From 'Awareness Economy' to 'Intelligence Economy' and the prospect of a $100 billion revenue stream - OpenAI's Denise Dresser takes OpenAI's ad pitch to the Mad Men in Cannes
Denise Dresser at Cannes LionsOpenAI founder and CEO Sam Altman once famously declared that having advertising inside ChatGPT would be a “last resort”, insisting:I kind of hate ads just as an aesthetic. . .
Denise Dresser at Cannes LionsOpenAI founder and CEO Sam Altman once famously declared that having advertising inside ChatGPT would be a “last resort”, insisting:I kind of hate ads just as an aesthetic choice. I think ads needed to happen on the internet for a bunch of reasons, to get it going, but it’s a momentary industry. The world is richer now. I like that people pay for ChatGPT and know that the answers they’re getting are not influenced by advertisers. But that was 2024, a lifetime ago in AI hype cycle terms. How times change - and how quickly shameless pivots can be executed when there's money being left on the table. Back in April, OpenAI told investors that it expects to generate around $2. 5 billion in advertising revenue this year, with that projected to rise to $100 billion by 2030. That ambition was articulated only a few weeks after the firm began experimenting with adverts inside ChatGPT. So, any doubts are at an end - OpenAI is in the advertising game now, and as such Chief Revenue Officer Denise Dresser has made the obligatory pilgrimage to the Cannes Lions annual junket for the Mad Men of Madison Avenue and their ilk, where she’s articulating the OpenAI ad story. . . so far. It’s, inevitably, a story framed by the twin notions of disruption and transformation, she says:It's the work, it's the knowledge work. So, you think about a Marketer who's thinking about planning a campaign - preparing, researching, looking at the execution - all of that can be done so much better and faster, and a Marketer can have empowerment and agency. The creative space is going through “such an exciting moment”, she opines, with creators able to go from idea-to-execution, and at speed, scale and with wider audience reach, powered OpenAI’s Codex code generation model:A creator can actually use Codex to think about, 'How do I build, how do I view this creative content, how do I think about the placement, and then how do I iterate?'. It's a tool to help them be even more scalable. . . Codex started with the software development lifecycle, automating that entire process. We started with developers, but now what we're seeing is that this is expanding to all knowledge work. This is the capability of bringing agents into how you market, how you build that marketing campaign, how do you create, how do you think about go-to-market? How do you build a customer presentation? If I'm an ad agency, how do you think about building a client proposal? How do you think about the deliverables and doing that in Codex, automating that capability, and really having a conversation and an agentic experience. When it comes to enhanced audience reach, Dresser pitches:There's a new form factor of advertising that's happening. We've gone from an 'Awareness Economy' to an 'Intelligence Economy'. That means that, as users come in with deeper intent, thinking about the form and the way that you reach users with more usefulness in ads, that's a new opportunity for all of us. It's great for consumers when you think about getting help right there in the moment when you need it. And that Holy Grail of personalization also gets a boost here, she claims:Personalization is also having this evolution, which is not just about how do you reach multiple people in a broader campaign, but how do you create experiences that are actually personal in ChatGPT? One of my favorite examples is Maybelline. So, we work a lot with L'Oréal, and they're doing something where you can actually try on the product and actually see what the mascara looks like. It's on the website now. It's going to move into ChatGPT. Think about that - in the moment, you can actually have that virtual AI-powered experience. That's an entirely new experience, and it's because in ChatGPT the users are going not just to search for a keyword, they're actually going to do something. Twenty percent of the users that go there are actually going there for some commercial intent already. Ads in the botBut it’s the topic of ads embedded into ChatGPT that is the more interesting point of discussion. Now that Altman has changed his mind, what happens now - and how quickly? Dresser concedes that it’s taken time to establish a party line on the role of advertising:We wanted to be very thoughtful about this whole area. The mission of OpenAI is to bring intelligence to all of humanity, for good. One of the ways in which we do that is by scaling our business and providing an ads version of that, to make it more accessible. As we think about experimentation, we want to make sure that first of all, it is useful, is not intrusive, is helpful, and that it allows you to have this experience that is not invasive, but really within the flow of what you're doing. A few months into the experimentation phase, the firm is tapping into feedback from its test markets, she says:We work closely with CMOs and ad agencies to really understand what they're trying to accomplish with their users, what is the right content, what is the right creative that is working, and we give people as much information that we've learned in the very short period of time we've been doing this, to help guide them and advise them. Like with AI in general, we're all learning together and sharing the information and using experimentation and being very thoughtful[We're] moving very slowly, but the response has been amazing. We're