Loading summary
Transcript49 lines
- [00:00]
A
Foreign.
- [00:10]
B
Hello. Hello, and welcome to another episode of the Digiday Podcast, a show for anybody wondering, is OpenAI's ad business moving way too fast? I'm Kamika McCoy, senior marketing reporter here at Digiday.
- [00:22]
A
And I'm Tim Peterson, executive editor of video and audio at Digiday Media. Kamiko, do you feel like OpenAI is moving too fast with its ad business?
- [00:30]
B
Tim? I think the fact that we're probably gonna have to timestamp OpenAI conversations going forward says a lot about how fast they're moving.
- [00:39]
A
Yeah, I feel like we already have to timestamp. So for this episode, we talked to senior platforms reporter Crystal Scanlon. We did the recording last Wednesday. Since that recording, Crystal has reported that OpenAI's ad prices have dropped from $60 at launch to as low as $25 now. So that's how quickly these things are moving. That said, I feel like I'm now, after talking with Krystal, kind of of the mind that, like, is OpenAI moving fast enough on the outside?
- [01:15]
B
Yeah. I'm really grateful for her being able to hold our hand through this. I feel like I'm running through the woods here, trying to keep up with everything that OpenAI has got going on. Question here is just, like, how much of this is. You've got the playbook from other platforms, you take it and run versus them as the first AI chatbot ads business, having to build it out themselves. So Crystal does a really, really good job of explaining all her thought process there and allowing us to weigh into.
- [01:45]
A
Yeah, because, I mean, even there, like, is it the first AI chatbot ads business? Because. Oh, and you have perplexity, but that, like, went up in flames super quick. But as we get into the episode, like, as much as Google hasn't put ads in Gemini, its AI chatbot, it's effectively put an AI chatbot into search with AI overviews and AI mode, and it's putting ads there. Which is where I kind of come back to the idea of, like, okay, sure, OpenAI has been moving super quickly in the two months since its ad business has launched. That said, Google's so far ahead that maybe OpenAI actually isn't moving quickly enough.
- [02:27]
B
Yeah, that's a really good point and a really good question. And it's like, they're like, so much of the conversation is about them, like, burning cash and things like this. So pedal to the metal seems rational.
- [02:39]
A
Yeah, it's pedal past the metal.
- [02:41]
B
Yeah. Flooring it.
- [02:44]
A
I'm excited for folks to hear the conversation because, yeah, I came into this conversation very much thinking, eh, OpenAI is moving super fast like this. Is this too fast? And I think Crystal brought a lot of information and insights that has me rethinking things.
- [03:05]
B
Absolutely. Well, I'm sure our listeners have similar questions, so without further ado.
- [03:17]
A
Crystal, welcome back to the show.
- [03:19]
C
Good to be back.
- [03:21]
A
Yeah. We had to have you back very quickly to catch us up on what's been going on with OpenAI. Because it's been two months roughly since OpenAI rolled out ads in ChatGPT. But just within the past month there's been a whole ton of developments. Kimiko, help me out here. So what's happened? They've added Criteo as a demand path for Advertisers to access ChatGPT inventory.
- [03:48]
B
Yep. And then they got smartly on board for creative optimization.
- [03:53]
A
They hired former Meta exec Dave Dugan to oversee ads.
- [03:57]
B
They also launched a self serve ad buying tool.
- [04:01]
A
And then Crystal, you just reported on how they're launching Pixels for their ads as well.
- [04:08]
C
Yes. It seems like at this point they're like running at lightning speed, but I mean, unsurprising. Did we expect them to take long? No, just didn't think anything was going to happen this quick.
- [04:18]
A
Right. Yeah. Which kind of raises the question, Kimiko and I have been bouncing around. Are they moving too fast?
- [04:26]
C
I suppose I wouldn't necessarily say it's too fast, but it's definitely, I mean if we compared to how like, I mean with Meta, I think theirs was like three years launching after ads with their ad platform. TikTok at least took two years. Google I think did it within a few months when they launched AdWords. But I mean if they're going up against all these major players, they have to be able to compete in a space kind of almost adjacent like next to them. They can't have anything that's not equivalent to what the different tools and managers that they have. So in one sense it's almost like they have to catch up and catch up fast. I guess the issue that they have is that they also, it's a completely uncharted territory. They're doing it in a space that no one has been able to do it before yet. So that's the kind of bit where it's very trial and error but probably what works in their favor. I mean they've got a lot of ex. Meta execs there and they've obviously been the ones that were building out Meta's ad platform and ad capabilities back when, before Meta became the beast. That it currently is. So they have a playbook that they can use and then obviously add their own frills and bells and whistles and anything to make it an OpenAI standard.
