The Daily: GenAI Search Ads or GenAI Search Subscriptions?

On today’s podcast episode, we discuss when and where we will start to see ads alongside ChatGPT, how other AI startups are rolling them out, and what kinds of GenAI ads will be most impactful. Tune in to the discussion with Senior Director of Podcasts and host Marcus Johnson, and Analysts Jacob Bourne and Daniel Konstantinovic.

Subscribe to the “Behind the Numbers” podcast on Apple Podcasts, Spotify, Pandora, Stitcher, YouTube, Podbean or wherever you listen to podcasts. Follow us on Instagram.

Episode Transcript:

Marcus Johnson (00:00):

What does data collaboration enable? Glad you asked. Innovative companies to unlock powerful new insights that transform customer experiences and fuel business growth, that's what. With LiveRamp, marketers get the industry's only interoperable platform for data collaboration across every cloud, wall, garden, and media platform. Learn more at liveramp.com.

Daniel Konstantinovic (00:21):

It really needs another revenue stream to make that happen. And just for some context, I mean, generative AI search, deploying that costs 10 times more than deploying traditional search. And so that gives you a sense of what the main differentiator here is.

Marcus Johnson (00:44):

Hey gang, it's Thursday, December 12th. Jacob, Daniel, and listeners, welcome to the Behind the Numbers Daily eMarketer podcast made possible by LiveRamp. I'm Marcus. Today, I'm joined by two folks. Let's meet them. We start with our senior editor of the Marketing & Advertising Briefing. He is based in New York. He is called Daniel Konstantinovic.

Daniel Konstantinovic (01:05):

Hello, I am called Daniel Konstantinovic.

Marcus Johnson (01:07):

It's true, the whole time. We're also joined by one of our technology analysts. He is based on the other side of the country in California, and we refer to him as Jacob Bourne.

Jacob Bourne (01:19):

Thanks for having me today, Marcus.

Marcus Johnson (01:20):

Yeah. Of course, mate. Thank you guys both for being here. We start with the fact of the day, how big is a fully grown blue whale? So the largest thing-

Daniel Konstantinovic (01:30):

Pretty big.

Marcus Johnson (01:30):

... on the planet. That's the correct answer. Pretty big is the exact correct answer.

Jacob Bourne (01:34):

Many tons.

Marcus Johnson (01:38):

Easy with the specificity, gents.

Daniel Konstantinovic (01:39):

No problem.

Marcus Johnson (01:40):

They're very, very big indeed. But actually a blue whale, a fully grown blue whale is the same size as a Boeing 737.

Jacob Bourne (01:47):

What?

Marcus Johnson (01:47):

So next time you're looking at a plane-

Daniel Konstantinovic (01:50):

Wow.

Marcus Johnson (01:50):

... You can just think, blue whale. That's also about three school buses end to end to end. However, in terms of weight, a blue whale is 15 fully loaded school buses.

Daniel Konstantinovic (02:06):

Fully loaded. That's so fun.

Marcus Johnson (02:08):

The length of three to put 15-

Jacob Bourne (02:09):

Fully loaded with children or adults?

Marcus Johnson (02:11):

Yeah, fully loaded with, I don't know, adults, I think. They taking school buses? Maybe. Speaking of them being compared to planes, blue whale being the same size as a Boeing 737. A jet engine can reach up to 140 decibels, a blue whale can reach 180.

Jacob Bourne (02:33):

Wow.

Daniel Konstantinovic (02:33):

Step it up [inaudible 00:02:34].

Marcus Johnson (02:34):

It ain't quiet.

Jacob Bourne (02:34):

Well, I mean, you got to communicate under in the deep ocean, right? So it makes sense. You got to-

Marcus Johnson (02:39):

That's true, yeah.

Jacob Bourne (02:39):

... Make some noise down there.

