Loading summary
Transcript62 lines
- [00:00]
A
Foreign.
- [00:10]
B
Hello. Hello and welcome to another episode of the Digiday Podcast, a show for anyone who's asked Chachi PT are you being paid to say this? I'm Kameka McCoy, senior marketing reporter here at Digiday.
- [00:22]
A
And I'm Tim Peterson, executive editor, Video and Audio Digital Media. What's up, Kamiko?
- [00:27]
B
Hi, how are you? Tim, I am indeed asking ChatGPT if it's being paid to say this. How are you?
- [00:33]
A
I'm not, because I'm on ChatGPT plus, which means I am not seeing ads. But I don't know how many people are seeing ads. But people are starting to see ads on ChatGPT, and so we are here to make sense of it. Well, maybe not so much you and I as our intrepid senior platforms reporter, Crystal Scanlon, who is with us back on the show today. Welcome back, Crystal.
- [00:57]
C
Good to be back, guys.
- [00:59]
A
All right, so Krystal, you have been all over this ChatGPT ad story for months now at this point. And finally, on February 9, after the super bowl, after Anthropic threw all the shade at ChatGPT for rolling out ads, ChatGPT has finally rolled out ads.
- [01:21]
C
It's like the beginning of the story that now is the next new story that we keep kind of going through.
- [01:27]
A
And so, Chris, like ChatGPT's initial ad product isn't super complex. It's like pretty limited. And there's precedent for this. Normally when platforms or streaming services like Netflix initially roll out their ad products, they're pretty basic. I imagine some of that is because they still have to figure out what works and what doesn't. Some of that is also to, like, manage expectations when it comes to performance. Can you give us a quick overview of what is this ad product that ChatGPT has rolled out?
- [02:04]
C
Yeah, of course. So basically they literally just sponsored ads or sponsored services that are literally just going to appear at the very bottom of the answers of ChatGPT. And the one thing that they've been very clear about is that they're going to be very clearly labeled and separate from the organic answer, because any messaging that we've seen from OpenAI is making the point that this is not going to influence what you're saying to ChatGPT and how they're going to help there, how it's going to respond to you. It's completely a very neutral. Well, ChatGPT is supposed to still remain neutral, but then the ads are obviously going to come in as and when it's relevant to whatever your conversation is within ChatGPT at that time, I'm curious.
- [02:48]
B
And I know it's early days, but given that it's a sponsored result within search, what does the pricing for something like this look like because you got on or how you even pay for it? Because on the one hand it's CPCS, CPCS and that type of deal. So how is money showing up here?
- [03:05]
C
Right, well, this has definitely been a massive conversation so far because apparently the starting like CPMs, because we're going with CPMs for this model, it's 60, $60. And literally anyone I've spoken to, I mean, that's putting it in the same sort of region as like premium tv. And like we already saw like we. When Perplexity tried to come out with high prices, no one liked it. Even when Netflix first did it, when they had their ad product, no one liked it and they had to lower theirs. So for a start, that's already just kind of we know what bucket that they want to be considered as that this is a very separate model and very different to what the other ones that are already out there in terms of why they've chosen cpm. I mean, anyone that I'd spoken to before literally was saying like, you know, for a search type model, you'd kind of almost expect it to be a cpc.
- [03:52]
B
But.
- [03:52]
C
But CPM is actually the model that is kind of leaned on for if you're doing brand sponsorships and things like that. Which is obviously where they're gearing this too straight up, ironically, albeit that they are still trying to go for search budgets. But it's probably just the most succinct, easiest way to at least test something to see if it's worth their time, how it's going to work, and if users are going to be okay with it before starting to delve into more, I suppose, formats that might be more complex.
- [04:16]
A
Yeah, it seemed like ChatGPT set the $60 CPM as a way to like set a really high bar in terms of who are the advertisers that are gonna wanna be in on this and what are they gonna be willing to accept here? Like basically, if you're paying A$60 cpm. And for context, when Netflix first launched its ads business, it came in with a $65 cpm that got a lot of pushback from TV and streaming advertisers who were just like, are you crazy? You know, Hulu at that time was like a 20, $25 CPM. And so there was like a big difference. Netflix has since come down, it's I think around 40 or it's in the 30s, I think of a CPM now. And generally the premium streaming services are like in the $30cpm range. So half of what ChatGPT is asking for here. But to your point, Crystal, there's. It's also being priced on an impression, so then it's an awareness play. So it's similarly like there's a, like a status that an advertiser would have in being one of the first advertisers on ChatGPT. And so it seems like it's almost as much about that as about like actual ad performance for an advertiser at this point.
