Latest / Shiny New Object - a Marketing Podcast / Stop just accepting what AI gives you - Havas Media's Des Bateman
Transcript
- 0:00Tom Ollerton: Hello and welcome to the Shiny New Object Podcast.
- 0:02My name is Tom Ollerton. I'm the founder of Automated Creative,
- 0:05the creative effectiveness adtech platform. This is a
- 0:08weekly podcast about the future of data-driven marketing. Every
- 0:12week or so, I have the pleasure and the privilege of
- 0:14interviewing one of our industry's leaders about where
- 0:16they think this industry is going and this week is no
- 0:18different. I'm on a call with the lovely Des Bateman, who is
- 0:23SVP of Growth Havas Media Network. Now, we've never had
- 0:26this before. We never had a sales guy. We’ve never had a new
- 0:28biz guy on the show. Like in 300 odd episodes... apart from me.
- 0:31But when I first met Des, he's worked at Lego in like a
- 0:34marketing function, which he can't tell you about. But
- 0:36anyway, Des, I’ve introduced you in quite an odd way. Could you
- 0:39do a better version?
- 0:40Des Bateman: Yes. And thanks for having me on the show. And I
- 0:43know it's been, been a while, sort of to get this conversation
- 0:45happen. Yes, so essentially, I've been into various different
- 0:48roles throughout my career, largely within the media
- 0:50agencies. I've been a client twice. I've been head of
- 0:54strategy in various different roles for the likes of
- 0:57Volkswagen, but also Sparks Foundry over at Publicis. I did
- 1:00Lego for a year. Most NDAs I've ever signed. which is great
- 1:04because some of that's about to come out which I'm quite excited
- 1:06about and I'm currently two weeks into a role as SVP Growth,
- 1:09Havas. So that's about going and getting new business on behalf
- 1:12of the media agencies to, you know, to bring in new clients,
- 1:15new revenue. So I'm surprised you haven't had, anyone else on
- 1:19that's been as a salesperson.
- 1:21Tom Ollerton: No, senior marketing people. That's why I
- 1:22wrote to you, and now you're not. You're a senior
- 1:24salesperson, so we're here now, aren't we? Do you know what I
- 1:27mean? But with that, with that background, I have very high
- 1:30expectations. So lots different roles, brand side, agency side,
- 1:35all over the shop. Are you a marketing book person? And if
- 1:39so, do you have a book that you talk about a lot or recommend or
- 1:42buy for people?
- 1:43Des Bateman: Yes, I think so. It's one of those things... I
- 1:44think there are so many books and recommendations as to as to
- 1:48what you should read and very much, that's almost like, you
- 1:51know, you have to read, you know, Byron Sharp's How Brands
- 1:53Grow and you have to read all the follow ups, genuinely act,
- 1:57you know, distinctive assets. But what I've sort of found in
- 2:00my career is that everyone sort of gets very obsessed within
- 2:04kind of the marketing books and what you're supposed to do. And
- 2:08actually, the book that I end up recommending to people is, I
- 2:10keep, I go, "But what fiction have you read recently?" And
- 2:13they sort of look at me, they go, "What do you mean?" I say,
- 2:15well... It's all well and good having sort of the basis in, you
- 2:18know, the science, the understanding, making sure you
- 2:20know what's going to actually work based on the kind of the
- 2:22principles and what's happening. But often, and especially in
- 2:26marketing and advertising, we're always thinking about the lives
- 2:30of people who we are not. Right? It's always like so, we you know
- 2:33we imagine ourselves as, you know, there's a... There's a mum
- 2:35and you know, she's looking at the kids and what's that life
- 2:37like? And often many of those marketers are thinking about
- 2:40that are in their twenties and they don't have kids and they
- 2:43don't know what that life is like. And they can't really
- 2:45imagine it. But actually, a lot of the fiction books that you
- 2:48read, especially the ones that are sort of about, you know,
- 2:52real life, I mean, the one that stood out for me, this is sort
- 2:53of, the one that came to my head when you asked me that question
- 2:56was Gymkhana. Like, I didn't really think about what it was
- 2:59like to be, you know, a Nigerian living in New York. What does
- 3:03that mean? Like, where do you go and get your haircut? What does-
- 3:05What are the conversations? What are the societal expectations of
- 3:08America versus you as someone from Nigeria going to America?