Marcus Johnson (02:40):

Screaming at each other. Anyway, today's real topic, gen AI search ads versus gen AI search subscriptions. All right, gents. So we're talking about gen AI search ads and maybe search subscriptions as well where we start to see those emerge more and more. We'll start with this. Our colleague Jeremy Goldman just wrote that OpenAI is exploring advertising as a potential revenue stream according to CFO, Sarah Fryer, in a Financial Times interview as the company, valued at over 150 billion, continues to chart its path to profitability. However, Ms. Fryer later told the Register that OpenAI had no immediate plans to implement advertising, going on to say that, "The company is placing a lot of thought around when and where we implement ads." Daniel, I'll start with you because she's written a bit about this very, very recently. Where and when will we start to see ads alongside ChatGPT?

Daniel Konstantinovic (03:40):

My guess is probably sometime in 2025. I would be surprised if it doesn't launch by the end of the year because just days after this news where the CFO let slip that they were thinking about advertising and then walked it back, OpenAI made a big deal about hiring its first chief marketing officer, which definitely signals a effort to not just market OpenAI and ChatGPT in a more meaningful way, but also potentially to bring advertising features to its products.

Marcus Johnson (04:12):

Yes. And you were saying that she's got very strong links to the digital ad space. It's not just some random CMO-

Daniel Konstantinovic (04:18):

Yes.

Marcus Johnson (04:19):

Someone who is an indicator that the advertising is definitely on the cards.

Daniel Konstantinovic (04:22):

Yes, her name is Kate Rouch or Roosh, I hope I'm pronouncing one of those two correctly.

Marcus Johnson (04:27):

There are some options.

Daniel Konstantinovic (04:27):

She worked at Metta for over a decade and at one point was the global head of brand and marketing across its suite of apps. And before this, worked at Coinbase and while she was there, Coinbase had its big Super Bowl ad. So definitely, someone who has elevated a tech company to a prominent world stage in terms of brand marketing.

Marcus Johnson (04:52):

Yeah. And Danny, really quick, I mean, it was interesting because you'd mentioned that we've not really seen this big consumer facing push from OpenAI. And I hadn't really thought about it too much, but they've not really been seeing in all the ads telling people who they are, this is how the product works, the same way that a lot of other brands do. And so, I'm assuming that her hire means that we're going to start to get to know, if you will, OpenAI as a company a bit better in terms of how it fits into consumers' daily lives.

Daniel Konstantinovic (05:23):

Yeah, I would say so. I mean I would expect something like that from them, especially because we've seen numbers that show that use of chat GPT is getting up there in comparison to things like Google Search and I think that they definitely want to increase those numbers and a good way to do that is to just increase brand awareness about it. I think a lot of US consumers are probably familiar with OpenAI just because it's in the news everywhere. I'm sure a lot of people are encountering it at work. But yeah, it really is not in marketing much itself, as a brand, and you're seeing a lot of companies like Apple or Samsung marketing things like phones with AI features and really putting AI out there as this big consumer feature. We'll see how much interest that generates or if consumers really bite. But I would be surprised if OpenAI is not interested in being out there in the same way since, even without those big marketing efforts. I think it's pretty safe to say that OpenAI is synonymous with AI.

Jacob Bourne (06:25):

And I think, Danny, you're right about the 2025 launch for ads. I think that is when we will see it. And there's two things driving it. One is, we're seeing this coincide with OpenAI's push to become a exclusively for-profit company. And the other thing is, it's OpenAI, like its competitors, is really struggling to meet the high costs of deploying a commercial generative AI. And of course, it's setting its sights on building an artificial general intelligence, which is also very expensive. And so, it really needs another revenue stream to make that happen. And just for some context, I mean, generative AI search, deploying that costs 10 times more than deploying traditional search. And so that kind of gives you a sense of what the main differentiator here is. The $20 per month subscription model is just not quite cutting it. The $200 per month subscription, I think, is just, people are just not going to pay for that in droves-

Marcus Johnson (07:23):

Yeah.

Jacob Bourne (07:24):

... I think is the take home there. So they need another way to make money, essentially.

Marcus Johnson (07:28):

To your point about how much they're making, they've been able to make money from subscriptions and licensing [inaudible 00:07:34].

Jacob Bourne (07:33):

And their API as well.

