LLMs, media plans and creators: are brands walking into a trap?

Speaker 0 · 0:00

Hello. Welcome to the Media Unfiltered podcast. I'm delighted to be joined by Shane Skillen, who's the CEO of Hotspecs. He is an AI expert in the field of media advertising and branding. He works day in, day out with global brands. And today, he'll share experiences around the impact of LLMs on consumer buying, what value brand has in these environments. We also discussed the impact of AI in media planning, where it's heading, and exactly what agencies are doing today. And we finish off the discussion talking about the creator economy and why that is perhaps a major threat to the television industry. Shane, hello. Welcome.

Speaker 1 · 0:42

Thanks, Justin. Great to be here. Seeing you at the Pathfinders event in Toronto was was amazing. You on stage re... Just revealing the truths about what's going on with some of the platforms and pulling back the curtain, the the wizard of Oz metaphor where Google, Meta, and even Amazon are creating these profit machines for themselves and destroying brands is something I think you and I are both very focused on as, we're all headed for trouble. So I I run-in two agencies, Hotspax Insights and Hotspax Media, and we help Fortune 100 brands build singular brands, and that is brands that are consistent with a set target of emotions so that the work they do in advertising media, with their creators, their product, their packaging, their content, everything is aligned. And when you do that, you achieve consistency, which we call singularity. And we have, you know, big brands that we work with, like P and G are terrified of something called fragmentation, where Mhmm. The amount of things that get get created with AI across an organization is just staggering. And so how do you leverage emotion to make sure you're consistent across all of these touch points that consumers are exposed to thousands of times in any given week. And it's been a really interesting time because AI is just changing everything.

Speaker 0 · 2:10

Well, let's talk about what is actually changing and what it isn't changing. So firstly, let's talk about this brand issue. So will AI reduce the need for brands to have strong brands? Right? You know, do you believe that AI has fundamentally disrupted that?

Speaker 1 · 2:25

Well, I think that, Meta and Google would like that to happen so that they could be for all purchases that happen. Meta Meta just launched a new product that your listener speech said to no on October 28 unless they adjust the setting. Meta is just gonna pull actual product catalogs onto the platform to try to get a purchase with inside the meta platform and not sending that traffic off to a website where, people can have better exposure to brand. And they're gonna take credit for the sales there, thus boosting the ROAS, thus creating their flywheel of spend more with us. We will sell more for you. But, you know, brand is still out there in the world. People still have to actually go and buy a product. They have to know what that product means to them, how it makes them feel. And with with AI, that there's so much opportunity to destroy these brands because, you know, they they're just being pulled in all these different directions. And so one element is that, the platforms, Google and Meta, they've got their own internal AIs that are run to maximize Google and Meta profitability. And so I think that they would prefer that brands not be strong so that that our clients and all the brands in the world become ultra reliant on them driving all of their sales and commerciality.

Speaker 0 · 3:50

But you... You've got some research on this, right, on the consumer sort of discovery, using LLMs, that we do use it. But is that the final point that we... The final reference point that consumers actually make before deciding what to purchase? Yeah. It's a great question. So, more and more consumers are asking chat g p t, Gemini, flood.

Speaker 1 · 4:12

Yeah. We did some research in banking. You know, what's what's the best credit card for me? What's the best mortgage for me? And the answer engines give very good recommendations. They look like good recommendations. So remember how they come up with that recommendation. It's almost like a a zero moment of truth where they are scanning everything that's known about that brand and who's used that brand, the ratings on that brand, and then use that to make a personalized recommendation to the user, based on what they they... Those large language models believed to be true. And so we ran a simulation where, we we asked, 500 people how they felt about five banking brands, measured the emotional connection. And then we, simulated, ChatChiPT, which is the fastest growing, most popular LLM in the world, and I believe it might continue to be. Mhmm. We'll see what happens with Muse. It's growing very quickly as well too. So we simulated a, recommendation back to those consumers, and then we wanted to see would they override the first recommendation. Mhmm. Say you should go use RBC, for your credit card, or you should go use a lesser known bank in Canada called EQ Bank, that had weaker equity. And we found that strong brands, were over... Would... The... The stronger the brand was in the emotional connection, the more likely they would be to override it. In fact, 71% of consumers overrode their recommendation from the LLM if they had a strong emotional connection to that brand. So what that indicates is if if there's less brand strength, then the brand that people will not override what the LLM says and your... The brands will be coming at the mercy of the recommendations from these LLM platforms.