- [05:39]
B
The fact that you don't have whiplash from all of this, truly astounding world class reporter. But I do have to you bring up like that they've hired, they've got VGCMO from Instacart and they've got MetaExecs and they've really built out the tame for their playbook. But the question still kind of remains for me, like how much of this is because they had a playbook versus an act of desperation because they're still burning through cash?
- [06:03]
C
Yeah, I definitely think that. I mean there's no question like the ad businesses, I mean it's the one proven track record and one proven driven path to have consistent revenue. And OpenAI is no different when it comes to that. And the fact that they've already I think forecast that they're going to make what, 102 billion in ads by what, 2030. I mean looking now, we're already in April 2026. That's only like four years away. And to be able to get to that caliber, you cannot do it with just Holdco money and independent agency money and just relying on those big deals because let's be honest, the way that the economy is, budgets fluctuate from time to time. And then if, if one of those major budgets goes, that could be a massive detrimental impact to their bottom line. They need to have the SMBs on board because then you have the scale. And that is exactly what's been proven with whether it's TikTok, whether it is Snapchat, even Pinterest, they've all now started saying it on their earnings. We know that they are providing us with consistent revenue and, and when they start having their doctor platforms and everything available to them, that is what they have seen already proven and works really well via Meta and Google. That's how come Google and Meta have been able to become as big as they have done so far. Because they've been able to rely on that revenue.
- [07:20]
A
Right? Yeah, exactly. It's the Google Meta playbook. And it's when there was the Facebook boycott a handful of years ago, why Meta didn't actually see that big of an ad revenue hit because it was big brands who were boycotting Facebook. But you had all these, this huge long tail of small and medium sized advertisers that were still spending there.
- [07:43]
C
Yeah, I mean the amount of times I've spoken to people and like, say if I even just take with Meta's Advantage plus, for example, it is very difficult to have a conversation with anyone about Meta's Advantage plus and for them to not bring up how infuriating the level of bugs that they have in that platform. And I've even often asked them, well, if it's so problematic, why stay there? Why not go elsewhere? You've got enough platforms in the landscape now, why not just take those dollars and move them elsewhere? And it's the same answer I'll always get. I can't do that. Meta still performs. I can't just go to another platform because I won't get the same level of performance as I would on Meta. So the bugs have kind of just been as much as we get, we get annoyed by them. I just have to put up with them because I know I'll get the results on Meta.
- [08:26]
B
Well, here's something else because like, if OpenAI is making this pitch for smaller businesses and whatnot, didn't they also just reduce what the kind of the barrier to entry was, where they had gobs and gobs of money that you had to pay them even just to get your foot in the door and now they've reduced that. You reported on that too, right?
- [08:42]
C
Yeah, it was literally so in the initial app pilot when it was brought in, the asking price was literally, I believe, between 200 to 250,000 for commitments. And obviously at that stage, as much as it's kind of a new, everyone's really excited, it still does fall in the bucket of being just a very emerging platform. It's still very untried, untrue. We still don't know how it's going to play out or if it's even worth our time. So that's why obviously it did put a lot of advertisers off even considering it if they were asked about it initially. But now they've obviously got Critio, which naturally has then brought in a lot of other advertisers already. So advertisers now have different options to get into the pilot initially anyway. They can go direct via OpenAI, they can now do it via Critio and they can even do both if they feel like it, going both directions. But now to make it even more accessible, it's now been dropped for the initial commitment to 50,000.
- [09:39]
A
Yes, Lord of Mercy and the self serve tool. So then like small and medium sized advertisers love a self serve tool. I mean, even in the streaming ad market, we're seeing self serve tools be rolled out. Hulu did this years ago, but more, I mean, even just this morning as we're recording it, the ad Exchanger reported on Televisa, Univision rolling out a self serve tool. And that kind of indicates also with the developments and how quickly OpenAI has been moving, that it has a lot of inventory to fill. Like it wouldn't be bringing on Critio as a demand plath. It wouldn't be adding a self serve ad buying tool if it didn't have enough inventory to fill. It seems like its issue at the moment is it doesn't have enough demand.