Marcus Johnson (07:35):

And their API, absolutely. Just not enough to turn the profit. CNBC's Hayden Field said OpenAI expected to post a $5 billion loss this year on nearly $4 billion in revenue. Miss Field was also noting though, that revenue is expected to jump to 12 billion next year. You'd imagine the ad's going to try and help with that, as a source with knowledge of the matter was confirming. So a big jump from the 4 billion that they expected to make this year, 12 next year. And to what Jacob's talking about, Angrej Singh of Axios saying, "OpenAI is bringing the o1 reasoning model series out of preview in ChatGPT and is launching a new pro tier," the one Jacob's talking about, $200 a month for unlimited access to the oh one model GPT-40 and the advanced voice mode. Jacob, it sounds like you don't think this is going to make much of a dent in terms of uptake.

Jacob Bourne (08:24):

Yeah, I mean, certainly, this is geared towards enterprise users, organizations, not individual consumers. Maybe some high-powered high-end users will pay for that. People will try it, but I don't see that being a real moneymaker at that price point.

Marcus Johnson (08:40):

It does seem prohibitively high, $2,400 a year for this thing. Danny, do you see OpenAI leaning heavily on the subscription side of things in the years to come? Do you see ad revenue when it does, when they do turn the taps on, flooding the share of how much money they're making, the balance between ads versus subscriptions?

Daniel Konstantinovic (08:58):

It's kind of hard to say because the CEO of OpenAI has historically put up a lot of resistance to advertising and is very concerned about how advertising could mess with the user experience on ChatGPT and its products. But the sort of context around turning on advertising is important to take into account because it's not just OpenAI. This is a lever that so many apps and tech companies have pulled because it is just a very easy, I suppose, way to generate revenues. You have a lot of user information that you can leverage for advertising. If you have a large volume of users, you can entice a lot of companies to come on board and advertise on the platform.

Jacob Bourne (09:38):

Yeah, it's very lucrative and I think it comes down to just the competition. ChatGPT is the most popular generative AI platform, but there are so many and so people will want some variety in terms of the kind of generative AI tools they can use and aren't going to want to pay a $20 subscription for five different platforms. And so, ad revenue gives people the sort of flexibility to create a free account and then use different tools from different providers.

Daniel Konstantinovic (10:06):

Yeah. And I think moves from some competitors have shown that there is interest in AI advertising in its many forms, whether that's using AI to create advertisements to target people or appearing next to AI search results, for example. And as much as Sam Altman might have a personal distaste for ads, the market pressure against these companies to show that there is a mass product here and an opportunity for significant revenues, is going to force OpenAI down this road eventually, no matter how the people at the very top feel.

Jacob Bourne (10:43):

You're absolutely right. I mean, Wall Street is just really looking for a clear evidence of monetization of generative AI. They've grown more skeptical of generative AI capital expenditures and OpenAI is certainly, with this relationship with Microsoft, is one of the players that their scrutiny over on that front. So it's almost a certainty at this point that we're going to see OpenAI and others launch ads on their platforms and rely heavily on that revenue.

Marcus Johnson (11:11):

Yeah, I mean, we've seen a couple of examples of that, right? So others starting to implement gen AI ads, Ben Sherry of Inc noting that AI search engine perplexity recently began experimenting with sponsored follow-up questions. Meaning, if you were to search, "Help me find a job," for example, the chat bot could suggest that you next ask it, how can I use Indeed to enhance my job search? And then Zak Stambor, our colleague, writing about how Amazon has quietly begun to weave sponsored ads, it's core search ad format, into Rufus, its Gen AI powered chat bot. It launched earlier this year. So shoppers might ask, "What gear do I need to go camping," in the forum would get, Patagonia jacket recommendations, REI camping stoves. Jacob, what do you make of these examples and others?

