Loading summary
Transcript34 lines
- [00:00]
A
Foreign.
- [00:10]
B
Hello. Hello and welcome to another episode of the Digiday Podcast, a show for anyone wondering why they're seeing more ads in ChatGPT. I'm Kamiko McCoy, senior marketing reporter here at Digiday.
- [00:20]
A
And I'm Tim Peterson, executive editor of video and Audio Digiday Media. Kamika, are you seeing more ads in ChatGPT?
- [00:27]
B
I am. I feel like after every handful of questions, ChatGPT is like, oh, by the way, here is an ad. What about you?
- [00:36]
A
I have not. I've also, like, yesterday was the first time I've used ChatGPT since the Pentagon deal was announced or they announced they were doing the Pentagon. That's when I was just like, nah, you're Team Claude. No, I was veering towards Claude anyways because I had gotten like, really into Claude code late last year after doing explainer videos on, like, AI agents and model context protocols. Call. But yesterday I was just like, I knew we were going to have our senior platforms reporter, Crystal Scanlan on the show for the monthly check in on ChatGPT. Welcome, Crystal.
- [01:10]
C
Always great to be back. Now I feel like I'm part of the furniture.
- [01:13]
A
Yeah. So having read your prolific reporting on how ChatGPT's advertising business has progressed, I was just like, all right, it sounds like more ads are starting to appear on ChatGPT. Let me get on there and see if I can trigger that. And so I tried to be like extremely commercial with my query. I'm currently in the market for a new pair of sunglasses. And so I was just like, this seems like a perfect thing for ChatGPT to be like, here are some sunglasses and here are the ones that this brand really wants you to buy. And is the sponsored placement got none of that. I didn't even get any, like, ancillary ads of, like, well, if you're in the market for sunglasses, how would you like some sunscreen or, you know, what have you? I got nothing. Which Kimiko was kind of a theme that came up when we were at the Programmatic Marketing Summit earlier in May where we had our town hall session where we have brands and agency execs. It's under Chatham House rule, so everything everyone says in there is kept anonymous. But we are able to report on what's said and what a number of people were saying in the room is, hey, We've been running ChatGPT ads. Well, we've been looking to spend money on ChatGPT ads. The ads aren't running. Crystal, you and Sam Bradley, fellow reporter, put out a piece within the past week or so reporting on these under delivery issues. What's happening? ChatGPT doesn't have inventory because I feel like the last time we had you on the show, we were talking about how ChatGPT is doing so much and going so fast in building up this ads business. But it seems like the stopping point had been short of actually showing people ads.
- [03:02]
C
Yeah. So there's definitely been some real under delivery issues. They definitely seem to be ironing out a little bit more now. But I think also we're over three months into the pilot, but especially for advertisers that were in the earlier stages, I mean, there were some that were literally just, I think not even getting a tenth of their 250,000 commitment even used at that point, which, understandably, if that money is just kind of tied up and you can't really move it or do anything with it, it's quite frustrating in that sense. From the OpenAI perspective, it's very much from their side. We're just being very cautious, we're being very careful. We're not trying to damage the user experience. So hence they are. It's kind of almost built into this is how they want it to be, because they want to get ads. Right. Albeit probably not ideal for an advertiser wanting those ads to be out there and to get the reach from those ads. But yeah, I guess, I mean, there was even one person that I did speak to, they were literally telling me that they've got blue chip clients, they had that experience. It was so incredibly frustrating for them, especially with the basics. I mean, the very basic reporting that they had at the time because of the kind of negative experience that they felt that they had from that they were. To be honest, it's not kind of, I guess it's left a bit of a sour taste in their mouths that they want to kind of hold off again if they can, and, and wait until OpenAI has kind of better partners, or not even better, just partners that they know and already trust to bridge that gap so that then things feel more like they're running a lot smoother as opposed to trying to go direct and then have all of these kind of bugs, which, as you'd expect with something so new that they're practically building it in real time so that they can try and remove those kind of pressures and issues. I guess.
- [04:45]
B
I think that's so interesting and I wonder how much of that is like part of the course. Right. Because when I talk to advertisers and folks who were joining on for the pilot. So, like, we had the not. I talked to, like, Bear Paint and a handful of others, dsw and I was like, as you guys are testing this, like, what are the expectations? And everybody kind of gave a shoulder shrug. And this, the point of it was just kind of like, well, this is a new ad opportunity. So many people are using this. I think at one point ChatGPT had said that it had 900 million users and there was just like this massive opportunity to get there. So to see it unfolding. I am curious how much patience media buyers and advertisers seem to have here. Because you join for fomo and now it's just like, like, does that fear of missing out still exist? Krystal, what are you kind of hearing from your perspective?
