OpenAI showed up at Cannes Lions for the first time this week. That's not just a branding play — the company is actively building out its ad sales operation, and the pitch it made to CMOs and agency leaders tells you something real about where ChatGPT ads are heading, and how fast.
What OpenAI told CMOs at Cannes
David Dugan, OpenAI's head of global ad solutions, made the core argument: ChatGPT users arrive with "a job to be done." Around 20% of queries carry direct commercial intent — product research, comparisons, purchase decisions — concentrated in categories including travel, retail, beauty, healthcare, and financial services.
Denise Dresser, OpenAI's chief revenue officer, shared the stage at Cannes with Google CMO Lorraine Twohill for a session called "Winning the AI Discovery Era." The positioning was deliberate. OpenAI isn't pitching itself as a niche experiment anymore. It's pitching itself as the intent-rich alternative to a Google search stack that's losing clicks to AI Overviews and a Meta feed that's losing attention to broadcast video.
The ad format itself is unchanged: clearly labeled sponsored text at the bottom of a ChatGPT response. Organic answers stay above the fold. CPC and CPM bidding are both available, with no minimum spend requirement. The platform already has thousands of advertisers across seven test markets. Expansion to Brazil, Mexico, Japan, and India is planned for Q3 2026. The company filed confidentially for an IPO in early June — ad revenue growth before that debut matters financially.
The measurement gap that just closed
The bigger news landed a few days before Cannes: LiveRamp became ChatGPT ads' first Conversions API partner.
This is the piece that's been missing. Browser-based click tracking can tell you someone clicked an ad. It cannot tell you whether they bought anything, especially in-store. Every brand running Meta, Google, TikTok, or Pinterest seriously has a server-side CAPI integration. ChatGPT didn't have one. That's not a minor gap — it's why most performance marketers have stayed in curiosity-test mode rather than allocating real budget.
LiveRamp's CAPI Hub connects ChatGPT ad exposure to verified purchase data via hashed email matching and RampID. In-store conversions come first; online conversion measurement follows. It's the same infrastructure LiveRamp already runs for the other four platforms.
If you're already running LiveRamp's CAPI Hub for Meta campaigns, connecting ChatGPT is an extension of what you've built, not a new project.
The "super intentional" argument
The intent claim deserves scrutiny, not dismissal.
When someone opens ChatGPT and types "best running shoe for flat feet under $150," they've already done the mental pre-work. They're not passively scrolling. That's a structurally different moment than a Google Search query on "running shoes" — which spans the full spectrum from early curiosity to ready-to-buy. ChatGPT queries, by nature, tend to be further along that spectrum.
What's still unverified: actual conversion rates at scale. The "super intentional" label comes from query intent analysis, not from closed-loop purchase data. That's exactly what the LiveRamp integration is designed to generate. Until that data exists across enough campaigns to draw real conclusions, the intent argument is a hypothesis with good supporting logic — not a proven channel advantage.
You're testing to find out, not betting on a certainty.
Who should run a test now
The categories Dugan named — travel, retail, beauty, healthcare, finance — share a common structure: the buying decision involves research and comparison before purchase. If your product fits that pattern, you have a reasonable case for a test.
Practical setup: connect LiveRamp CAPI before you spend a dollar. Without conversion tracking live from day one, you'll have click and CPM data but no purchase signal. That tells you nothing useful about whether the channel works for your category.
Budget logic: OpenAI has no minimum, so this isn't a $50K pilot commitment. A 30-day test weighted toward CPC bidding, with $2,000–$5,000 allocated, gives you enough signal to compare ChatGPT's cost-per-acquisition against your Meta and Google baselines. Run the test at a category level, not a brand level — you want to see which product types generate intent that converts, not just whether your brand name gets clicks.
The IPO context is worth factoring in. OpenAI needs to show ad revenue trajectory before going public, which means the account management and optimization support available to early advertisers right now will be higher than post-IPO. That window doesn't stay open indefinitely.
The timing argument
Cannes is where platforms court budgets. Every major platform — Meta, Google, TikTok, Pinterest, Reddit, Snap — showed up in Cannes this week with roadmaps and case studies. OpenAI was there for the first time, which signals they're past the internal debate about whether advertising is part of the business model.
The brands that tested ChatGPT ads in February, pre-CAPI and pre-geo-targeting, paid $60 CPMs with no attribution to show for it. CPMs have since dropped to $25, CPC bidding is live, geo targeting exists, CPA optimization is in early access, and now conversion measurement works.
The platform is four months old and is adding advertiser infrastructure faster than Google Ads did at the same stage of development.
If measurement was your reason to wait, that reason is gone. Run your audit at /audit first to confirm your current attribution stack is solid — if your Meta and Google CAPI data is noisy, adding a third channel creates confusion, not clarity.
Get the foundation right. Then test the new channel.
Sources: Marketing Dive, Digiday, Storyboard18, AdWeek, Adgully, Cryptobriefing, June 2026