Jacob Bourne (11:55):

Yeah, I mean, this is just further evidence of what we're saying. I mean, there's Google with its AI overviews, they have ads. Microsoft added ads to its co-pilot Perplexity with its sponsored followup questions. It's the industry trend now to explore this. I think another interesting player to note is actually Meta, though in a different way. Meta just recently said that its AI assistant reached 600 million active monthly users and it's a free AI assistant to use, but the notable thing about it is, it's integrated into its existing platforms and so the monetization there is on the existing social media platforms [inaudible 00:12:36] ads, right? And so, Meta doesn't necessarily have to monetize its AI assistant directly because there's already that happening with the greater ecosystem where AI assistant is just basically to serve as a product enhancement. And it's kind of demonstrating indirectly that it's finding some success in terms of more time spent on its platforms because of that. So it's kind of sort of an indirect way to monetize generative AI that I think is pretty viable.

(13:06):

I think the big takeaway from all of these different players doing this is that players like Google and Meta are really going to have an advantage out the gate in terms of integrating ads into generative AI just because they already have that advertising framework well established. They have-

Marcus Johnson (13:07):

Yeah, that practice.

Jacob Bourne (13:23):

They've established trust with brands in a way that Perplexity and OpenAI have not done. And so I think, although long-term ads are going to be a big revenue driver for OpenAI and a player like Perplexity, I think it's going to take them some time to get there because they have to really build up that framework and that trust in the industry.

Daniel Konstantinovic (13:47):

I think there are a couple of interesting things about what you just said, Jacob, the AI assistants, personal assistants, this is a feature that's cropping up everywhere. Reddit is also adding this. Yelp just announced that it's going to do a similar thing, and I think you should look at those launches as something that could eventually be a vehicle for advertising. Like, any place within a platform that a consumer is directly interacting or interfacing with AI and knows that that's what they're doing, that is a potential place for advertising.

(14:21):

The other thing that I think is really interesting is Google. So Google has its AI overviews, these AI generated summaries of the search results at the top of the page, and it has launched advertising for the AI overviews in some capacity and I'm sure we'll roll it out more broadly and it's attracted some interest. We don't have numbers about how much has been spent on this or how many advertisers are using it or anything like that, but I think the thing that makes it really interesting is the question of, are marketers interested in this because it is AI and because it has some kind of unique angle to it that they can tap into, or is it because it's the most prominent real estate on Google search?

(15:06):

I'm not sure where I fall on that question. I mean, it is now the top thing that shows up in most Google results, which is very sought after ad space that Google can sell. So is interest in AI overviews ad placements, reflective of advertiser interest in appearing next to AI in general? I think it's tricky to say.

Jacob Bourne (15:29):

Yeah, just to take a stab at that, Danny, I'd say it's the prominence of the AI overviews that's really the draw there because I think that this is sort of an experimental thing. There's a lot of risks too, in terms of AI hallucinations and being able to maintain brand messaging alongside AI responses. So there's some risks there, but I think they're willing to take that risk because of the prominence and just the interest in AI overall. This is my guess.

Marcus Johnson (15:57):

We started by talking about the kind of where and the when AI search ads show up, it really, really matters. Ina Fried from Axios was making the point that a chat bot-turned-salesman could be both lucrative and annoying. An AI companion that sneaks in a sponsored message or an AI agent that makes choices for you based on ad dollars could be predatory. And so given everything we've talked about, Jacob, I'll start with you, what kinds of Gen AI ads will be the most impactful? What are the things folks need to keep in mind when they're rolling these things out or when they're buying ads on these platforms?

Jacob Bourne (16:31):

Two examples. I mean, I think Perplexity's sponsored follow-up questions is a really impactful way to introduce it. The questions are, first of all, they're relevant to the initial prompt, and so that will, I think, ensure user engagement, but it's also non-intrusive because it's not part of the output itself, it's kind of shown separately. So it's not really part of the catboat's core response.

Marcus Johnson (16:55):

Interesting.

Jacob Bourne (16:55):

Another thing about it is that Perplexity has said that it's not going to share data, which could be reassuring for some users who are concerned. And I think the final thing, which is sort of maybe a double-edged sword is that the sponsored follow-ups are actually AI-generated. They're not generated by advertisers. So that could create some concern for advertisers, but I think that the way that Perplexity's doing it is probably the least intrusive. And so it'd be interesting to see if it can actually get advertisers to spend money on its platform.