- [05:35]
C
I guess it makes sense as well, just because, I mean, even looking at who they've been targeting for their first ad partners, like, they haven't gone to small brands really. They've gone to like, as we've even heard from adthena, who some of their clients have definitely taken them up on the idea. They've gone for the biggest of the big clients. We've got all the hold codes. They've come out now with public statements saying that they are literally their clients. They've got a number of clients, each of them that are now doing the ChatGPT launch. But I say launch ads, the pilot ads. We've got other brands. There's one called Fever Up, a live entertainment discovery platform. They've come out saying that they're doing it Williams Sonoma, that's broadly known. Bed Bath and Beyond the Not. So they're not targeting tiny brands by this. So obviously there is a status thing at play.
- [06:19]
A
But Bed Bath and Beyond being part of that group, I mean, you filed for bankruptcy how recently? And FeverUp, I've never heard of that company before.
- [06:31]
B
Well, is it a matter of like these premium brands, if you will, or is it about Whoever has the $200,000 minimum commitment to spend?
- [06:43]
C
I mean. Yeah, that's definitely a thing, isn't it?
- [06:47]
A
Yeah, yeah. And so that's been. So Adweek reported on the $200,000 minimum commitment. The information was the one who reported on the $60 CPM. Crystal, earlier this month you reported on what the go to market strategy has been. You mentioned how OpenAI has been often going directly to advertisers instead of agencies. Now some agency holding companies are on board. Omnicom, WPP and TO have all been announced as having clients in this initial test. But the idea that OpenAI was also going directly to brands is really interesting.
- [07:25]
C
Yeah, I suppose, like, it just goes to show that There was. They weren't just trying to just go for any old person. They obviously had like specifics as to who they're targeting and where they kind of envision this because they're clearly trying to see themselves as very separate to any of the other platforms out there. Any kind of interview. Whether I've seen Fiji Simo, their CEO of applications, anytime I've heard her say anything or anyone in that region, it's literally just we are doing things differently. We're not going to do the same model that every other platform has done. They're not necessarily saying that other platforms have failed, but they've obviously come with different learnings as to what has worked and what hasn't. And she herself, Fiji, she came from Facebook, she kind of has seen the playbook. She's got that literally at her disposal. A lot of her team are also ex meta. So in that regard, they've seen what works firsthand, they've seen what hasn't worked. And even on the most recent interview with her with sources, even she was explaining that the way that they came up with their principles about doing ads, it was down to roundtables of over 100 people within the OpenAI company. That's how they decided on their principles, because they were so passionate about ads. But they also want to do it in the right way that won't completely disrupt the user experience and effectively alienate their customers or their users. Because without users, well, they'll just become another sad platform story of we no.
- [08:49]
A
Longer exist, at least for now. I feel like Facebook made similar claims initially. Google made similar claims initially. To what extent they've held to that, I think anyone who's used Facebook or Google can probably attest to at this point.
- [09:05]
C
I mean, to be fair, I say that, let's be honest, every single platform up until this point has also said they'd never touch ads and look where we are.
- [09:12]
B
Well, I want to kind of compare and contrast this to perplexity, right, because they kind of took the first stab at what this looked like. And to your point, OpenAI has kind of had some learnings. What, in your opinion, kind of went wrong as you guys went through the reporting with perplexity, that's kind of set the playing field for OpenAI now.