- [05:30]
C
I think I'm definitely hearing similar to you in some respects. I think I've definitely heard from one bucket. It's very much a case of, well, it's kind of baked in. You kind of knew it was going to be a bit of a slow process. While they iron out all the little, like, snags and everything within the pilot, others are definitely have been telling me that they want to hold off, or at least they did want to hold off, until they started seeing things like more performance advertising and like CPAs. Because from their perspective, if they've got clients that aren't like, massive global brands, they need their dollars to work harder and they've got a lot less dollars to work with. So I kind of think it's been definitely in those two buckets that I've seen most of the people that I've spoken to with.
- [06:07]
A
Yeah, and it seemed like a lot of agencies were kind of advising clients, like, look, if you want to be part of this pilot, cool, but let's manage expectations. Don't expect to get much out of it. Even OpenAI kind of indirectly seemed to indicate that by being so limited with what the ad offering was and what, you know, the measurements that are going to be available, all of that. It seems like the frustration, though, from the buyers, brands and agencies has been, wait, why are you asking me to commit $250,000 and you're not going to run these ads? Like, it'd be a different thing if they were running the ads and the ads weren't doing well, but they're not even running the ads. Krystal, have you gotten a sense from, like, the OpenAI side why they would be lining up all these advertisers? Even, like, in this process, they've added critio and Other, you know, I think Stack Adapt has been one of the more recent ones. Like they're making it so that more advertisers can access this inventory, but isn't really much inventory for them to access.
- [07:14]
C
Yeah, it's definitely a tricky one. I know when I've sat in on a roundtable with OpenAI and even was putting questions to them about that, the very clear message that they say throughout the entire thing, any conversation I've had with either them or even people that have had conversations with OpenAI, it's very clearly this is a test. We are taking our time with it, whether you like it or not, essentially, because we have to be careful with how we do this. This is like a very uncharted territory. We need to make sure that we get it right because if we don't, we are going to screw up the user experience and we cannot afford to ultimately lose the users. And I guess in terms of when it comes to the actual commitments they were asking for, they asked it from that. I mean, some people I've spoken to did kind of give the impression that they felt it was just a bit of a give people FOMO essentially, like they want to be in on the action, whilst not a lot of people have the opportunity to. But even what that was, in February they were asking for 250,000. Then it reduced once we started seeing what about a month later to 50,000. Since they've opened out the ads manager in the US widely, they've scrapped it all together. So now you've got partners also scrambling to work out well if they don't even have a minimum commitment. And that was kind of potentially what we were offering as a good way to kind of join the pilot. What can we do and how can we become competitive so that people still want to actually do the ads but use us to do them versus just going direct.
- [08:40]
B
I wonder how limited. Well, how tied their hands are here. Right, because to the point about not pissing off your user base, because to one, their user base has been a little bit finicky. Because even to your point earlier, Tim, about the. The deal with the Pentagon and they kind of took a hit and became lost their spot within the App Store and things like this. Those are all things to like really, really think about. I'm curious, how does this compare, and I know we talked about this before, but how does this compare to how Netflix played out their playbook? Did they also have kind of like spin commitments they had to roll back? Tim, what did that look like from your reporting Just to kind of compare
- [09:13]
A
and yeah, so Netflix took a very different act than OpenAI has taken here. So when Netflix first rolled out, its ad supported tier lined up, advertisers got them on board. And this was in Q4 where that money is very precious because it's not only oh cool, we get to be on Netflix, get to be 1 on the first on Netflix, but also this is the big quarter. This is the quarter we need to be driving sales or whatever other business results. And so it was like by early December of that year when there were advertisers who were just seeing under delivery issues, like they, you know, Netflix just wasn't filling the spots, didn't have the inventory for them. And Netflix went to them and said, hey, look, we don't have enough inventory to fill your demand. So if you want to take your money back, that's totally cool, take your money back. If you want to hold it here, we'll try to fill it or like we'll move it to Q1 or what have you, but basically like, let us know what you want to to do with this money. We don't need to hang on to it because we understand this money means a lot to you. And Netflix got a lot of goodwill from doing that, from like being willing to work with advertisers. There were some who took their money back. There were others who were just like, nah, let's just, let's move it to Q1, like let's stick here. But the gesture really held for Netflix. So that, I mean, through the next year, Netflix was still having inventory challenges, supply challenges with advertisers, but brands and agencies were willing to stick with Netflix because Netflix was willing to work with them. And OpenAI seems, I haven't gotten the sense that like OpenAI has ill intent here. It just seems unaware of, oh, some of these brands, that $250,000 is meaningful money. And they were very willing to spend that money on ChatGPT and whatever the results would be were what they were, but with the like very basic stipulation of I would like this money to be spent and ads to run because I have a brand to build or sales to products to sell, things of that nature. And Crystal, it just sounds like OpenAI wasn't as aware of that as Netflix may have been.