- 3:12What does it also mean in your reflections back when you got to
- 3:14Nigeria? And I've always done this, you know, since I started
- 3:18in advertising is I'll read a business book. It's like, you
- 3:21know, what's the latest business book? And, you know, what's the
- 3:24latest science? You know... You know, the Daniel Kahneman books,
- 3:27you know what we all believe about, you know, system one,
- 3:29system two. You know, the latest book on, you know, does
- 3:33everybody know about everything otherwise it doesn't really work
- 3:36and what happens in a society where, you know, you're all
- 3:38potentially in bubbles and not all agreeing that this is
- 3:41something we should all know about, but equally, I, then,
- 3:43always follow that up with a fiction book and I think, I
- 3:46would always recommend to people to make sure that you're not
- 3:48just sort of reading the books, which were about marketing, but
- 3:51actually going, are you reading books which are telling you
- 3:53about people's lives to give you that empathy and understanding
- 3:56of lives that you don't understand? And I've always felt
- 3:58that's made me, you know, a better, more rounded person
- 4:02because I'm aware that you go into a lot of meetings where
- 4:04people give opinions, and that opinion is generally based on my
- 4:09family, some of the stuff I've read, but a lot of it doesn't go
- 4:12sort of deeper than that. And that always concerns me, and I
- 4:15always want to ensure people are reading kind of more widely than
- 4:17they probably do.
- 4:18Tom Ollerton: Well, I think I've got an idea for a product there
- 4:20where it's just like you're pitching for a mum's in Belfast
- 4:23for Flora. These books are like a first-person description of
- 4:29what that life is like. All right, we're going after dads in
- 4:31Newcastle, so inescapable and so on. I've never heard that, Des.
- 4:34That is a... That is such an interesting, interesting way of
- 4:38researching an audience through fiction. Brilliant. I did this
- 4:41weird thing with my nephew last year where we listen to a brand
- 4:44new podcast every day for a year. It can be a serial.
- 4:47Completely different show every day for year, and you have to
- 4:50listen to something for 5 minutes and if after 5 minutes,
- 4:52you were done. That's fine. You did it. And I've gone really
- 4:56deep in Sleepy Science. It's a whole sleepy category of
- 4:59podcasts where they're interesting... Not interesting
- 5:03enough to keep you awake, but interesting enough to stimulate
- 5:06a mental chatter. You know what I mean? Like...
- 5:08Des Bateman: Yeah, yeah, yeah.
- 5:09Tom Ollerton: Lower pitch in Belfast. Sleepy silence, it's
- 5:12like, right, I think they talked about, I think they talked about
- 5:14quantum chemistry, and the relationship between... The
- 5:21science behind the, the world of quantum, you know, was it... was
- 5:25it super- superpositions where you can be both two things and
- 5:30all this stuff, and how that actually fits into psychology
- 5:32and how things like manifestation might indeed
- 5:34actually be related to science directly. And you just go to
- 5:37sleep and go "wow!" and yeah, I'm going to crank that out...
- 5:43but just put in your brain, I think in a completely different
- 5:46place, a real benefit or, like, not immediately somewhere at
- 5:51some point that information will resurface as a useful, useful
- 5:55Des Bateman: Well, all I'm saying, because actually, so you
- 5:55way of thinking.
- 5:57mentioned, when you mentioned that podcast, it made me think
- 5:59about, a lot of people read and, or listened to "Breath." I think
- 6:03it's Daniel [James] Nestor who writes a book on any sort of
- 6:05investigated all the things about breathing because there's
- 6:08a lot of things now.
- 6:09Tom Ollerton: I heard about this, yeah, yeah.
- 6:10Des Bateman: All that stuff and also the breath work. But what
- 6:12was interesting was when you, when I finished the book, which
- 6:16is very interesting, the things that he sort of discussed around
- 6:19like nasal breathing, making sure you're breathing through
- 6:21your nose...
- 6:21Tom Ollerton: Oh, that's right! Yeah, yeah.
- 6:24Des Bateman: What amazed me was and I think it shows you how far
- 6:27you need to- So the basic premise is sort of summed up in,
- 6:31you know, breathe through your nose at all times.