Speaker 0 · 6:15

So that's interesting because if you have a category with lots of weak unknown brands, people will pick whatever the, the, AI recommends. But what you're trying to say is to have a competitive edge in any category you're in, strong brands can can and will be overridden by by the consumers themselves.

Speaker 1 · 6:33

Correct. So... Yeah. Correct. And so I think I think brands are more important than ever. I think a brand becomes something that marketing can hang their hat on up to the CEO and CFO and say, so it's like, look, we are the unified theory of how we bring consistency and singularity consumers across all of our touch points Mhmm. So that we can create long term profits and pricing advantage, as the world changes in this agentic shopping environment that we're in and what... But what Meta really wants to happen is for people to do all their shopping on Muse. Like, if you go to Muse and download it and say, hey. I'm looking for, a new fan for my room or... Well, I'll use a different category. I'm looking for a new pair of shoes. If if if you don't say I want a pair of Nike shoes and the brand doesn't make it into that prompt, then, you're just gonna get whatever you get. And, you know, Gary V's been saying this for a while, ten years before the LMS got here because he predicted voice would be the new way that we transact where you're just asking Alexa or Siri to help you shop. And there are recommendations that come back, if if there's no strong brands in it, people are just gonna get whatever they get. Mhmm.

Speaker 0 · 7:49

Yeah. How how do you feel about the LMS LLMs, offering advertising or being like an ads business? It it... For me, I look at this and think, why does every canvas that has human attention need to turn into an ads business? It's it's just... For me, it it might muddy the waters, but what do you think?

Speaker 1 · 8:07

Yeah. It's a great question. I mean, the the joke I was just in San Cisco a couple weeks ago was that no matter what you do, you'll eventually be in advertising. Even Netflix. Yeah. Reed said we're never gonna do ads. And of course, they're they're running ads. There's a lot of money to be made in advertising as we see the growth of Meta and Google, because eventually you've gotta create commerce. So these LLMs have a voracious appetite for cash to build out their data centers, and to build out their compute and their inference. And so the the source of revenue to allow that is, is advertising. And I think if you look at the capital expectations from Google and meta, they've got these, you know, profit machines that that are advertising. And I think that they see that commercial opportunity and will do everything they can to capture as much of that as possible. You've got great stats too in terms of the growth of these platforms. It's just... It's it's incredible what they've been able to achieve.

Speaker 0 · 9:09

Yeah. We, yeah, we're just discussing this privately, and, looks like Meta, by the end of this year, will grow another 50,000,000,000, which is just staggering. The the top three now will account for an incredible share of media spend globally. One thing that, it just... When you look at search now, search continues to grow which is just staggering but, you know, when you look at the product, it's barely usable. So I'm super surprised that search is still continuing to climb. I think in Canada alone, it's grown about 10%. Does that surprise you? You know, with, like, again, the growth of AI platforms and the ability to actually have a conversation with a bot, and then search still growing at 10%. Does that does that shock you?

Speaker 1 · 9:54

It it doesn't shock me because that's Google's revenue model. So what what, you know, my friends at work at Google will say that your first day there, they they put out these philosophical questions and say, you know, we at Google could have, you know, set our company up to bring the world a balanced opinion on all topics and educate them holistically, or we could have just said make us as much money as possible. And Google, of course, said, let's make as much profit as possible. So Google's set up to make search work for them because they can go back to advertisers and say, look at the ROAS that we're creating for you, and their systems are designed to to almost go and find consumers as they're ready to buy a brand and get Mhmm. Get media in front of them and let's take credit for a purchase that might have might have already happened. And so I I expect that these platforms have more powerful AIs than you or I or anybody outside the organizations have been exposed to that are constantly looking for new ways to monetize cars. And, you know, Google Google search, some of that is in AI Gemini. Like, if you go and search for things in Google, you're getting Gemini answers. Yeah. And it's making product recommendations in there. Mhmm.

