# OpenAI ads boss David Dugan on third-party measurement: ‘it’s a natural step’ > Source: > Published: 2026-06-30 04:01:00+00:00 # OpenAI ads boss David Dugan on third-party measurement: ‘it’s a natural step’ Self-graded measurement is the norm for ad platforms: they sell the ads, then report back on how well those ads did, with no independent auditor checking the math. CMOs usually have to fight for years to change that. OpenAI didn’t wait for one. Speaking to Digiday, the AI firm’s boss Dave Dugan said the arrival of third-party measurement is “a natural evolution” from where the platform is today. Right now, advertisers get bidding and outcome data. None of it proves an ad reached a real person. What’s more, it’s OpenAi grading its own homework. Third-party measurement would change that, confirming ads were seen by real humans, in view and in brand-safe environments. Or rather it would up to a point since verification vendors have their own track record of disputed methodologies and blind spots. While Dugan was coy on when marketers would get the chance to make that call – but he did hint at with whom. “As we evolve, of course, we’ll think about what are the most trusted third-party partners that we would look to collaborate with and integrate with,” he said. “So I don’t have any names to announce on that, but I think it’s a natural step that working with trusted industry partners is often an expectation of advertisers or agencies or partners. We respect that.” That candor points to sooner rather than later, which wouldn’t be a surprise. OpenAI’s ads business has come together faster than most. In just four months, it has launched in [seven markets](https://digiday.com/marketing/openai-expands-ads-manager-to-u-k-adds-cpc/) (so far), hired ad execs, struck commercial partnerships, built [measurement tools](http://google.com/search?q=digiday+openai+pixel&oq=digiday+openai+pixel&gs_lcrp=EgZjaHJvbWUyBggAEEUYOTIHCAEQIRigATIHCAIQIRigATIHCAMQIRiPAjIHCAQQIRiPAtIBCDM1OTZqMGo0qAIAsAIA&sourceid=chrome&ie=UTF-8) and [launched an ads manager](https://digiday.com/marketing/openai-has-quietly-launched-its-ads-manager-as-it-races-to-build-out-its-ads-business/). It took the likes of Google, Meta and the rest months — if not years — to do the same. “So as we think about the roadmap, we’ll certainly think about what are expectations from advertisers, and then how can we build that,” said Dugan. “I do think this is such a new experience [for users] that we will have to think about how we evolve some of the metrics. How does the industry have to evolve and perhaps think differently from what other channels like search looked like.” It’s an important point because it gets to one of the more fundamental challenges for OpenAI: this isn’t search so the old playbook for measuring performance doesn’t map cleanly onto it. Dugan expanded on the point: “How does the industry have to evolve and perhaps think differently from what other channels like search looked like. I think that’s going to be a really interesting collaboration between us and some of these partners, because the playbook will have to be different.” A lot is riding on that playbook. If and when OpenAI goes public — it has already [confidentially filed](https://openai.com/index/openai-submits-confidential-s-1/) an S-1 form to the U.S. SEC — it will need to prove to investors it can keep advertisers coming back, and turn a profit quarter after quarter, not just that it got an ad pilot off the ground. But that only happens if they trust the ads work as intended. Third-party measurement is what makes that trust legitimate. “OpenAI has refused to share chat contents with advertisers (it cannot risk eroding user trust), making external verification a requirement to prove effectiveness and build the necessary advertiser trust to transition into being a core part of media budgets,” said Claire Holubowskyj, senior research analyst at Enders Analysis. Still, it would be wrong to think third-party measurement is a silver bullet to OpenAI’s challenges to building an ads business. Yes, it buys credibility and a check against the platform marking its own homework. But it doesn’t answer the harder question: what’s the right way to measure influence inside a conversation in the first place. That’s still being invented. “It proves performance but can’t deliver it: OpenAI is still developing the model for chatbot-native advertising and is yet to land on a truly differentiated format,” Holubowskyj said. ### More in Marketing #### [ How the Chicago Bulls retooled their sponsorship business to meet CMO data demands ](https://digiday.com/marketing/how-the-chicago-bulls-retooled-their-sponsorship-business-to-meet-cmo-data-demands/) As sports sponsorship spending rises, CMOs are looking for hard evidence to justify their deals. Teams and franchise owners are responding, but there’s competition. #### [ Cannes Briefing: The Cannes confessional ](https://digiday.com/marketing/the-expense-of-cannes-is-getting-out-of-control/) Cannes is soft win. Attendees showed up and so did everyone else and the showing up is the whole transaction. #### [ Future of Marketing Briefing: CMOs are still haunted by hard questions about value of ad creative ](https://digiday.com/marketing/future-of-marketing-briefing-cmos-are-still-haunted-by-hard-questions-about-value-of-ad-creative/) While interest in AI-enabled media and creative effectiveness measurement is rising, 49% of senior marketers say they can’t back up their ad creative with hard data.