- [09:30]
C
I think the one thing that I suppose there are similarities actually, and differences. One of the main kind of problems, I think that definitely that Perplexity came up against was just, I suppose the attitude that they came with. It wasn't like, by all means, be very bold, be Very confident in your own product. But when you're already going to a very well established market, if you are kind of going with the attitude of, well, if you don't like what we're doing, then we don't want you anyway, that's not really the best way to get anyone on board with anything. And that was kind of the impression that I was given anyway with the messaging that I'd seen with just how bold they were about like, well, this is what we're offering, we know it's going to work. And just very kind of outlandish statements like that. The similarities, I guess, that I've seen between the companies is just the lack of anyone being able to get in touch with the right people. Because with perplexity, I mean, I have yet really to speak to anyone that has had proper, proper discussions with them when they were doing their ads, kind of their ads project. I mean, the few that were involved were like, we're going to spend minimal just to see how it works. But even they weren't necessarily convinced by it because it didn't follow this proper search CPC model. And then that along with the high pricing just really put them off because they were like, you're asking us to think of it in a search way, but you don't want to manage it or measure it in a search way. So that just doesn't add up for us. And then granted, we've seen what happened there. They've kind of had to put a pin in their ads strategy and then now they're focusing on their other things, like their browser and things instead. But with OpenAI, I mean, we know already that they're supposed to be looking for an ads boss. I have been trawling literally since I heard this like late last summer to find out who this individual is. If by any chance they're listening to this, please reach out. But at the same time, I would say that I don't even know if this person has been hired yet. Like from anything that I'm hearing, it sounds like Fiji is somewhat potentially part of those conversations. I mean, another conversation that I had literally even before this. The individual I was chatting to was trying to get in touch with OpenAI to set up a meeting about the ads in Q1 of this year. And even with them, I was like, how did you get in touch? And they were like, literally had to trawl through LinkedIn and try and find someone on their go to market team. Which, I mean from my experience and from every trawling that I've done a lot of that is to do with their enterprise sales. So it seems like they may be not they don't necessarily have a completely separate ads team as we know it from any other platform, at least not right now. However, I would imagine that they are in the works of trying to get that in place because obviously we've got possible coming up and can and from my conversation that I had with OpenAI last year, they were very much already considering what they're positioning and what they're going to do in terms of showing up at Cannes this year.
- [12:19]
A
And it's going to be interesting to see to what extent they run ad sales as a separate purchase path or if it becomes a bundle thing where similar to how Google and Amazon can be with their respective cloud platforms. But hey, sign up for OpenAI Enterprise to use to have GPT power platforms inside a business and as part of that you get ad credits or even for OpenAI to just go straight to Omnicom, WPP Dentsu and say, you know wpp, you have WPP open, that's your AI platform. You're going to be using GPT as one of the models in there. Let's do an enterprise deal for that. And as part of this we'll also give you credits for your clients to be running ads on OpenAI. Because OpenAI really likes its circular deals. And so I could see there being a circularity to how the ad deals get structured too.
- [13:26]
C
I suppose in that regard they're not wrong in the way of if you've already got business with them already, why not just capitalize on it and get as much out of the one kind of partners as you possibly can because there's already a certain level of trust built there. And I guess the fact that they are trying to be the pioneers of AI, be the front runners with absolutely everything, there probably is a kind of, do I want to say innate trust. There may be a certain level of we trust you more than others right now in terms of how well they will be able to pull this off.
- [13:55]
B
I'm also, I'm so curious about like the inner workings of this team that may or may not exist. I'm not, I don't know, there's no clarity there on who their ads teams, if their ad team exists. Right. Because one of the pain points that I hear when I talk to media buyers is that, you know, especially with like meta's notorious for it, I can't get in touch with anybody for, you know, whatever is happening with my campaign, which becomes like A frustration point. I think media buyers just kind of like take it on the chin with meta because it works. It is unclear if that same grace will be extended to something like OpenAI until there has been a test that's been kind of proven. Right?
- [14:34]
C
I think. Yeah, like, I suppose right now, because they've got that kind of, you know, oh, this is the shiny new thing. There's so much buzz around it that that's obviously kind of given them the front point of everyone just saying, like, we just want to be a part of it. So that's probably worked in their favor because there's like limited supply or so they want us to believe there is limited supplies to who they're going to work with in this initial test. How long they have that grace, though, is tbd. Because I guess really as much as any advertiser, as big as they are, as much as they may want to be quite happy spending $200,000 on just an initial test run over a number of weeks, that's fine for now, but I'm sure that they're not just going to keep handing that out if they're not going to get sufficient returns and they're not just going to like pop that back into matter if it's not really perform how they want it to.