Speaker 0 · 11:12

Yeah. Yeah. I just wonder when, when all these platforms start offering advertising, whether that will muddy the waters, whether whether their independence will be compromised. Let's have a quick look at media planning because this is this is a big area, particularly for the traditional media law. How today, like, right now, how influential is AI in media planning? Or are we are we at just like the really early stages of it now?

Speaker 1 · 11:33

Yeah. I think it's it's it's growing. It's a great study from Futurian. By the way, after this podcast, I'll create a, companion website with some of the stuff and some tools for all your, listeners. It'll be at shaneskillin.com/justin. So I'll put that futuri setting here. One of the interesting points there was that 63% of media buyers are using AI to plan media, and that's only growing. And if you think about things like WPP Open and some of the other platforms that these big hold codes are using, they've... They're they're... What underlies them are all these foundational LLM models. And then what happens, when you when you ask it to build media plans, like, for example, we ran a a test saying, build me a media plan for QSR with a $10,000,000 media budget to maximize my ROAS and short term sales lift. Yeah. You you get a media mix, putting 57% of that into Google, Meta and Amazon. Mhmm. Right? So 57%. But then when you alter the prompt to say, and now do the same thing, but use proven econometric models from the IPA, and my marketing mix model if possible and other proven respected sources, the TV spend, the meta spend drops to 20% and TV and traditional, like, out of home goes up to, 38%. So what... What's happening is these models without propping them correctly are massively defaulting to these digital channels. And the reason why that is is because there's so much, in fact, a 100 times amount of marketing effectiveness data in the training that these models have learned on to say how effective digital is. And traditional stuff like TV, radio, and out of home just don't have those same datasets. And then even more interesting, because that was across five elements I've just talked about, When you ask flawed, without the comment or econometric model comment, Claude puts 21% of TV in plan. And Gemini, obviously owned by Google, only puts 6% of the TV in the plan and massively recommends YouTube, which is obviously owned by Google. And you just you you just wonder why that is. And, you know, these these models are not open. These are closed LLM models. And, you know, in others, in fact, if you go and ask these LLMs where a great pace... Place to start your career in AI is, Gemini will recommend, as, like, a geography. Gemini will recommend a location worksite DeepMind. Meta will recommend, an opportunity in San Francisco near their office. And even Grok will recommend if you go to Texas where XAI is. So these models can be tuned to show users what they wanna show you. And so it's, it's not too far fetched that there's a line of text sitting in the Gemini master LLM to say when any user anywhere asks about building a media plan, can you please massively over index a recommendation to YouTube and our other own properties?

Speaker 0 · 14:37

Yeah. Well, that doesn't surprise me, but I I guess the challenge here, like you said, is all the information that's available to these platforms, around digital, but not available from sort of the traditional media, which you were talking about their radio, out home, and television. So Correct. You can't, I... You can't rely on the planners to go, hey. Can you incorporate these studies? So what do what do... What do this... What does the sort of traditional media need to do, to to help sort of educate the AI planning tools?

Speaker 1 · 15:08

Yeah. I mean, I think if the traditional media could put more of this data in into the training material by publishing more studies so that they'd learn it. But mostly, I think brands, because they're by the media eventually need to ask the media agencies, how did you come up with this plan? Yeah. And how are you incorporating proven econometric modeling and even our marketing mix model, when designing this and and maybe even having brands sending the futuri study into their media buying group to say, is is this affecting your recommendations back to me?

Speaker 0 · 15:45

Yeah. Yeah. Okay. So Brian Lesser, who's the CEO of WPP Global, he said about a year ago, five years' time. So in four years' time, no human will touch a media plan. Just quickly, do you agree or disagree with that?