- [10:28]
C
Yeah, I think in general, the only reason where, I mean, obviously there have been the reports out there that the performance hasn't been great. Oh, it's not what I expected. But that was kind of part of the course with doing this pilot. Everyone knew it was like, we need to see how this is going to work. Because ultimately what they can't do is destroy user trust. This is going into a completely different type of a platform. It's a chatbot. They obviously have to stick to guidelines. Well, I say guidelines, the guidelines that they are creating about certain types of conversations, if they're too sensitive, if it's about health, there are certain areas where they can't really put advertising in it or else it's going to really mess up the way that users even feel about how OpenAI is handling them as users and their data and what they're doing with it. But I think even with their terms and conditions, I've now seen a recent display of the terms and conditions for advertising. They've definitely said they're not going to sell the user data. I mean, who knows? I mean, I don't know whether or not terms and conditions, if they can be tweaked or how often they can be tweaked. I would have thought if this gets a lot of traction as time goes on, then surely at some stage we will start seeing various different tweaks. If, say, for example, there is a hell of a lot of demand then I mean, because at the moment the pilot is still only in the us, we've now obviously seen a few different other countries added. The UK still isn't even one. But I would assume if this is going to go global, we'll start seeing certain things be tweaked, maybe certain very strict regiments, that they'll relax slightly, shall we say.
- [12:02]
B
Do you get the sense that there's a method to the madness? Right. Because even thinking about their move with Sora, where they had inked a deal with Disney and everything seemed to be on course and they were like, you know what, nevermind, back out. We have something else more important to do. So I do wonder, like, given that this is a new space and there's still kind of like a high threshold, even though the barrier to entry has been lowered, there's still questions and things like this. Realistically, how, how fast can you afford to move? How like, let's move fast and break things. Can you afford to be here?
- [12:33]
C
I would say it's one of those things of I. Weirdly, it takes me back to what Mark Zuckerberg said like years ago, like move fast and break things, that's the end of it. And it seems like they're actually doing that. Like, why not? I mean, nothing at the moment is stopping them because they are like the front runners. I mean, obviously, I suppose for them they've also got competition. As much as Anthropic doesn't have advertising, they do have user support and obviously after the pentagon deal that OpenAI did. Yeah, there is an element where, I mean, I don't know, obviously the updated numbers, but there was a definite spike in terms of uninstalls from ChatGPT and then a lot of users suddenly appearing on Claude, for example. So they've obviously got the user support there. We can't discount Google because Google is traditional search. And yes, at the moment Google's Gemini doesn't have advertising, but at the same time that's only because they still have AI mode, AI max that they're focusing on. There's nothing to say that they won't have advertising on it in the future. Meta's already brought themselves back in the game with their recent launch of their latest version of their model. Muse is the family and then Spark is version one. So I've even heard from people that, you know, before it was kind of like Meta, what are you doing now? From that launch, it's their best one that they've had yet. The first one that they've literally launched through Super Intelligence. And also it's like way better apparently than Llama from when they've previously launched that. So Meta's now kind of like back in the game with all of this. I think when it comes to AI at the moment, it's such not only just a buzzword, everyone wants to be the front runners, the ones to actually break through and actually make it successful first because no one's ever been able to that and maybe it's also down to the fact that other than really what Google Meta, Twitter in its heyday, and TikTok during the COVID pandemic, it's been very hard for any platform to really break through and not only break through, but stay and have staying power, like for absolute longevity. So maybe that's also part of the issue that we're seeing here because they're all racing to basically be the first.
- [14:35]
A
As much as we started this conversation with the question of Is OpenAI moving too fast? There's also the maybe even more valid question of like, is OpenAI moving fast enough? Because you have people like Ben Thompson from Stratecheri and Eric Seufert from Mobile Dev Memo who have been banging the drum for years at this point of like, OpenAI needs to introduce ads to ChatGPT, it needs to introduce ads to ChatGpt. And then it finally did and they're just like, oh God, this is kind of late to introduce ads because Google especially already has a search ad business that ports immediately into AI. And Chris, like you said, ads aren't in Gemini. They don't necessarily need to be in Gemini. Like, I'm still kind of of the mind of I'm sure Gemini as a product will continue to exist, but I don't think that's going to be Google's AI product. I think search is going to be Google's AI chatbot with what they're doing with AI mode as like a Trojan horse. So it's like OpenAI has to catch up as quickly as it can to where Google already is.
