It’s rare that the marketing industry has a chance to ask such a blank-canvas question, but the rapid acceptance of AI-driven chat experiences like ChatGPT presents just such an opportunity. And with both Microsoft and Google planning to launch at-scale versions of the technology in coming months, it’s a question every marketer should be pondering.

Innovations in the media business regularly pair with the emergence of new technologies — radio in the 1920s, network television in the 1950s, the world wide web in the 1990s, social media in the early part of this century. But while new media formats drive new ad formats, the most powerful and disruptive shifts in the media business are driven by rare, complicated and interconnected technological forces that build over decades, then break out in what feels like an instant.

“Generative artificial intelligence” — ChatGPT and its kin — is having such a moment. Over the past 30 years, I can think of just one similar disruption: Search. That gave us Google, the largest media company in history. More than a trillion dollars in market cap later, search remains the world’s most powerful digital marketing channel.

Search not only augured a fundamentally new interface to knowledge, it also upended decades of advertising norms. Instead of paying by impression, you paid for an action taken by a prospective customer — a click. Search allowed for unprecedented targeting, unprecedented experimentation, and unprecedented results. Within a decade, every marketer — from the smallest retailer to the largest consumer packaged goods company – became a search marketer.

So what role might marketers play in the generative AI revolution?

More than likely, they’ll help pay for it. When Google launched its eponymous search engine back in 1998, no one could have predicted that a highly disruptive, multi-hundred-billion dollar market would emerge from an ad unit built on the back of the humble blue link. Two years and ten billion queries later, Google struck gold with AdWords.

OpenAI’s ChatGPT is following an eerily similar trajectory. Launched just five months ago, it’s quickly amassed more than 100 million users and a torrent of punditry claiming that the “conversational interface” is the future of, well, everything.

This puts Google in a classic “innovator’s dilemma.” The company must protect its market share, but it’s facing a disruptive technology that it helped create (Google invented the “T” in “ChatGPT”). Google must thread the needle of its legacy cash cow — AdWords — with whatever product emerges from AI-driven search. So far Google’s announced Bard, an upcoming (and for now, experimental) chatbot similar to ChatGPT. As I’ve written before, it will be difficult for Google to disrupt itself, but if any company is well positioned to succeed, it’s the company that helped invent AI and brought modern search to billions of people.

Over the course of the past few weeks I’ve spoken to search and marketing executives, deep thinkers in tech and media, and leaders at Google and Microsoft. Our conversations all started with “What will the ads look like?” And while their answers aren’t definitive, some patterns certainly emerged.

Let’s start with the big players — Google and Microsoft. I reached out to both, and neither is officially commenting on how they might monetize their new offerings. “Google search has long incorporated LLMs (large language models, a staple of artificial intelligence) to better hone queries and answers…with search ads appearing alongside them,” a Google spokesperson said to me, insinuating that when Google does launch monetization for Bard or similar AI tools, the engine behind AdWords will be part of the mix. That’s to be expected, but it doesn’t tell us how the actual ad experience might change for a consumer engaged with Google’s chatbots. Alas, Google remains tight lipped: “We’re way too early to be speculating about the consumer use case,” the spokesperson said. No doubt scores of product managers at Google are hard at work on the problem, but none are talking to the press quite yet.

Microsoft has also remained quiet, at least to reporters. But an article earlier this year claimed Microsoft is telling major ad agencies that its Bing Chat will incorporate traditional paid links in the results. This would be step one — forcing the old model into the new. But it doesn’t take advantage of the new consumer behaviors inherent in chat-based search. As I wrote last month:

The consumer experience of ChatGPT, Google Bard, or Bing Chat is fundamentally distinct from how search works today. Traditional search’s prime directive is to convert the user as quickly as possible, and to deliver revenue as quickly as possible. There’s a reason a search service never pops up a refining dialog box to ask “Wait, did you mean….?” Roughly 50 percent of search users will abandon a query if they don’t get the right result after just one prompt. 