Speaker 1 · 16:04

I wonder. I mean, it's fantastic what's been happening in AI. I mean, it it... It's it's... So all of the hardest math problems in the world were to solve by Chachapiti in, like, a day last week. And so these are these are things that humans have been trying to solve for decades. The... There's, you know, new digital synthetic human beings where they're, using Gemini's, AI, Google's AI and behind it, which lives in something called isomorphic labs to cure all disease. And, you know, some of the people I know in San Francisco were saying, you know, we we might not age anymore because AI is helping us to figure out how our biology works. And so if it can figure that stuff out, you know, is media planning any more complicated than that? And it it, you know, the... If you think about what media is trying to do is trying to put the right ad in front of the right person at the right time at the lowest possible price to get a sale, the AI will look like assuming incredibly good job at that. And so I think AI is gonna always... You know what happens? It looks amazing. It sounds amazing. I think you still need human beings in the mix to make sure that it could be trusted because all of these models, as I was saying before about Gemini, are gonna have bias in them. And Meta and Google, they don't care about brand building. They care about profit maximization. In fact, weaker brands are are beneficial to them. And, you know, these agency networks, a a few of my mates are running some of the biggest ones. They've got pressure on cost pressure to prove value. And so these platforms will show up like AI platforms that look like they're doing a great job and they'll show internal efficiency at these agency groups. And so I think there will be a push and motivation to that to say, you know, we we should be using AI. But if we're using AI, then what's the point of the media agency? Because men and Google have their own AI. And I I think men at Google would love nothing more than to be able to go brand direct to just say, give us your website, give us your goals. We'll take care of the rest. And this is this is what Mark Zuckerberg said, you know, a few years ago. Just just give me your products, and your bank account and I will fill it up for you. But I think the human in the loop needs to be in there to audit this and see what's really happening. And we don't fall into this trap of AI psychosis where we just rely on these models to be saying what's best for brands.

Speaker 0 · 18:43

Yeah. No. It's it's a good point. And, the... You know, it's sort of already started to happen. You know, you need to have a massive account with Meta to have actually a human look after you as... Just as customer service. Most of them are being serviced by AI. Correct. The the black boxing, if you like, within Meta and Google has increased because these AI tools are super sophisticated. And, that's Performance Max. I can't remember what the meta one's called. Meta Advantage. It's called Meta Advantage. That's it. Advantage Plus, I think. Isn't it? Advantage Plus. That's exactly right. Advantage Plus. And, yeah. And that's how they've grown their business, to be honest. So let's talk... You do a lot of work also in the creator economy, which is also being impact by by AI. And also, we've got creators going rogue. So there's a lot of danger to to to brands as well. I I don't know if you saw... You see the Callaway example? Did you see that one? Yeah. The Callaway example. Yeah. There's things like that. There's things like... But, like, I mean, I don't... Those things are are isolated

Speaker 1 · 19:39

and Yeah. But they do happen. I think, you know, Unilever has famously said, you know, we're gonna we're gonna take 10,000,000,000 as a media spend, and we're gonna have half of that go towards our creators. And then we're now working with, you know, instead of 30,000 creators, they've got 300,000 creators they're working with. And what they're doing is they're chasing attention on these platforms. So if you wanna think about, you know, attention is everything. That's that's how advertising works. And you've gotta find creators that are making interesting, new, fresh content so that people watch it. And and then you can actually create through that. And, you know, what what we're seeing across all of our clients now is just to focus on, okay, who are the right creators for us? How do we brief them properly? And they use our hot specs maps. So so what are the... What's the authentic emotions that they bring to life across all of their content? And then what are our emotions we wanna bring in? So where's the synergy between their brand and our brand? And then that brief creates, the means for them to crave an effective video and certain things need to happen in there. You've gotta use the distinct brand assets in appropriate way in the first few seconds. The content has to have a certain amount of attention throughout it, And they need to say the product reason to believe in the exact same way after they say the product name. Axe deodorant is the best smelling, fragrance for teen boys to make their first move with girls, for example. And that's not actually what they do, but all brands have specific reasons to believe. If they don't do that effectively, as all this content gets out there, it's gonna create fragmentation because muse, the meta meta AI assistant is gonna be training on this content, and Gemini is gonna be training on this content. And if 300,000 creators are out there, say you've got a a 100 on a brand, but all saying different reasons to believe. If you go ask what's a good deodorant for me to buy my next shopping trip, or Muse fill my shopping basket on Instacart, and it's confused about what Axe does and what what its product benefits are because it's all over the the map, so to speak. It won't make that recommendation. So you've gotta be ultra consistent with that. And then the next trick is you you audit the creator content for those things I was talking about earlier, and then you bonus them based on their ability to adhere to getting attention using the strict brand assets, and then using the reason to believe. And then you figure out based on that how much paid media to put in behind that content. And if if brands do that, I think that's gonna be the brand building platform of the future because all... As you've said, all this money is flowing into Google and Meta. And it's it's the creator content that, you know, we hear all our clients are ultra focused on because that is what's gonna keep his attention right now.