- [15:19]
B
And is this like a. Do you have any insight onto if these are managed, serviced or self. Managed service.
- [15:26]
C
Self service right now, as far as I know. I think there is kind of like slight rumors swirling that they are going to be building something to enable them, enable advertisers to do that. But I think even just having seen how any of the other platforms have behaved over the years, building anything like that takes such a lot of infrastructure and such a lot of investment. They would never be able to do that as their first offering. I think it's really just when it comes to this test, it's great to be a front runner in terms of an advertiser. Yes. We were one of the first to test. We help to really position and decide what it looks like going forward. But it's probably OpenAI getting far more out of it because they're the ones doing the learning versus the actual advertisers who are just like, we just want to see if this works or not.
- [16:13]
A
And I wonder if it's historically, yes, it took a lot of time to develop an ad platform. I wonder if at this point they could just throw Codex 5.3 at it and say, build me an ad platform, go get a coffee, come back and bam, they have their ad platform.
- [16:32]
C
I mean, it wouldn't surprise me. I mean, the fact that they're literally wanting ultimately, like ChatGPT and like their other apps and everything to basically be a personal assistant to help you run life efficiently. I mean, that would be the most efficient way buy ads going forward.
- [16:45]
A
Yeah, and one thing that stood out to me. So Omnicom, in their press release announcing the ads pilot had a quote from an OpenAI exec. Usually this would be a quote from the head of ad sales or the head of ads. It was not in this case, at least not in terms of. It wasn't a Linda Yaccarino, Peter Naylor type. It was Asad Awan, whose title is Ads and monetization lead at OpenAI. Asad is the former VP of creators and monetization for Facebook, and According to his LinkedIn profile, that meant he led the product and engineering teams on creators and monetization for the Facebook app. Before that, he was the director of monetization and public content for Facebook, which according to his LinkedIn bio, meant that he worked on, quote, all the way from ads in feed to stories to content monetization and direct user funding for content and communities. So this is a person who's a product engineering person, not a salesperson. Another person who posted on LinkedIn from OpenAI is Vijay Raji, who's the CTO of OpenAI applications, and he posted a week just days before the pilot program officially launched that he's building a monetization team with Asad Awan. And in his LinkedIn post he wrote, quote, this isn't about retrofitting old ad tech models. It's about inventing something new aligned with our ads principles around privacy, trust and user control. So they're building something. Maybe not a sales team so much as a sales product or a sales platform. To Kimiko, your point? Maybe this does go self serve.
- [18:34]
B
That'd be interesting. And something that a lot of buyers are pushing for, especially in the idea of like, data and how this is measured and how it performs and things like that. Being able to do that yourself as opposing to having a managed dashboard where they just kind of like, you know, hand you your lunch. Best of luck. May be a better selling point for open AI. Not that I think that they need a heavy one right now, given that they're the shiny new object here. To your point, Crystal?
- [19:07]
C
Yeah. I mean, if they did come out with something like that a lot sooner, which by all accounts they do have the capability to make it happen, I mean, it would kind of go along with their even narrative. We're not doing everything the same way as everyone else has done before. I mean, I find it quite funny in general. Whenever I see execs come out and say we're not copying everyone else, like what everyone else has already done, I'm like, yeah, for now. Is all kind of the way I see it.
- [19:29]
B
Because usually, and Tim, I'm going to lean on you here. These start as managed services. No. Because it's easier of a lift than having to actually build out a platform where somebody can do it. An advertiser scheme can do it themselves.
- [19:41]
A
Yeah, yeah, they start as managed platforms but then when they want to scale in terms of bringing on more advertisers, and the reason to bring on more advertisers is that's more competition, law of supply and demand. You get more demand and their prices are going to go up. Then you move to self serve, you know, open, open it up to DTC advertisers, small and medium sized business advertisers, the advertisers. But for that you need scale in terms of audience. If there's five people who are on the ads tier, that's not going to work for ChatGPT to get advertisers on board because then all the advertisers are going to be competing just to reach those same five people. But it is going to be interesting to see how ChatGPT develops this ad offering in terms of the capabilities because right now it's, it's not really offering much to advertisers. Crystal, correct me if I'm wrong, but like advertisers get to know aggregate views for their ads and aggregate clicks, but they don't get information like what was the conversational context in which my ad appeared? Or anything about like who was the person who clicked on my ad or was shown my ad.