- 6:37Tom Ollerton: And yeah, yeah. I met a guy. I met him on this
- 6:38podcast. He had like a tape for his mouth.
- 6:39Des Bateman: Tape, yes, they do taping. So yeah. They do taping
- 6:41at night to make sure you still breathe through your nose. But
- 6:43what was amazing was in many ways you could write the sort of
- 6:46summary of that book in one page, which is breathe through
- 6:50your nose, here are some rhythms to change to kind of how you
- 6:54feeling emotionally around the way you breathe. But what was
- 6:57amazing for me in the book was what he was prepared to do to
- 7:01prove this and write chapters. So like he blocked up his nose
- 7:05for like a month and like tracked what happened when he
- 7:08wasn't able to nasal breathe and it was like, okay, you're taking
- 7:12this to the next level of like not only am I going to write
- 7:15about the simple answer, which is breathe through your nose,
- 7:18but I'm going to now demonstrate exactly what happens if you
- 7:21don't which I thought again is a lot of, you know, the way we're
- 7:25currently interacting with A.I. is it gives you a really quick
- 7:30answer. And you don't necessarily need to go that
- 7:32depth. And that book just made me sort of reminded me of like,
- 7:36you can easily accept, we can all easily accept things and go,
- 7:38"Oh, here is an answer. It's a good enough answer. We'll use
- 7:41that answer." And that book was a real kind of interesting
- 7:44explanation on well, you could accept that answer. But I'm not
- 7:48going to just accept it. I'm going to go research it, and I'm
- 7:51going to go prove it so that now, you know, now that person
- 7:54is the go-to person on, on breath because they've really
- 7:58done it. They haven't just gone, do you know what? I think it's
- 8:01breathe through your nose.
- 8:02Tom Ollerton: Yeah. Yeah. What did he follow it up with?
- 8:05Anyway, moving on. So I always ask this question, what is your
- 8:09top data-driven marketing bit of advice to become a better
- 8:14data-driven marketer that you find yourself sharing?
- 8:16Des Bateman: So for me, often is I mean, we interact, I have so
- 8:21much different types of data. So you'll get, you know, the
- 8:23classic sort of sources. GWI, Nielsen, you know, you're
- 8:27looking at research studies. And the one thing that I always say
- 8:30is often there's a lot of numbers and it's the
- 8:33interpretation around those numbers that's more important
- 8:36and it's a bit like so you've read those numbers, but you
- 8:39know, have you really watched what that really means in daily
- 8:41life? Like have you actually seen what's happened you know,
- 8:45in a bar or in a pub if someone's going to buy a pint?
- 8:47So you might have all this data about, you know, how people
- 8:49choose to buy beer. You know what they do. It'll say, you
- 8:52know, these are the top beers, right? But then if you actually
- 8:56work in that category, you realise, you know, there are
- 8:58things that the data doesn't tell you, right, because it
- 9:00doesn't. So an example is, you know, say for example, for a
- 9:04while when I was working on it, Stella was the most popular
- 9:06premium drink that people would go in. And so the, the data
- 9:09would tell you you're likely to sell it more often, all these
- 9:12things. But what happens if someone goes in a pub and there
- 9:15is no Stella. Right? Does the data tell you that people walk
- 9:18out to find the pub Stella and the data- No, no, and it's like,
- 9:22well, they must walk out to go and get Stella, because they're
- 9:25a Stella drinker and they, they're going to walk out of
- 9:27this pub because Stella's not there. That doesn't happen. What
- 9:29happens is that person looks and goes, well if they don't have
- 9:32Stella, what else do they have? And at the time it was, oh, they
- 9:35have Kronenberg, Kronenberg, Stella, pretty similar. I'll
- 9:38have Kronenberg, and it's so much of what we do with looking
- 9:42at data. It's not then looking at it and going... Well, one, do
- 9:46I believe it? Right, and I think because you know, often I
- 9:50remember when I first started... This is a very good example,
- 9:54TGI, I think when they first were doing their surveys, just
- 9:56to like, how do you interact with media and what do you do?