- [15:39]
C
Definitely, definitely agree with that. And even, I guess, even if we take Meta for example, with any of the platforms, say with Meta, when it launched Threads, everyone was like, oh, well, what's happening about ads there? Completely different platform, I get it. But also it's Meta. Meta manages to monetize anything it has, so it was only a matter of time. So Google's the same. Wouldn't be surprised if at some point down the line Gemini has ads, but it doesn't like. It's not its main bulk, as you said, it's not its main bulk of revenue. So there's no rush in a sense for Google to make that happen. With OpenAI, it's in a very different situation. It's not yet a profitable business. It literally is burning through so much cash and advertising is the one way that it can potentially relax some of that stress, especially with obviously there are numerous reports right now the company is potentially going to go public. I was going to say Private go public in Q4 this year that, I mean, we're already, what, halfway through Q2, so that doesn't really give them much leeway or much time.
- [16:40]
B
Whose playbook are they able to pull from here? Right, because what I think is interesting is that this is they're monetizing AI, which hasn't been done before. So realistically, like, do you pull from Meta Social playbook? Do you play from Netflix streaming playbook? Who are they able to turn from or who are you kind of seeing them pull from?
- [16:57]
C
I would say right now it's definitely Meta, but that's also because Meta as much as, and obviously Google and Meta are definitely on par with each other in terms of their successes. At the same time, Meta is always that one that's used as a prime example of this is what a robust advertising business looks like when it comes to anything to do with any of these sorts of conversations. If they weren't leaning on Meta's playbook, they wouldn't have had so many people from Meta literally coming into the company fair. So it's sort of like, I suppose the way I kind of, when I'm kind of thinking about it, I just see it more as like, it's like Meta 2.0, but under like an OpenAI guise. Like it will definitely have its own little like flair, but ultimately like the real nuts and bolts of how to put an ad business together. These are the same people that made Meta successful. Meta is now flourishing thanks to these people that are now working at OpenAI,
- [17:51]
A
which makes a lot of sense because like Google obviously is successful. We'll see how things pan out. But E marketers projecting that this will be the year that Meta overtakes Google when it comes to ad spend all kinds of asterisks on that because that's just forecasting. And we're only what, three and a half months? Plenty can happen between now and then. But Google's dominant advertising business is search, which is direct response, it's last click. It's very like pure performance in a lot of ways, to the point where Google still kind of has to make the case of like, there are brand opportunities in search to advertisers because advertisers still think of Google search as I'm just trying to like sell product here, sell a service here. Whereas Meta has been able to strike that balance between performance like you were talking about, but then also brand by getting in people's feedback to expose them to new brands, but also to get them to then buy from those brands. And it seems like that's where AI search is kind of landing at the moment, is like. So OpenAI has come in going hard at brand advertising to the point of being criticized for not offering enough for performance advertisers. But Dan Taylor from Google, who we've had on the podcast before, was just on the Mobile Dev Memo podcast with Eric Suefer and he was making a case around what they're seeing with AI overview ads is that they're more suited to brand advertising than necessarily performance, that they're more brand suitable or they're more brand advertising like, than traditional search ads. And then Business Insider had a story in the, I think the CMO Insider newsletter where Lara O'Reilly cites a study from FTI Consulting that was like 22% of, I think it was like 18 to 29 year olds. So like Gen Z, AI search changes their perception of brands. So there's a brand angle to AI search, to these AI chatbots. And so when I think about it in that respect, Meta should be the model for OpenAI is probably the model for Google when it comes to AI. And maybe not so wild that OpenAI went so hard at brand advertising off the bat and didn't have performance tools. But you've been reporting that performance tools are coming.
- [20:30]
C
Yeah, I think also as well, what can't be forgotten, it's great that they started with brand advertising, albeit, yes, it's been criticised. I think there's also the kind of the thing that a lot of people forget of just how expensive and difficult it is to actually build out all of these different ad tools normally, like a brand advertising perspective. Like that part is probably the easiest part to start with in what started out as a small pilot, they're going to. This is still very trial and error. All these other things can come. But I mean, we've even seen, say if we take Snapchat for example, Snap hasn't. It took how long to build out all of these different things? It wasn't until more recently that we've seen a real emphasis on the direct response side of their business. And it's because they needed to have the infrastructure in place to be able to put that on their platform for advertisers to have access to. That took a lot more investment in terms of money and also a lot more time and having the right tools in place to actually put it all together. That's why with all of the other platforms we've pretty much seen them, they normally build out all of these different features and tools separately and then they group them all together at some stage and then you see this major thing like an Advantage plus or a PMAX or something or other like that. But OpenAI has almost done everything backwards. Could it be crazy? Potentially. But at the same time, is it necessarily as crazy when you think that they already have the playbook of knowing that it's these smaller advertisers that will get them the consistent revenue that they need to probably help them get in the black in terms of their bottom line of revenue. If they have that, then they can actually sort of have a bit more time and a bit more effort, I guess in other different projects, whatever they might throw, who knows, Sora may or may not come back, but at the same time, like they need to have that consistent revenue so they don't have so much pressure. Because if they are going to be going to going public soon, they're going to need to see investors are actually interested in it. Investors want to see that the company is able to turn a profit or at least has the potential to do that, and that they're not basically going to be putting in money, money, money, money the entire time and they're not just seeing any return whatsoever for the foreseeable.