almost four months in, we have thousands of customers now, we've expanded in seven countries. We're going to continue to expand now. . . but I think it's still important to make sure the user experience is great. Just from the user experience, we've seen a 50% reduction in when someone closes the ad, meaning that the ad was not useful, so we're seeing that relevancy improve. The form factor has to be based around usefulness, she insists, both to the creative and the consumer:Ads in ChatGPT need to be useful. What I mean by useful is like relevant to what you're talking about and helpful in the discussion that you're having. [Compared to a search ad] it's gonna probably feel less attention-grabbing, more useful, more intelligent around how you're using the product and conversational, so it's in the conversation. Every single industry is going through a transformation right now, and I think when you think about advertising and marketing and creative, there is such an opportunity here to think about Marketing, the creative, and the reach and the new way to reach people in a different form factor, and that's really exciting right now. . . . really we're just getting started, but the response has been really great. Bigger pictureAway from the advertising sector. , Dresser, formerly CEO of Salesforce’s Slack business, is in reflective mood about the wider AI business transformation journey for enterprises:Initially, rightfully so, I think everyone thought about maybe this is productivity or this is hours saved, and that was great, but I don't know that it unlocked all of the economic value that every company wanted to get. I think the stat right now is 70% of the economic value is in 20% of the companies in the world. If I think about a lesson learned, it is that this generation of AI and capability is not just a technology transformation, it is a business transformation. So thinking about it in the early chapters as just technology, did not yield the outcome that companies want. Now CEOs and the boards have entered the room. The business executives are in the room, and they're thinking not about incrementalism, but exponentialism. . . Forget about productivity for productivity's sake, every workflow, every industry can be re-imagined. So, how do we lead organizations through that?So at a time when tokenomics paranoia is kicking in and early enterprises experimenters are expressing dismay at lack of perceived value from their AI investment to date, what does a firm like OpenAI need to do here? Dresser advises:The way that I think of helping companies understand how to transform is thinking about the business process and the outcome and really guiding companies along the way. I think we have a lot of education that we still need to do to help everyone understand the capability of AI. What we've seen is this AI maturation or adoption if you will, and it has got to be bottoms up, and it does need a CEO sponsorship. But the evolution that I see happening, first of all, starts with individual productivity. . . You have to think about literacy first, making sure that everybody knows how to use the tool, has their own experience with it. But where we see companies really starting to take off and unlock this capability is when they move from just my own productivity at work to thinking about, 'How can I share an agent with a teammate?' ,and then we start collaborating around AI, and it gets to be more embedded in the work that we do. The next step is embedding AI into workflows, she goes on:Think about a marketing workflow, a campaign creation, or a creative workflow, or you can think about software development lifecycle. . . When companies begin to get to that level, they start to really unlock the opportunity in their business, and then it enables them to think about industry transformation. It doesn't mean that you have to go slower, but what you have to think about - and when we work with companies, we do this, - where are you today? It's sometimes just good to know where are you at on this maturity curve, so you can plot yourself. , [This is] helping companies understand, 'Well, if we're really just at AI literacy, but we don't have people working together and embedding AI and workflows, what are the two or three high-value use cases that we can lean into and actually start to begin to go up and up into the right?’. It is important to remember is that the technology is moving so fast that it is hard for companies to keep up, she suggests, but adds that the biggest mistake any enterprise could make right now is to do nothing:Being afraid to not get it right the first time, not staying with it. Sometimes I think people are like, ‘It didn't work' and the reality is, this can be really iterative. When you build an agent or you leverage AI, it's kind of like hiring an employee - they're highly qualified, they've made it through the process, but you still have to give them education, knowledge, and learning. When you use the capability, it gets smarter and smarter. It knows you personally. So [the trick is] staying with it, not giving up, and realizing that this is a journey. :My takeWe're at this moment where everyone uses Chat GPT, literally millions and millions of people, so you have a bottom-up experience where employees are already using it, often in their personal lives, but then you're also having a top-down conversation with CEOs. After a decade in various roles at Salesforce, Dresser cut an impressive figure during her two year tenure at Slack where she helped to grow that product into perhaps the brightest star in the Salesforce firmament. She was lured across to OpenAI last year with a mission to expand its enterprise footprint. That, in itself, will be a journey worth tracking.