- 9:59If they had one tick, like people like to think that they
- 10:02read The Economist, so loads of people would go, "Oh, yeah, I'm
- 10:05the sort of person who would read The Economist" tick. They
- 10:07never read The Economist because if they did read The Economist,
- 10:11then it would be the best selling magazine the world has
- 10:13ever known, but obviously they now have to manipulate that data
- 10:16and go, "Well, all these people have said they've read The
- 10:18Economist, but we know they don't read The Economist so
- 10:21we’re gonna manipulate the data and numbers so that we get
- 10:23around about the right number of people that are saying they read
- 10:26The Economist." Right? And that's why you always need, you
- 10:28know, sort of that Bayesian mathematical kind of look of
- 10:30going, "Yeah, we know you said that, but we also have a, have a
- 10:35data point up here which says you 100% can't all be doing
- 10:38that." And it's really sort of not just accepting the data and
- 10:41going. Right, that's definitely, well the data says that, it's
- 10:44going... Okay, do I believe it? How has that data been
- 10:47collected? And I think we live in a world where so much data is
- 10:52digital and everyone has an acceptance of digital data as if
- 10:56the digital data is true or, you know, like eye tracking data and
- 10:59going, well, we've tracked their eyes and we know what's going
- 11:02on. And the example I give on that is if you watch TV tracking
- 11:06and eye tracking and people watching, there's context that
- 11:10doesn't get included into what's happening in the room, right? So
- 11:13the example I use is as far as the eye tracking is concerned,
- 11:16I'm staring intently at the ad and so I'm definitely noticing
- 11:20the ad, the ad's going in 100%. No, the ad is right. What the
- 11:24eye tracking doesn't know is I've just had an argument with
- 11:27my wife and all I'm thinking about as I'm staring attentively
- 11:30at the TV is how do I get out of this argument? How can this end?
- 11:34How can we go back to being like having a conversation? How can
- 11:37we deflate the atmosphere of what I've just been created? The
- 11:40ad is actually then not doing the thing that the eye tracking
- 11:43would tell you the ad's doing. And I think we're in an industry
- 11:46at the moment where when I first started, there was a general
- 11:48acceptance that you couldn't accept everything is true. It
- 11:52was things that you knew that weren't true. And the danger of
- 11:55the digital data is like, well, it's facts. It's all fact. This
- 11:58is a fact. And I think the reality is it's not facts. There
- 12:02are things that you still don't know like there's a failed
- 12:04example, I think from the, I wanna say the 70s, there's an
- 12:07article... And in one of the points he says, even- And I
- 12:11think this is true today, we don't know when somebody fancies
- 12:14a Coca-Cola. Like, they could be walking along and they suddenly
- 12:17think, I fancy a Coke. And we still have no idea as to why
- 12:22that moment specifically happens in that person's head at that
- 12:25specific moment that they've gone, "Do you know what? I
- 12:27really fancy Coke." And they go and buy one but we act sometimes
- 12:31because of digital that we can be like, "Well, we know of
- 12:33course we know when they're going to buy a Coke because
- 12:35we've advertised them in this and done this. And we know that
- 12:37when this area... and the temperature is at a certain
- 12:39level, it makes it more likely." But if we're honest, we don't
- 12:43know but we know what's more likely to cause that moment. And
- 12:47we know that we can keep top of mind and we can have awareness.
- 12:50So that should that moment come that they're thinking, I fancy a
- 12:53fizzy drink on a certain day and I happen to be walking down the
- 12:56street, Coke's the one I think of. And it has availability in a
- 13:00store near where they've had that thought.
- 13:02Tom Ollerton: Oddly, we just wrote a book on this subject
- 13:04called "Using Data and Creativity in Marketing." And I
- 13:07interviewed 40 very interesting people, Rory Sutherland and
- 13:10various SVPs and CMOs and agency boards. That was the main thing
- 13:14that came out because I went into it because I run Automated
- 13:17Creative and you know, that's all about ones and zeros and
- 13:22data and proof. And, you know, we can tell our clients
- 13:25specifically what should go in their ads from a visual
- 13:28perspective to get them to go through the site or whatever it
- 13:30was or do the thing. And I sort went in going, "Oh yeah, data, I
- 13:35know all about this." But it came very clear like as I was
- 13:37doing the research for the book that... I don't know whose line
- 13:40it was but said something along the lines of don't look at what
- 13:43the data tells you, look at what it's obscuring or what it is
- 13:46hiding from us. But in the advertising industry we love the
- 13:49easy button, right? Yeah. I'll open a spreadsheet. There is all
- 13:52of the information. It has to be perfect because it says Google
- 13:56in the top left hand corner or Meta. We're spending so much
- 13:58money with these people, it must be right, but there are so many
- 14:02data points that aren't captured and can't be captured. And the
- 14:06assumption is because they aren't and can't, they're not as
- 14:09important.
