Engage the Conversation(al Ads)

Written by Brook Schaaf

Sometimes you catch a glimpse of the future on X. Such happened when one Roy Carrilho bewailed the looming likelihood of what he called a “conversational ad.” (The term apparently already exists to refer to a chat baked into an ad creative, though I had never heard of it before. If I’ve ever seen one, I’ve never interacted with it.) Carrilho meant something different: “a conversation with somebody you trust, and out of nowhere, they suggest a product for you to buy. It’s subtle, connected to the conversation, and delivered so well that you will genuinely do it.” [sic] This came in response to a comment on a Sam Altman post about Pulse, a sort of next-gen homepage or info feed. The comment predicted that ads are soon to follow. 

Ads are indeed sure to follow, but they might arrive later rather than sooner. Reflecting on their absence so far, it occurred to me that ChatGPT may quietly be postponing them for as long as possible. Sure, Altman has teased about the possibility of affiliate links, and OpenAI has reportedly been in on-again, off-again talks with affiliate tracking platforms, but it hasn’t made a significant move in ads or revenue sharing. 

I don’t count the much-hailed Instant Checkout announcement. To me, this is a fancified wallet—and wallets have a poor track record of success, unless you count Amazon as one of them. Instant Checkout strikes me as a small play hoping to be a big one; OpenAI would like to change user habits, but is unlikely to do so, at least with this lever. It’s less a grab at cash than a faint echo of Google’s tour de force that shifted users off desktop apps and into web apps—Gmail, Maps, Drive, Docs, Photos, YouTube, and of course Chrome. “Part of the motivation was that they were looking to make something that would make Google stickier,” as stated on the Acquired podcast.

Instant Checkout seems unlikely to get much yardage if for no other reason than Amazon, jealous mistress of its data, will not participate. Not to mention, merchants are notoriously bad at managing even their own product feeds (more on this in a future blurb), and it’s unclear if there will be much traffic to begin with. That said, is it still a good user experience and viable monetization strategy? 

Google famously disdained and then delayed advertising for fear of corrupting the content, as warned of in Brin and Page’s “The Anatomy of a Large-Scale Hypertextual Web Search Engine.” This prophecy came to pass, though ironically not as the ads corrupting the search results, but as the search results corrupting the ads.

Given that what were once clearly sponsored links now look like native ads (i.e., endemic content) on most walled gardens, conversational ads could hardly be any worse. The key would be to dial in the creepiness factor. In a private chat with a GPT, the ads would probably be fine, and of course, the affiliate marketing channel has monetized links and incentives aplenty. In a semi-private channel, they might also be fine, given that a participant knows they are, say, in someone else’s Slack Workspace. I’d personally be put off seeing something unprompted in a one-on-one message, but it’s easy to imagine a button that turns a snippet into a contextualized link. In fact, this could be an excellent user experience. 

One doesn’t suspect that OpenAI has the same concern as Brin and Page did, in part because the answer engines rely on a somewhat different basis. But it probably is concerned with anything that might keep wide-open land from being claimed. Contextually-relevant, privacy-respecting, non-creepy affiliate ads might be just what OpenAI needs, even if it’s not quite worth a trillion dollars.