- [20:57]
C
Yeah, it's really bizarre in one sense. Like, don't get me wrong, I get this is a very early pilot, but at the same time these are kind of like the basics, basic things that advertisers would want to know, to know whether it's worth their time or not. And it's just so unclear right now as to what ChatGPT or OpenAI, what they're actually going to give them in terms of information. Who knows? I mean, whether they keep it really gated, I don't know. I would assume not be in their best interest to do that, but how much they're willing to share because I mean, I guess the other side of it is if it's not performing as well as they had hoped, yes, you could say it's a learning curve, but if it isn't, it's. You've got more of a chance of these advertisers just saying, you know what? I'm just, I'm out, like, I'm done with this.
- [21:38]
B
But on the opposite side of this, they, I mean, all platforms right now, but most notably ChatGPT and OpenAI, are fighting kind of an uphill battle with consumers and trust. So if there becomes. And part of the thing with the sponsored post, like going back to our original question at the top of the top of this show, is this, you know, being sold to me as a sponsored response as opposed to an organic response. So at the same time that, you know, it would make a better deal to provide that information to advertisers to make the juice worth the squeeze, you also risk alienating consumers. If you dig too deep into that and like, you know, then there becomes like a breach of data privacy and things like this.
- [22:16]
C
I mean, Fiji's definitely come out saying that, like, from all accounts, like, they are not selling anyone's data to advertisers. She's also said that as far as she's concerned, as far as the team is concerned, when it comes to these ads, ChatGPT remains neutral. And the example that she gave when speaking with Alex Heath was the fact of, say, if there were two ads that showed up within a conversation and if that user then asked them, like, oh, what do you make of these ads? ChatGPT isn't going to have a clue because it's not actually looking at the ads. You have to literally actively say, look at them. Click here to look at them. And then that ad, the ChatGPT will then come out and say, like, you know, this ad means this, it's good, it's bad, whatever it wants to say. This one is slightly better than the other one's worse than this one. Where you could probably get an issue there, though, is obviously if ChatGPT is coming out with, say, potentially some negative stuff about one company versus another, that's the whole thing down to what it's aggregating in terms of information about the company's reputation across the Internet. From Fiji's perspective, I don't think they should really be worrying about what ChatGPT is saying about them. I think they need to be worrying about changing that reputation so that ChatGPT can say more positive things.
- [23:24]
A
Oh, which then gets into the whole answer Engine optimization, generative engine Optimization conversation. But part of those optimizations are going to be, I can already hear the advertisers and agency execs saying, well, then give us more data so that we can make these optimizations. Because without that data, we don't know what levers we should be pulling or not pulling.
- [23:50]
B
Do we see us full circle starting to happen here?
- [23:53]
C
100%.
- [23:54]
A
I mean, it's always going to come back to the advertisers are always going to say, we need more data. And the platforms or the ad sellers are always going to withhold the data because they understand that's their leverage. If as soon as ChatGPT or OpenAI starts sharing with advertisers, not only here's how many people saw your ad, but here's what they clicked on, here's the context of that information, all of that, then now it's set the standards in terms of expectations from advertisers. And so if that performance starts to degrade at all, the advertisers are going to say, you know what? We're not getting enough. Our prices need to come down. And so by ChatGPT keeping its cards close to the vest, it can keep its prices high. It'll still get pushback and obviously advertisers opting out, but until it'll start sharing those numbers, when those numbers are big and favorable.
- [24:54]
C
I feel like it's literally kind of even part of their narrative as well, though, because even the guidelines, which we've seen before, they had this pilot, even in the guidelines, the OpenAI was very clear in the sense that, no, no, if you're going to say anything publicly about this pilot, number one, you make sure that you only talk about it as a pilot. It is not a launch product. They were very clear about that. And then the ultimate thing was any kind of public messaging has to be kind of, you know, we have to approve it before it goes out, and you can only do yours after we've actually said publicly that we're doing ads. So it's OpenAI controlling the narrative, which makes sense that they would want to, because they already know bringing ads in is already going to be a risky business for them in terms of whether or not it's going to alienate users. There's already like, how many subreddits rather out there where people have already just been saying pages and pages? Once they come, I'm out, I'm going to Claude, I'm going to Gemini. They want to avoid that happening because at the end of the day, they need users because Users bring advertisers.