- [11:33]
C
I do. Part of me does kind of question whether or not could that have even been because Dave Dugan wasn't actually there at that point. So he's obviously knows what he's doing in the ad business. He knows how people operate and what they want to do with different budgets and things. So I don't know if the former
- [11:47]
A
Meta executive, Meta ad sales executive who OpenAI hired in, he started in what, April as the head of ad sales at OpenAI.
- [11:55]
C
Yeah, literally. So I mean, what we're now what, a month in really to his role. So he's super, super early, but obviously he's a Meta exec. He knows exactly what he's doing. All of the commitments were put in place back in February, like when it was still very much. They were trying to get someone to be head of that team and build out the team as they were doing it as well. So it was almost, I guess, like really building from the ground up with people, not that they don't know what they're doing, but not people that are as seasoned as him in an advertising industry.
- [12:22]
A
Although Meta has a different track record when it comes to a lot of these tech platforms have different track records when it comes to issues with advertiser spending and they handle it. Or like there was the, what two years ago or so there was, I think it was like around Easter, an issue with the Facebook ads manager where a lot of money was being spent like past budgets and, and Facebook or Meta was basically like too bad or like the measurement issues that it had, you know, so many measurement issues last decade and it wasn't willing to like refund advertisers for a lot of those things. Whereas like Netflix at the time it had Jeremy Gorman who had come from Snap and from Amazon, but it also had Peter Naylor who had come from Hulu and NBCUniversal and so especially had the pedigree of traditional TV advertising where there is more relationship building and more of kind of the like one to one, a hand holding and white glove service that I think the tech platforms don't really come to. But so under delivery has been a criticism of OpenAI also just OpenAI not having performance minded advertising capabilities, both from a, you know, initially you could only buy based on impressions, you couldn't buy based on clicks or based on actions. The kinds of things that a lot of digital advertisers, especially search advertisers, come to expect. It didn't have conversion measurements and it also was a fairly limited ad product anyways. Yeah, that all seems to have shifted. Like, you know, Kimiko, when we were in the room with the brands and agencies at Programmatic Marketing Summit, one thing some of them were saying is like, yeah, we saw these under delivery issues in that first phase, but. And this was early May when we had this conversation. They were saying like within the past couple weeks at that point. So like sometime around mid April things started to shift. Crystal, have things started to shift?
- [14:32]
C
Yeah, and I also, I do think that that was kind of also around the same time that they started showing ads to logged out users as well. So that kind of probably would have bumped it up a lot quicker. Anyone that I've spoken to recently has definitely said like there was definitely that period where it was just like, what is going on? But things definitely seem to have ironed out and they are improving. But it's almost like people will say, yes, it's improving, but with the caveat this is still very new
- [14:57]
A
and there's a lot of very new things as part of OpenAI's ad product portfolio at this point.
- [15:04]
C
I mean it literally feels like it's almost a new thing every day. With how many stories even I've had. What have we had since last week we had CPAs, added the self serve Ads manager that's now widely available in the us. They've got a pixel obviously up and running. There was a bit of code in the pixel that I had verified which kind of gave me the indication that they'd be starting to move into Europe because we need that for GDPR over here, unsurprisingly. Then a few days later they announced officially that they are expanding their pilot to the uk, Japan, Mexico, Brazil and South Korea. Reporting is slowly improving, albeit, I mean it's still very, very like, it's like night and day if you're going to compare it to like a Google or Meta situation. But yeah, but the surprising thing, as much as they're moving very quickly, they are still only asking for test budgets. They're not asking even when they started having CPCS in, I did start asking questions like, are you now starting to see the pitch change to say that they want to start getting search budgets even? But OpenAI is very, very determined right now and the only thing they are saying consistently is this is still a very new product, this is still test. So make sure you just take it from a test budget just to get there and then we'll figure out the rest. So even in that kind of language you can already see it's still, it's not a fully fledged thing by any means. Yes, they've definitely built this entire platform and all the different features far quicker than even. I think the last one that I heard that was quick within my time at Digiday was TikTok and that took what, four years. So this has been within like two months. So in that sense it's insane. But also is it surprising when they are kind of hoping and expecting to go public before the end of the year? That could also be an indication as well why they wanted to lock down certain budgets. Because if you're going to have to have forecasts and targets about what you're going to be able to achieve revenue wise, you need something for investors to be interested in or else it's just not going to go go well.