- 14:10Des Bateman: Yes, and even, and even to the point of... So there
- 14:13are counter-intuitive things that so... When we were first
- 14:17building websites, websites were often how do you, how do you
- 14:20make this experience as frictionless as possible? So how
- 14:23do you make it do something that you get to a sale, right? An
- 14:28example where it's all captured within cars actually. Along
- 14:32cars, there's a certain amount of time if you configure a car
- 14:36online, if you make that really super simple that you can
- 14:39configure a car. So it takes you like a minute and 20 to be like,
- 14:44I want this, this, this... It configures your car. You can now
- 14:46decide whether or not you want to go test drive, order from a
- 14:49dealer. The amount of sales you get from that isn't as high as
- 14:52if you make it more difficult. Now, if you make it too
- 14:56difficult so to configure the car and decide on things like, I
- 14:59want this stitching, I want this bit, I want a bespoke this, I
- 15:02want this engine, these colours. Again, you'll get less sales
- 15:07because you made it too complicated but there's an
- 15:08amount of time that someone has invested that then makes it more
- 15:12likely they'll go to a test drive because they're like, I
- 15:15have, I have spent enough time building this that I'm not
- 15:18annoyed by you. You haven't, you haven't wound me up, so I'm not
- 15:20going, God, I just want to get to the end of this process. You
- 15:23have made it too easy for me to go, well, I don't really need to
- 15:26go through because I can, I can come back at any point and do
- 15:28it. So it's just a matter of time that allows me to go, oh,
- 15:31right, okay. I've invested enough of my time to feel like
- 15:35I'm now... I've built something that I don't want to lose, and
- 15:38I'm now going to book a test drive. And that's something that
- 15:40all the advice in the data before you've, you've tested
- 15:43that and AB tested in different times and tried different
- 15:45things, you wouldn't do. And also, there's very few people
- 15:49that would want to take that risk. There's very few clients
- 15:52who want to go, we're going to make this more difficult.
- 15:55Tom Ollerton: We had a lady on the podcast two years ago. She
- 15:58was at Bravissimo, which is a ladies bra manufacturer and
- 16:03gosh, so sorry I can't remember her name, it was 200 episodes
- 16:06ago and their whole thing was they found... Her shiny new
- 16:10object was something friction or deliberate friction. Whereas if
- 16:13you just, if you'd rattled someone through the sale, they
- 16:15would get an ill fitting bra and then they would go, oh, it's
- 16:18not, it's not my fault. It's the, it's the brand's. I don't
- 16:21like it. I'm not coming back. Whereas they had like lots of
- 16:23videos, like, you have to watch this video on how to measure
- 16:25yourself properly and fitting and all this kind of stuff and
- 16:28that, they had less sales, but their returns went massively
- 16:32down. So completely on those sales was much higher. So,
- 16:35anyway, Des, we're going to spend 4 hours talking about not
- 16:37your shiny new object and your shiny new object is
- 16:41interrogating audiences with A.I., so what is that?