Marcus Johnson (17:28):

Yeah.

Jacob Bourne (17:30):

So that'll be the key thing. But again, going back to Meta. Meta's having the monetization already baked in with its existing products, I think is, it's really the ideal solution that others really can't replicate in the same way necessarily. But having that revenue stream there while not necessarily having to incorporate it closely to the general AI output is probably ideal from a risk perspective.

Marcus Johnson (17:56):

Right. Danny, I keep thinking about how advertisers, they can't be running faster on AI than consumers can keep up though, right? Some data points here, the share of brands willing to advertise alongside AI generated content went from 33 to 57% in the past year according to advertiser perceptions. However, only 35% of US adults are comfortable with AI generated content in advertising and those civic sites. So advertisers have got to make sure that they're bringing consumers along for the ride because even though they're used to ads being on platforms where they're looking for things, it's a new type of experience.

Daniel Konstantinovic (18:34):

It is a new type of experience and exactly how consumers feel about it or will feel about it or will choose to interact with it is still pretty unclear. And something that I feel is also worth mentioning is that there is a real lack of rules and regulations around AI and advertising in AI. Regulators are very slow to catch up to the pace of this. We've seen numbers that show that consumers tend to feel more negatively about advertisements when they know that AI has been involved in some capacity. But those same consumers are also not great at identifying when an advertisement is AI generated or had AI involved in it. So if you are a platform like Google or Meta or whatever, or a brand that's using AI to advertise in some capacity, that gives you an incentive to not necessarily disclose that.

Jacob Bourne (18:34):

Right.

Daniel Konstantinovic (19:32):

I think that raises some questions, like if you're a consumer, the way that your ad experience and the information that you get online could be affected even without your knowledge, I think, is concerning. And I'm sure it's also concerning to some extent for advertisers too, because you don't want to use AI, not disclose it, and then have it found out that you are in a X amount of money partnership with Google or whatever or that you made this ad with AI and said that it wasn't, or something like that. For now, we've seen brands very publicly disclosing when they're doing this and to the credit of these big platforms, there are disclosures when there's an AI sponsored response, at least in Google AI overviews. So we're not seeing a backing away from that at the moment, but the temptation exists.

Marcus Johnson (20:22):

Yeah, transparency is going to be key. I want to end here with something that Zak Stambor, one of our colleagues I mentioned earlier, said, basically saying there's going to be somewhat of a learning curve. He said, "Gen AI powered chat bots are still in the early days, and consumers aren't used to seeing advertising integrated into the responses to their queries, but they are used to seeing ads in their search results, both on search engines and retailer's websites. That's why they're likely to see advertising within a chat bot as a natural extension of search ads. However, execution matters." He says ads that are well labeled, well integrated and relevant are more likely to drive strong results and avoid blow back but consumers may grow frustrated if they can't distinguish down to what you're saying between an ad and a natural result, or if the ads aren't relevant.

Jacob Bourne (21:11):

Well, I just want to say, we've talked all about the ad format, but I think the subscription format, while might reduce in relevance or prominence, I should say, it's still going to be around. They're going to be generating revenue certainly from it, but I think that the focus there is really going to be on niche use cases or topic areas like academia, enterprise users, I think, are going to gravitate towards the subscription-based platforms models.

Marcus Johnson (21:42):

Yeah. A great final thought. Thank you so much, gents, for being on today's show. That's all we've got time for. Thank you first to Jacob.

Jacob Bourne (21:49):

Thank you, Marcus. Thanks for having me.

Marcus Johnson (21:51):

Yes, sir. Thank you to Danny.

Daniel Konstantinovic (21:52):

Thank you. Always a pleasure.

Marcus Johnson (21:53):

Yes, indeed. Thank you to Victoria who edits the show. Stuart runs the team, and Sophie does on social media. Thanks to everyone for listening in. We hope to see you tomorrow for something. What is it? Weekly Listen, that's what we hope to see you for. We hope to see you tomorrow for the Behind the Numbers Weekly Listen, an eMarketer podcast made possible by LiveRamp.