OpenAI officially launched advertising on ChatGPT for free-tier and Go plan users in the United States on 9 February 2026. One month into the rollout, enough early data has emerged to give marketers a clear picture of how this new advertising channel actually performs — and whether the premium pricing is justified.

The short answer: the economics are radically different from anything in the existing digital advertising landscape.

The pricing structure

ChatGPT ads operate at a premium CPM (cost per thousand impressions) of approximately $60, according to multiple ad-tech partners involved in the pilot. This is roughly three times higher than typical CPMs on Meta platforms, which average under $20.

OpenAI is currently testing a cost-per-engagement (CPE) model, where advertisers pay when users interact with ad units embedded within conversational responses. Traditional CPM and CPC (cost-per-click) models are expected to follow, but the initial emphasis is on contextual engagement over demographic targeting.

The pricing reflects a fundamental difference in the advertising environment. Unlike social media feeds, where users scroll passively, ChatGPT interactions are high-intent and conversational. Users are actively asking questions and seeking recommendations, which creates a significantly more receptive context for advertising.

Early conversion data

The most significant data point from the first month comes from Criteo, the first major ad-tech platform to integrate with OpenAI’s advertising pilot. Criteo’s network, which includes approximately 17,000 advertisers, reports that users referred from conversational AI platforms like ChatGPT convert at approximately 1.5 times the rate of other referral channels.

This higher conversion rate partially justifies the premium CPM. If a brand is paying three times more per impression but generating 1.5 times more conversions, the effective cost-per-acquisition (CPA) may still be competitive with traditional channels, depending on the product category and margin structure.

Early brand participants include Williams-Sonoma, The Knot, and retailers across Shopify’s Shop Campaigns network. Walmart has also entered into a partnership with OpenAI for AI-powered shopping experiences within the chat interface.

The data transparency problem

Despite promising conversion rates, a significant limitation remains: the data provided to advertisers is currently restricted to high-level metrics such as impressions and clicks. Detailed conversion tracking, purchase attribution, and audience segmentation data — features that are standard on Google and Meta — are not yet available.

This opacity presents a challenge for performance-focused marketers who need granular data to optimise campaigns and justify spend. For brand awareness campaigns, the high-intent environment may be sufficient justification. For direct-response advertisers, the lack of attribution data makes ROI measurement difficult.

The contrast with Google’s AI Mode advertising approach is instructive. Google’s system extends existing Search, Shopping, and Performance Max campaigns into AI-generated responses, allowing advertisers to leverage their existing measurement infrastructure. OpenAI is building its measurement stack from scratch.

What is coming next

OpenAI has outlined an ambitious feature roadmap for the remainder of 2026:

• Dynamic Creative Optimisation (DCO) by late Q2 2026 — ad content will adjust in real time based on the conversation context.
• Direct-to-chat purchase integration by Q3 2026 — users will be able to complete transactions within the chat interface without leaving ChatGPT.
• Enhanced audience syncing using first-party data and bidding strategies optimised for conversation depth later in the year.

The phased international expansion will extend to the UK, Canada, and Australia, though specific timelines have not been confirmed.

Should brands invest now?

For marketers evaluating whether to allocate budget to ChatGPT ads in 2026, the calculus depends on objectives:

• Brand awareness and consideration campaigns are well-suited to the current format. The high-intent, conversational context delivers strong engagement metrics.
• Direct-response campaigns should proceed with caution until conversion tracking and attribution improve. The $60 CPM demands demonstrable ROI that current data transparency does not fully support.
• First-mover advantage is real. Early advertisers are establishing category presence in a channel that will inevitably scale. Brands that test and optimise now will have a significant data advantage when the platform matures.

The broader shift towards AI-native advertising surfaces means that ChatGPT ads are not an isolated experiment. They represent one node in a rapidly expanding ecosystem of conversational commerce channels that marketers must learn to navigate.

Which verticals are seeing early traction

Early performance data — while limited in granularity — points to specific categories where the conversational context of ChatGPT creates a natural advantage for advertising.

High consideration purchases perform best. Product categories that benefit from substantial research and comparison — home furnishings (Williams-Sonoma), travel, financial products, and high-ticket electronics — align well with how ChatGPT users are actually engaging. When a user is mid-conversation asking about home office setup ideas, a sponsored product placement for an ergonomic chair delivered in that context is highly relevant in a way that a display ad on an unrelated website is not.

Event and occasion purchases see strong intent signals. The Knot’s inclusion in the early pilot is not coincidental. Wedding planning involves hundreds of complex, interrelated purchase decisions made over months. A user using ChatGPT to research venues, photographers, or caterers is exhibiting exceptionally high purchase intent that is much harder to capture through keyword-based search alone.

Categories requiring explanation and comparison are natural fits. Insurance, financial services, and software-as-a-service products — where the decision involves real complexity — are categories where AI-mediated recommendations carry significant weight. A sponsored listing that appears when a user is asking ChatGPT to explain the difference between term and whole life insurance reaches a consumer in a research mindset that is closer to purchase than a demographic-targeted social impression.

Pure impulse categories are a worse fit. Fast fashion, low-cost consumer goods, and categories driven primarily by visual discovery are less natural for the ChatGPT context, which skews toward information-seeking. These categories belong, for now, on social platforms where visual browsing behaviour is built into the experience.

A framework for media teams evaluating ChatGPT ads

For media planning teams deciding whether to pilot ChatGPT advertising in 2026, a practical evaluation framework:

1. Map your customer’s purchase journey. Where does your buyer do research versus where do they discover? If research is a significant pre-purchase phase, ChatGPT ads belong in the test budget.
2. Set brand awareness KPIs, not direct-response targets. Until attribution improves significantly, measuring ChatGPT ads against cost-per-acquisition benchmarks from Google or Meta will produce misleading comparisons. Measure reach, first-touch attribution, and qualitative brand recall instead.
3. Prepare your creative for the format. Conversational ad environments reward concise, benefit-forward copy. This is not a display or video format — it rewards precision over production value.
4. Start with a $10,000–$25,000 test budget. This is large enough to gather meaningful impression data across multiple weeks but small enough to limit exposure before attribution tools mature.
5. Monitor the roadmap. The DCO feature expected in Q2 2026 will significantly change the creative and optimisation possibilities. Brands that have already established accounts and baseline performance data will be positioned to move quickly when that feature launches.