Speaker 0 · 22:30

Well, it isn't it isn't. I would argue with that because attention is actually driven by the platform and media type. It's not by, by what you deliver. Obviously, what you deliver can help, but ultimately, they're they're low attention environments. I don't know if you've seen... I don't know if you're on TikTok or on these platforms, but soon as an ad comes up or a creator ad, they can be flicked away within a nanosecond. So it's... They are low attention environments and making them high attention will be really, really challenging. So that... That's one of the problems. And then the other problem is, as we invest heavily in this, the the platforms are done, like, the the the the creator environments are dominated by a few platforms, and then the brands themselves become very reliant on the algorithms. Right? It's it's a it's a it's a vicious circle, isn't it? So do you think, you know, obviously when you come back from Cannes and stuff, it's the... All the talk is creator and stuff. Do you think as a as a industry, we've gone too heavy into the creator world?

Speaker 1 · 23:26

I do. I think I think I think it's a little bit too heavy. I'm not sure what the exact optimal mix is. And I think if you think about, like, a TV ad and the effect of it as a TV and it's, you know, it's been proven to be extremely effective in terms of driving short term sales lifts and long term brand building. Yeah. And it's it's like a shared viewing experience, which which does get better attention. In many cases, you're watching it, on the big the big screen at home. Mhmm. That that almost could be the North Star for a brand because that... Everybody will see that. And then I I think brands do have an opportunity to go and find different creators that can, be additive to that. I call it a tent pole, marketing and advertising event, which is which is the traditional TV buy. That's it. Some people you just can't reach on TV. And so Yep. I think creators are are gonna continue to grow because it's in the platform's interest to push them. And I agree you can't flick creator content away, but it... They have more engagement than than, say, a traditional ad where it's obviously just an ad, particularly if the people have seen that creator before. And interestingly, what we're seeing in our clients' data, and and I've heard this from some of the platforms too, is that your number of followers doesn't really matter anymore, because the AI inside Google meta and TikTok will, scan content and figure out what's the most relevant audience for this content that's just been created. And how can I link back a commercial signal to showing this to people? And so creators you've never seen before, you might notice when you're on these platforms, you'll start getting their stuff because the algorithm algorithm has decided that you you should be seeing that right now because that content fits with what you're interested in seeing and even even more so if you're about to buy something in that category, you you're because you've shown signals other where you're just gonna be seeing more and more of that content. So I I expect creators to continue to grow. I think it's up about 28% globally. It's still a very, very small share of total spend, but Yeah. We we just... We are all across our clients. We're gonna be using more creators. We think this is this is how we're optimally able to get attention, and we can put the faith into the platforms that they're gonna get this stuff in front of the right people at the right time, and we can create media attribution back to that content.