- [25:57]
A
So we've talked a bit about the outlook for OpenAI in terms of what's it going to need to do to develop its ad offering to get more advertisers on board to the extent that it wants to. And really that's going to be more capabilities, more performance based pricing, I think will be the main thing that would get advertisers on board because then they're going to say, okay, I don't really care about the price if I'm only paying if people clicked on my ad, because then I know I'm paying for performance, I'm paying for results. There's a clearer ROI to that than just like maybe people saw my ad, because I think I don't Even know how OpenAI is loading these ads in ChatGPT, where if a person scrolls down to that ad, is that when the ad effectively gets served and measured as an impression, or is it just if it's slotted into the conversation but doesn't actually appear on screen, does it still count as being served? I don't think we know that.
- [27:01]
C
No, I don't. And I'm literally going to have to rely on you guys when you actually get them because being in the uk, they're not even doing this over here yet. So how this actually appears is going to be interesting. But I wouldn't, I mean, the fact that we've seen brands come out already now starting to say that they're part of it, I wouldn't imagine it's going to be that much longer until we start seeing screenshots of what they look like across the Internet.
- [27:23]
A
And so then, like, I guess, you know, another question is going to be how is this going to impact the competition for these ad dollars, specifically Google and then also Bing, because there was an agency exec I was talking to who said they had clients get approached by OpenAI about this pilot. And in the initial outreach, OpenAI asked those clients for their total monthly ad spend, but also their total monthly search spend. And that the idea was OpenAI was going after share of search spend.
- [28:01]
C
Yeah, I guess to be fair, that does make sense. I mean, even with the verticals that they've said, the agencies have come out and said our clients are across these certain verticals, they all lean into search in that regard.
- [28:14]
A
Yeah. Well, I wonder if Google has such a hold on search ad budgets at this point that ChatGPT is going to really face an uphill climb in getting that money. Or if. Because. Or if those advertisers are looking for competition and are willing to take money away from Google search. And part of this is Google has said we're not putting ads in Gemini. That said, they are putting ads in. Yeah, well, although I feel like Demis Hassabis was pretty definitive that, like, that's not happening. That said, he's in charge of DeepMind, he's not in charge of the ad organization. There's always internal battles at Google between the product and engineering folks and the ads folks. But I mean, Google has ads in AI overviews and it's testing ads in AI mode. So it doesn't really need ads in Gemini to have ads in conversational AI. They're there.
- [29:14]
C
No, I mean, when I was even speaking to Dan Taylor from Google recently, I asked him straight out, like, are you doing this? Because last year I've even seen different headlines out there saying one minute that you're doing it. I've spoken to people that have suggested, oh, that when you decide to do it, you've got a playbook of what works. You're just going to do the same playbook. But when it comes to Gemini, and he categorically said, we are really right now focused on overviews and AI mode, we're not really looking at anything to do with Gemini. I kind of always, I suppose from anything that I've spoken to with any of the platforms, there's always the platform dance about ads and when they're going to land them in a certain platform and it's normally just once it's reached a certain level. Until what last year, Google had the antitrust trials, which obviously kind of put everything else probably on the sidelines just while it had to sort that out. So that took up most of their focus. Now that's completely clear. They've got a whole free run and we can already see how much Gemini really bothered OpenAI from the fact that Sam Altman literally called a code red over oh God, our chatgpt isn't even as good right now. So then that really caused them to up their game. But even with the idea of how ChatGPT has come out with ads now, even that doesn't seem like it's when it was supposed to happen. Something clearly went down between Christmas and 16 January when they announced that they were going to start doing this in the near future. Because up until that point it was very clear from Fiji's perspective that they wanted to make sure the commerce angle was sorted, that they had privacy down and that they had their as person in place without these things in place, they weren't even going to consider it. But as she said now, more recently, they've got 800 million users. So there gets to a point where if you don't want to have a complete divergence in the sense of how much intelligence is in the paying tiers versus the free tiers, you've got to find money from somewhere. And we already know that OpenAI is also burning through a lot of cash. And over and over again with every single platform, the ad model is the one that is consistent. It's the one model that has allowed every single platform to have consistent revenue so that they can focus on other things. Mark Zuckerberg from Meta even said on the most recent earnings call that literally it's going to be their advertising dollars that pay for their AI projects. So it makes sense why they need it.
