential ChatGPT ad revenue. However, the leaked deck shows that StackAdapt is targeting the same high-intent queries that drive Google's profits: 'best hotels in Paris,' 'cheap flights to Tokyo,' 'compare insurance plans.' According to Adweek, the deck includes case studies where advertisers saw 30% lower cost-per-acquisition (CPA) on ChatGPT compared to Google Ads, though these are StackAdapt's own figures.

The real threat is structural. Google's ad model relies on users clicking links; ChatGPT's model relies on users staying in a conversation. If OpenAI can prove that prompt-relevant ads drive conversions without disrupting the chat experience, it could siphon budget from Google's search ads. The timeline matters: Google is reportedly testing its own AI ad system, Bard Ads, but has not publicly launched it. OpenAI's head start of 12–18 months could be decisive.

What Remains Unknown About This System?

Several critical details are missing. First, the leaked deck does not specify how user data is handled. According to Adweek, StackAdapt claims that 'no raw prompt data is shared with advertisers,' but the system still analyzes prompts to determine relevance. This raises privacy questions: Does OpenAI store prompt-ad matching data? Can users opt out? Second, the pricing model is unclear. The deck mentions 'cost-per-prompt' (CPP) as a new metric, but no baseline rates are provided. Third, there is no independent audit of the brand safety filters. If a user asks about 'suicide prevention' and sees an ad for a pharmaceutical company, the backlash could be severe.

OpenAI has not commented on the leaked deck. StackAdapt declined to provide additional details beyond the Adweek report. This lack of transparency is a red flag for advertisers who need verifiable metrics before shifting budgets.

My thesis is simple: OpenAI is building an ad platform inside ChatGPT, and StackAdapt is the Trojan horse. In the short term, this gives OpenAI a path to profitability beyond subscriptions—crucial as training costs mount. The free tier becomes a viable product, not a loss leader. In the long term, this model could reshape how users interact with AI: every prompt becomes a commercial signal. The winners are OpenAI and StackAdapt, who own the infrastructure. The losers are Google, which faces a new ad competitor, and users, who lose the purity of an ad-free AI experience.

My concrete prediction: By Q1 2027, OpenAI will launch a self-serve ad platform for ChatGPT, bypassing StackAdapt as a middleman. The leaked deck is a beta test; the real product is OpenAI's own ad server. StackAdapt will be acquired by a larger ad tech firm within 18 months once its value is proven.

Predictions:

OpenAI will launch a self-serve ChatGPT ad platform by Q1 2027, cutting StackAdapt out of the loop.
Google will accelerate Bard Ads launch to late 2026, but will struggle to match ChatGPT's conversational context.
The EU AI Office will require opt-in consent for prompt-based ad targeting by mid-2027, citing GDPR concerns.

Late 2025
OpenAI-StakAdapt partnership announced
OpenAI announces StackAdapt as its official ad partner for ChatGPT, initially seen as a minor experiment.
April 2026
Leaked deck reveals prompt-relevance targeting
Adweek reports on a leaked StackAdapt deck detailing how ChatGPT ads are sold based on prompt context.
Q1 2027 (predicted)
OpenAI launches self-serve ad platform
Predicted: OpenAI will launch its own ad server for ChatGPT, reducing dependence on StackAdapt.

Article Summary:

StackAdapt's leaked deck confirms ChatGPT ads are sold by prompt context, not keywords—a first for AI chatbots.
OpenAI gains a revenue path from free users; Google faces a direct threat to search ad dominance.
Privacy and brand safety remain unresolved; the lack of independent verification is a risk for advertisers.
The partnership is likely a stepping stone to OpenAI's own ad platform, making StackAdapt a temporary ally.