- 16:46Des Bateman: So I think there's a lot of people are playing
- 16:48around in this space with sort of the agentic modelling and
- 16:51going like, okay, we've got an amount of data on people and we
- 16:54don't want to keep necessarily going at doing that sort of
- 16:57quality of asking those people more information we've asked
- 17:00them to fill in, you know, surveys, calls for information,
- 17:03data, but we don't want to interrogate them. We want to
- 17:05have, you know, people's ability to go, like, I've got certain
- 17:08questions. Now, the beauty now of like, you know, so I suppose
- 17:11I could call it, the ease of symbols, the ChatGPT layer or
- 17:14the, you know, the Microsoft Office kind of going in being
- 17:17able to interrogate that is very interesting to be able to sort
- 17:20of go, I'm going to ask questions of that and then going
- 17:23to find out more information. Now, we have to be very careful
- 17:25around it because I think a lot of the way that people are doing
- 17:27this modelling isn't necessarily, you know, it's not
- 17:31necessarily giving you correct answers in the way that you have
- 17:34to deal with and build it and train it. Now, I was working
- 17:37with a company called Brilliant & Bloom, I was advising them, I
- 17:39met a guy who, I felt like everyone should have him, he was
- 17:41the best and he was one of developers, called Dev, Dev
- 17:44Gadgil and the company's kind of run by a guy called Mark Watkins
- 17:48but what was interesting is that they were building such a high
- 17:50level of quality over the qualitative data that they were
- 17:53doing and the rigour around this data, but specifically around
- 17:56high net worth individuals. Now, I used to work on, on UBS and
- 18:01UBS, one of the biggest challenges is very wealthy
- 18:04people don't really want to fill in surveys and talk to you very
- 18:06often. And so you kind of are often struggling with how do you
- 18:11get information from high net worth individuals. Now, what
- 18:14this company did, which is why it's much like toys, they're
- 18:16like how can we remove the friction from that? You know
- 18:20what it is... Oh, there's Dev! He's looking very smart. I wish
- 18:22I look as smart as Dev. I don't look as smart and I'm definitely
- 18:25not, the brain power the same. And it's like what was amazing
- 18:29is it allowed you to go, there's all this information about these
- 18:33audiences and it allows you interrogate it. But also what
- 18:37they did is what was built building like, so if we say it
- 18:39does this and builds these personas, the personas then
- 18:42really useful for you to play around with an advertising point
- 18:45of view. So like does that work? But then also the feedback loop
- 18:48back to those people because those people are part of the
- 18:50network that you can then go to interrogate that as much as
- 18:53you've interrogated all the data sources, is that actually true
- 18:56or correct? And I think it's that sort of final loop
- 18:58backwards that makes it very exciting for me that we're in a
- 19:01point where we don't have to annoy people so much about all
- 19:04the questions that we have because large language learning
- 19:08models are able to go through all the information, extrapolate
- 19:11information out of that. If you've got enough data points
- 19:14and sources within there, you can extrapolate out further. And
- 19:18the trick, I think is really going, okay, how does that then
- 19:21give me a competitive edge? And that's what I think is exciting,
- 19:23is that we're in an almost, I suppose like, yeah, I don't like
- 19:26calling that. People call rooms like, you know, war rooms and
- 19:29arms race, that sort of thing. Don't we think we should be
- 19:31aligning ourselves with that being our goal in marketing? But
- 19:34you can see there is a competitive advantage race
- 19:37happening at the moment around the new tech that's available
- 19:40and what that's going to mean. And I think it's a very
- 19:43fascinating area to be playing around with a lot of these tools
- 19:46and systems that are going to allow you to do things that, you
- 19:49know, back when I started, you 100% couldn't do. You were like,
- 19:53okay, we've got a data point, we can't really interrogate it. We
- 19:57can maybe spend a bit of money to get people to come in and sit
- 20:00in a room and we can talk to them, but we can't do it at
- 20:02scale. And it's happening at scale and it's happening now
- 20:05much more easily multi market because of the ability for like
- 20:09instant translation. You know, it's like we speak to someone in
- 20:12Switzerland, someone in Germany, someone from Italy, and it's, I
- 20:15don't have to wait now for translators to tell me what was
- 20:18the meaning of that, what does that say, can I interrogate it.
- 20:21Tom Ollerton: So to push back on your earlier points. So there's
- 20:24two things with that. So A.I., I think A.I. is failing when it,
- 20:29when it's just a little speed and scale because everyone else
- 20:33can do it.
- 20:34Des Bateman: Yes.