- [31:35]
B
No pushback there. Agreed.
- [31:40]
A
And for this question of where that money is going to come from. So I'm sure it'll come from Google to some degree, but my expectation is it's going to come from Bing. So there was an agency exec I was talking to recently who said as OpenAI was reaching out to advertisers with this test. So this was after the January announcement that they were going to be launching this pilot program. This agency exec put an ask out to all of their clients. Hey, if we were to take your Bing ad budgets for a month and redirect them to OpenAI, how many of you would do it? And 45 clients said, yes, I would do that. And what's interesting about, I mean, this is Bing that we're talking about. So there's definitely a willingness to move money away from Bing. But these are also search advertisers willing to move money towards ChatGPT. And if they do that with their Bing budgets and they see performance from it, that could open up Google budgets because they're effectively going to be grading against the same measures.
- [32:53]
B
I think that all comes performance is like doing a lot of the heavy lifting here. Right. That is entirely what this is going to hinge on. Because one, you've got the continued theme of do more marketing with less marketing dollars. So if you shift those dollars out of Google, which is known to work and has been a proven and effective channel for God knows how long, and to put those into something with a high CPM and an unproven theory because they've just gotten started, you know, it's a big risk. So, you know, going back to your point, if OpenAI is able to prove that so like, they've really got a lot riding on this pilot program right now. That'll kind of be the determining factor if those dollars start to shift from Bing. I can see it from Google. They got a little bit harder of a fight. You know, the test dollars might happen, but to consistently usurp those dollars, that'd be a different story.
- [33:43]
A
Well, and then do they also chip away at TikTok is considered search? Do they chip away there? Amazon is in the search realm. Do they start to chip away there? Do they start to piece together the spending from all of these other businesses where at some point we look at it and they've built advertising Voltron.
- [34:07]
B
But also maybe that's a strategy that they do be, I don't know before they get, you know, go after Google. Because, like, to your point, I would imagine advertisers would be a lot more willing to let go of some of these other channels than Google. Google and Meta as we know, has the industry in like a chokehold. Right. So I feel like that would be what's a super villain that you got to go up against at the very, that's like, you know, search ad final boss. Yeah, that's Google is Thanos. So maybe you can, you know, collect the Infinity Stones Amongst Bing and TikTok and Amazon. Right. And once you've kind of proven the case for the dollar shifting there, then you move on. But that's my theory.
- [34:43]
A
But I think the buyers want four horsemen of the ad apocalypse. So we have the triopoly. Well, we have the triopoly with Google, Meta and Amazon. And the competition from Meta against Google was welcomed by ad buyers because it put pressure on Google to like manage pricing, make it more willing to cooperate with advertisers. Amazon was very much mentioned because then that created competition for, at that time, the duopoly of Google and meta. Now if OpenAI comes into it, that can create even further competition. And at a time where like Google's ad tech's business is under siege because we're still waiting for the remedies from the ad tech antitrust trial, the search business is just undergoing a pretty big change with AI mode and AI overviews. And then there's everything going on with Meta, which has been on a rebound since the pandemic from an advertising perspective. But that likely means prices are getting higher. And so advertisers may want more competition. And so there may be some goodwill that OpenAI could capitalize on here of just creating that competition for the big tech companies. But obviously all of this is going to have to remain to be seen. Crystal, we will be leaning on you to keep coming back to the show to help us understand how all of this is playing out.
- [36:16]
C
I'm sure this will be the first of many chats we have.
- [36:19]
A
Absolutely. Well, Crystal, thanks for coming on.
- [36:22]
C
Thank you for having me back, guys.
- [36:30]
A
Thanks for listening to this episode of the Digiday podcast. If you enjoyed it, please leave us a rating and a review on Apple Podcasts, Spotify or wherever you're listening. Get more from Digiday with our daily newsletter sent out each weekday morning. Visit digidair.comnewsletters to sign up.