- [17:07]
B
Yeah, they've also got their hands in a lot of pots right now. In addition to preparing to go public, they've also kind of got, not in comparison, more so in competition with anthropic about this push for like enterprise level partnerships and things like that, licensing deals and whatnot that they're working through. So I would imagine that a humble approach probably makes more sense here. I'm also curious because it seems like a chicken and the egg situation in terms of like the playbook of building out an ad. Right. Do you put the structure in place then kind of hire the humans to kind of help mandate it, similar to like bringing Dave Dugan in and things like that to kind of steer the ship. So it seems like they're navigating a lot right now. And you know, we'll see how it all pans out because they've also got some competition stemming from Google.
- [17:51]
C
They certainly, certainly do. And also I guess as well, we've all seen what happened with Perplexity last year. I mean like their ads business I think lasted for what, less than 12 months. So it's almost not that they'd be copying Perplexity. It's obviously a very different and a lot smaller as a platform. But if you're going to have that, you've already seen the learnings of what not to do in that scenario as well.
- [18:11]
A
Yeah. Although it's still interesting to me that, I mean again, in that story that you and Sam Bradley published about a week ago, you mentioned in that story how some ad buyers are seeing higher fill rates from OpenAI, like the under delivery issues getting kind of addressed. But even I think there was one person in that story who mentioned they're seeing fill rates improve to 30 to 50%, which still means half to more than half of the available ad slots are going unfilled. And so it's just I'm really fascinated at this approach OpenAI is taking where they have this inventory to fill, but they're still not filling it as much as they can And I'm sure ChatGPT users may like that. Although Krystal, you've also reported on how ChatGPT users seem okay with the ads. It doesn't like there hasn't been a blowback.
- [19:07]
C
Weirdly enough, the main kind of blowback that they really had was around the Pentagon deal and that's when there was a massive kind of, I guess, user dropped. Users moved straight from chatgpt over to anthropics. Claude, unsurprisingly from that scenario, but it seems almost as if the longer that these ads are kind of in place, people are just becoming a bit more accepting of it because I guess at this point they've, chances are they're probably on meta platforms. They're very well versed and well used to matter having ads, probably every other platform out there that they use. I think maybe it was just probably the, the perspective of the fact that if you've had a product that was very as all the platforms do at the very early stages, we're not going to do ads, we're going to do everything right. We're going to do it right by the user. Then to completely change that stance, it's going to get people's backs up instantly. But they're being really considered and that's probably why people are being a lot more receptive to it because they aren't just completely overwhelming their feeds with ads whatsoever. It's to the point now people are still even like Tim, like you mentioned, you're still waiting to see them.
- [20:10]
A
Yeah, and then I go back to cloud where I don't have to worry about ads whatsoever. But it's interesting because on the one hand, OpenAI is trying to be very considered and controlled and conservative when it comes to showing ads to people. On the other hand, it's still trying to secure ad budgets. It's still rolling out all of these ad products or options of you can now buy ads on a cost per click model, a cost per action model. You can add product feeds, attach those to the ad so they're more informed from a shopping perspective. You know, today, as we're recording this on Friday, May 29th, our executive editor of New Seb, Joseph reported on a deal that OpenAI has with the commerce platform sky that seems to bring in new demand from search marketers who, according to seb's story, quote, search marketers running campaigns across Amazon, Walmart and Google through sky, may never go near a programmatic platform. So it's like another new advertiser base that OpenAI is tapping into, but if it's not running enough of these ads, it's not. If it's not giving enough advertisers enough proof points, I wonder how secure any of that money is going to be. If you know, the general experience from advertisers with OpenAI is I can't really run there because as Kimiko is alluding to, Google is opening up new ad slots when it comes to AI mode, AI overviews and search, which is going over this huge overhaul. Kamika, you were following all the Google I O updates a couple weeks ago where May announced all these big search changes that you know, people were saying this is the biggest change to Google search in 25 years. It rolled out new ad products. What are these new ad products that Google is bringing to the market?