- [22:38]
A
Which I guess is why OpenAI needs to roll out a Pixel. So because the thing it hasn't moved quickly enough on that front is its pricing model, it's still impression based. It needs to move to click or action. And if you're OpenAI, sure, you could report on clicks back to the advertiser and let the advertiser then connect those clicks or correlate those clicks with product sales or action on site. But then you're letting the advertiser do that math. I would imagine if you're open AI, you want to be able to see those numbers for yourself, to be able to make those claims yourself. The pixel is your way of peeking at what people are doing on advertiser sites.
- [23:21]
C
Yeah. And it definitely doesn't seem like they're waiting for anything whatsoever. Like they're not waiting for things to be quote unquote finished, they're just rolling them out as they have them. I would assume that whoever's in the pilot right now, even from the conversation I'm having with some people that are in the pilot, they're getting feedback on what is working and what they're missing and what they need immediately, that naturally as time goes on is going to change the order of their product roadmap and how they actually bring all of this together.
- [23:50]
B
Where do you think We. Because they're working backwards in the playbook, right? Which part of me is like, are they working backwards? Are they kind of writing their own playbook? But what comes next? If they're working backward, where do they go from here? They've done the pixels. What happens next, in your opinion?
- [24:06]
C
I would say, I mean, I definitely know that CPC and CPA is coming soon, that I've definitely seen in the actual Ads Manager. So, I mean, I think in that regard, it's a matter of time. Because if we think about the initial criticism it was all about, we can't prove that these ads are doing much well, and not only can we prove that they're not doing much, there's not a big enough scale of it out there for us to even judge if it's doing good by a small margin. So they've obviously started to improve that by adding more countries to it. So far, they are literally now adding more advertisers in there. So really, and truthfully, this is growing. They're being kind of careful and cautious about it. But also, we have Cannes coming up as well. So naturally, I mean, I would be very surprised if David Dugan isn't actually there. He was there last year with Meta. Now he'll be there again, just catching up with all his, I guess, industry pals, industry peers, and saying, like, hey, I've moved here, now you know where I am. Come over here. Because now the one thing they have with him on their side is automatic trust. They know people in the industry, know how he works. It will be far easier for them to close deals with him. And if they have an ads manager, kind of almost in a slightly more open beta, I guess by then I mean that automatically you'll have smaller advertisers coming in, whether they're asking about it or not, because they won't be reliant on having a human being to actually put the ads in place for them.
- [25:28]
A
What will be interesting is how much of his meetings at Cannes might be with big brands and agency holding companies, and how many might be with vendors to put together the infrastructure, as well as with folks at, whether it's TV networks, streaming services, other platforms to try to poach them, to come over to OpenAI.
- [25:50]
C
I do know from things that I've known, from things and conversations that I have had, that in terms of partnerships, they're looking at, they're very open to partnerships. In terms of, say, working with DSPs, it's not on their radar. It's very much that they are building it from scratch. They want to own the ad tech stack. That is plain and simple. But they're not averse to say, a critio people that can bring in the different swaths of advertisers from all over the place. There may be other types of partnerships, but I think going down the DSP route and just partnering to scale up, it doesn't seem like that's going to be the way that they do it. It's more just they want support, but they will own the whole landscape of what they're doing under their platform.
- [26:35]
A
Okay, so maybe an OpenAI DSP somewhere, because that's part of the playbook is Google's got DV360, Amazon's got its DSP.
- [26:45]
C
Who knows? Here comes the second half of this year and that'll keep me busy.
- [26:50]
A
Yeah, well, I'm sure we'll have you on before even that, probably in a few weeks, I'm sure. But Krystal, thanks so much for coming on the show.
- [26:58]
C
Always a pleasure.
- [27:07]
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 newsletters sent out each weekday morning. Visit digidaire.comnewsletters to sign up.