Speaker 0 · 25:56

Yeah. But we're we're we're we're, again, gonna be chasing the sort of five to 50... Five to 25% of of those in market when we start doing that sort of heavy personalization. I understand the creative stuff. I understand going after younger audiences as well, different audiences too. But, I I also think... It feels very, like, early programmatic as well when you talk about the long tail of creators. You know, there's agencies out there who, like, look after tens of thousands of these people who have small audiences relatively. They're they're taking they're taking 70, 80%, like, of the, of the ad buy from these creators. It's crazy the business model's going out there. But very quick question, just to sort of yes, no. Do you think... Because, you know, if you look at Unilever as a good example, they've switched from being a broadcast first business to a social, but, first business, you know, money is coming out of of television to go to social. Do you think creators will be the death of TV advertising?

Speaker 1 · 26:53

Well, you just, you just wonder about what... What's the death of TV? And because, you know, is it dying? It's... So a lot of people watch TV. I think a lot of media buyers don't realize that. I'll put this, this is a great site that I'll put in the in the link that I'll that you can your get your guest can visit after this as well, too, is that the people buying the media are much less likely to consume TV. And they're much more likely to be on TikTok, Meta, and Google. And so they think that's where the audience is, but a lot of people still watch TV. It's it's it's obviously declining. I mean, we we all know that. But I do I do think it has a role. It's it's just hard to create attribution back to it. And it, you know, it'd be great if if you could get these real time reports after you ran a TV ad of what exactly happened to my short term sales life and what was the impact on my brand building? Because MetaGoogle, Amazon, and TikTok can show you that these these amazing reports that actually map back commerce, which marketers like to see CFOs like to see CEOs like to see and the media, agencies. This is this is their directive. Get get us cheap ads that converts, as as the brief into them almost universally. And so will creators, kill TV? I think TV is, has got a fight here, but I do think that more and more people are gonna gonna find creators that they said they like to watch. And I I think it's gonna I think it's gonna put a big dent in it, and I think that's gonna be the new big, most prevalent platform that brands are gonna focus on. And so, yeah, I actually I actually do think it's gonna make a big dent in TV advertising.

Speaker 0 · 28:35

Okay. So so just to wrap this up, so you got creators tied to retail. You've got the, AI platforms, LLMs making decisions for for for, consumers. Do you think brand in five years' time will be less important today or more important?

Speaker 1 · 28:54

I think brand in five years' time is gonna be more important, but but but in a different way, I think that now because the... These these companies are armed with all of this ability to create, like, you've got, you know, junior marketing managers that can go and make content, make ads, make make media plans, make strategies. What is gonna be that unified theory that links all those people working together? So that when people actually go to buy the product, like a can of axi odor, they've actually gotta hold this thing. And so how do you make sure that the content, the ads, the media, the shelf displays, the PR is all aligned so that when they get to that moment of purchase and they get to that that that usage experience, that there's singularity across all of that. And so I think it's a fantastic time to be in marketing because this fragmentation is just just... It's just exploding with the amount of stuff that's getting made. And so I think you can think about brand as as advertising, but, you know, Mark Ritzenol was... Likes to say that, you know, what what marketing and brand about is, you know, product price, promotion, and placement. And it... It's almost like marketers have just become advertisers. And I think there's a a a genuine opportunity, and we've seen this across a bunch of our clients, where marketing becomes the North Star of organizing the company so that they can go to market and go to their current customers and new potential customers in a unified way.

Speaker 0 · 30:26

Mhmm. Alright. Well, it's been it's been fascinating to to hear from a proper

Speaker 1 · 30:30

AI expert working brands, on a daily basis on this. Shane, I really appreciate your time. Thank you very much. Thanks, Justin. And just one point, so if you go to Shane Skillon dot com slash Justin, I'll put some resources in here for your listeners. And to everybody out there listening to this, if you can ever get Justin on a call to talk about his insight into the future of media and how it all really works and you wanna know the real truth in an unfiltered way,

Speaker 0 · 30:55

I... I've just never seen anyone on Earth be able to do that better than you, Justin. So it's great to know you, and thank you for this chat. Cheers, Shane. Flattery will get you everywhere. What I'm gonna do is I'll do an article for all the listeners with that link and, share the charts that you're you're gonna provide. That's amazing. Thank you, Shane. And thank you everyone for listening.