- [21:59]
B
Yeah, they've made some pretty big, big changes, big announcements here. So outside of the universal cart and things like that, that's like kind of upending the shopping experience. On the ad side they put some new modes do ad opportunities in AI mode, including like what they call conversational discovery ads which essentially work by like you are having your ad answer a person's specific question. So they've got on their ads commerce blog, for example, someone trying to say I want to make my house smell like those fancy spas or a rainforest. What are some low maintenance way to make my house smell? Amazing. Amazing. Excuse me. And then the Gemini model is built to kind of create tail creative tailored to that search. You've also got highlighted answers so I won't bore you with the details. But long story short, they've made this a lot more conversational specific. So when someone is typing in, you know, whatever their long wind questions are into Gemini, into AI mode, excuse me, that there's a ad opportunity there to be more than just kind of this stagnant placement that you see on ChatGPT. Google seems to be pushing for. Not seems to be, they are because they announced it pushing for more of a responsive, dynamic, interactive ad platform. So you know, in addition to having the ad space, they've also got some different formats that are coming forward.
- [23:18]
A
And so Krystal, I guess brings us back to that question we went through with you last time we had you on the show of like, is OpenAI moving too fast with its ads business or is it not moving fast enough? Because as you've been reporting, like it's built so much of the ad product and the ad infrastructure at this point. But if it's still not running enough ads and Google is introducing More ad opportunities for advertisers. At some point, is OpenAI still not moving quickly enough to capture the ad budgets or is Google moving so fast that it's going to help itself hold on to those ad dollars?
- [23:58]
C
I mean, when you think, I guess it wasn't that long ago that even Sam Altman was calling a code red internally. I think it was the information that had it and that was because the whole everything to do with Google, they saw it as a genuine threat to ChatGPT and they wanted to put a lot more of their efforts into ChatGPT so they didn't fall behind too much. I think with this entire space, everything is just moving too fast in a way just because it's hard to not only keep up, but probably keep up and keep the quality of what you're doing. With OpenAI, I think it's a double edged sword in the sense that yes, they're moving very quickly, like two months and they've got what they have already is insane. At the same time, they're not actually a mature platform by any means. They are so, so young still. They have a lot of growth room. Whether or not they can still keep that edge is still tbd, just given the nature of the space. But the fact that everyone, even over here in the uk, there is so much demand and so much interest in ads in ChatGPT at this point it's, I see a lot more excitement and buzz around that than I do around necessarily Google. But maybe that's also just because this is a very new type of advertising. This is a brand new like, you know, well, not brand new tech company but a brand new offering that they are doing versus Google, which is very tried and tested. We know what Google does, but it's just giving us more options at this point. So at this point I wouldn't be surprised if you see advertisers just to an extent sticking with Google because they know they're going to get their ROI from what they actually get. However, no one, from what I'm speaking to, no one will hold back on the OpenAI because they're just like, well, this is the future, this is the way ads are going. We can't miss out even if we're a bit frustrated. It's almost like the same attitude that people have towards the bugs in Meta's Advantage plus that it's become such a part and parcel with actually being on that platform. The amount of conversations I've had over the last, what, three or four years when I've said to people, are you just not kind of frustrated by it? And their answer is literally, even to this day, yeah, you just get on with it. It's just part and parcel. You just accept that there's probably an engineer doing something, tweak something, and it's just screwed everything up on our end. But it will fix itself at some point. But because Meta has always proved to give them their return on investment is why they still stick around.
- [26:16]
A
Yeah. So I guess the tension is like, okay, great, you have people who are excited to get their hands on with a thing, but how long will that excitement hold if they have to keep waiting to really get hands on or when they get hands on, to have the ads run and have that experience? On the other hand, you want to have an ad product that when advertisers get their hands on to it, they're happy with it, that it's delivering on what it is that they wanted. So I'm very jealous of the two of you because later in June you will both be at Cannes in south of France and have the opportunity to hear from all sides in terms of how are things really going with OpenAI and how are things going with Google, especially because God knows, Crystal, what you'll be reporting in terms of what OpenAI is going to be rolling out between now and then anyway.
- [27:06]
C
I mean, at this point, who knows? I know in terms of their presence there, they're not obviously having a massive beach spot by any means. They are obviously planning to do a lot. Definitely a of closed door stuff. I know that, but I think it's going to. It sounds like Dave is going to have his week cut out for him with lots of meetings here, there and everywhere, all the advertisers and I guess he's been in the industry enough meeting up with some old peers, old pals, and seeing if he can get some more ad dollars.
- [27:32]
A
Awesome. Crystal, thanks so much for joining Kona.
- [27:36]
C
Thanks for having me.
- [27:45]
A
Thanks for listening to this episode of the JJ 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 digiday.comnewsletters to sign up.