The interface model for ChatGPT-like services is quite literally the opposite of this approach. Consumers approach the service in conversational mode, honing queries and exploring concepts through informal dialog. The goal isn’t to immediately take action — it’s to find something out. The big shift is that we’re literally talking to the computer, and the damn thing is actually making sense when it talks back to us. That changes everything — not only what we say (the queries), but also what we expect when we approach the service in the first place. 

Beyond boring old blue links, here are four additional ways chat-driven search might incorporate marketing:

  • The affiliate model.  If you work with Google’s Maps or Flights features, you already understand this approach. The platform makes money when a searcher converts from exploring a high value session (“compare the best luxury electric cars,” “give me a five day itinerary for a trip to Sydney,” “is it a good time to refinance my home?”). The chatbot will deliver a summary, and when the user decides it’s time to move down the funnel toward purchase, Google or Microsoft gets paid for the lead. This model is already at scale all around the internet, but could prove highly lucrative given how rich the context can be in a chat-based search session. The platforms will have to be careful about when they insert these links and whether they interfere with a high quality experience, says Jim Lecinski, a professor of marketing at Northwestern’s Kellogg School of Business (and a former Googler). “The reason I like the AI responses is there are no ads in the middle of that text,” he told me. “But in some instances there may be a case where the ad itself is the right content.”  
  • The subscription. OpenAI already offers a subscription service to consumers for $20 a month, with premium features and faster response times. It’s reasonable to assume these services will be advertising free. But as with any subscription service, the vast majority of us won’t pay. That’s an opportunity: Sponsors could underwrite the premium version on behalf of consumers, in exchange for the right to collect an email, show a short video, or connect in some other way. This model could be enriched by the platform’s rich data on each user. Imagine, for example, McDonald’s underwriting free chat sessions for consumers who live in the delivery radius of participating franchises. Audi or BMW might be interested in underwriting sessions for segments of consumers identified as luxury auto intenders.  
  • The underwritten experience. This sponsored subscription idea brings us to what I like to call the “NPR” model. If you’re listened to public radio, you’ve heard the presenter thank brands which have underwritten various types of reporting – GM might underwrite coverage of the environment, or P&G coverage of healthcare. Imagine something similar, but at scale and in depth: Brands might compete in an auction to be associated with specific contexts of chat sessions, similar to how brands once competed to be in highly contextual sections of popular magazines. This would of course require brand-friendly ad units – the equivalent of full page ads — but if consumers are getting value from the service, I’d wager they’d be willing to tolerate brands being part of the experience, just as we did with magazines back in the day. The trick will be to find “native” executions of brand messaging — an ad unit that adds value and is appropriately contextual to the conversation that’s happening on the service. It’d be a creative and media buying challenge, but isn’t that what marketers are supposed to be good at? 
  • The branded agent. Chat-based search will certainly evolve beyond the rather flat call-and-response approach we’ve all been trying over the past few months, and when it does, intelligent agents will likely start to emerge. These agents are software programs that consumers empower to perform specific tasks such as scanning marketplaces for specific deals, monitoring health sites for information that might matter to a particular consumer, or any of a million other ongoing questions that will emerge as that magic wheel of technology, possibility, and entrepreneurial insight starts to spin. Smart brands will build branded AI agents that offer useful services to their consumers — Nestle might build an agent that leverages ChatGPT to deliver insights and tips around nutrition, for example, or Pampers could do the same for parenting. These kinds of programs already exist in the realm of apps and content marketing, but the added dimension of conversational interface could be a game changer, in particular for customer service and sales-based channels.

Of course, beyond ad units, platforms like OpenAI, Microsoft, Google and others will certainly reap untold petabytes of valuable new data about how we use AI, and will be tempted to use that data to create any number of new marketing products that benefit the platforms, but not the consumer or the advertisers that are paying for it all.  We’ve been down that road before, and Kellogg’s Lecinski echoed the advice of many I spoke with about this emerging medium. With AI, he said, “we will certainly find new customer behaviors that bring new ways to monetize.” He then paused, reflecting on the state of digital advertising to date, and added: “Maybe this time we’ll get it right.”