- 20:34Tom Ollerton: Right? And I've talked about this on the podcast
- 20:36before, a friend of mine, Amsterdam, reduces production
- 20:39costs by 750 grand to like, 30 grand, something like that, the
- 20:44numbers are loosely correct. And he's like amazing FD or
- 20:47something. And I'm like yes, so have all the other FDs at all
- 20:49the other brands who've done exactly same thing. And now
- 20:51you've, you've levelled the playing field for everyone with
- 20:53a 30 grand budget. So then the thing is like what can't you
- 20:56do... So what can you do that you couldn't do before and so I
- 20:59get your point, right. Yes, translations but that's, I'd say
- 21:02that's still speed and scale. None of that data is capturing
- 21:05the stuff that you were talking about before, which is the soft
- 21:07stuff that you can't see. Do you know what I mean? So someone's
- 21:10going, yeah, like I love watches that...
- 21:11Des Bateman: Completely, and that's, and that's what I liked
- 21:13about their A.I. push was the underpinning of the interviewing
- 21:17of the people still within that community and creating the
- 21:19community because without that underpinning, without the
- 21:23underpinning and not-
- 21:24Tom Ollerton: When you- Sorry, just, I'm being thick here. So
- 21:27when you say underpinning, do you mean like they would, the
- 21:30A.I. for argument's sake would go, oh, if you wanna sell
- 21:33watches, it's all about people with big arms or something. And
- 21:37then they, then they go... It's probably true. And then and then
- 21:41they go out and interview people go, oh, how big are you?
- 21:44Des Bateman: Yes. So essentially you have a large datasets
- 21:47collected around kind of what are the, what your interests
- 21:50are. So essentially, there's an agency that they use called
- 21:52Tribes who are actually very good, you know, qualitative
- 21:54agency... paid to survey, sort of all the research, you then
- 21:58got that, sort of, ground of data that's there that is based
- 22:01on that qualitative, you know, high in depth sort of studies.
- 22:05It's then put into like so the A.I. extrapolation of how do you
- 22:09make that usable at scale across you know, different sort of
- 22:12questions and way to interrogate that data in more interesting
- 22:14ways to get to your point, which is, you know, oh, look, we think
- 22:18the way this data is the extrapolation of that from LLMs
- 22:21and its ability to look at other data sources that are happening
- 22:24that aren't those people that are stuff online. These look
- 22:27like those things are going to be interesting. The bit that I
- 22:31think you can do, I think we talked about at the beginning is
- 22:33you can't just accept that answer. So you can't just then
- 22:36go, you can't just then go, all right, that's the answer. But it
- 22:40gets you to play around with interesting areas much more
- 22:42quickly that you can then go, we think these four Os are
- 22:45reaching, now we can loop back to the beginning of it, which is
- 22:47back into that, you know, community of people that we can
- 22:51interrogate. But we haven't had to go because the challenge I
- 22:54always found with, on the high net worth clients. It's always,
- 22:57you sort of go in for a little bit of information and you go
- 23:01away and you take that data and you go, okay, and then you want
- 23:03to go back again. And you almost are going back before you've
- 23:07really able to properly play around with what information is
- 23:10actually there and what happens with the high net worth
- 23:13individual especially, they are notoriously hard, mainly because
- 23:15they cost more, you know, to interview, to get their
- 23:18information and data, but they don't want to have that second
- 23:21follow up.. And if you've got like a tool or a system that
- 23:25allows you to go oh, okay, right, we can make you part of
- 23:27this community where interrogating stuff is
- 23:30interesting, that we can then come back to you less often, but
- 23:33with more useful stuff to come back to you with, that's a
- 23:36really useful thing. And I think that's what is that sort of
- 23:39slight point of difference that they had, which I thought was
- 23:41very, you know, especially having worked on those types of
- 23:44clients, I was always, I was like, oh, like 100% want to sign
- 23:47up for that.
- 23:48Tom Ollerton: So as of the 6th of Jan 2026, got to get used to
- 23:53saying that... Where are you at with A.I.? Are you A.I. positive
- 23:56or are you skeptical in any way? Because I saw a brilliant
- 23:58article from October, totally missed on businesses either and
- 24:02it was their early users of the... Meta, just give us an
- 24:07image, it will make all your ads for you, which... to a business
- 24:09like us where we, create ads at scale, but with that humans
- 24:12craft and strategic thinking to it, Meta doing that, you're
- 24:15like, wow, okay, this is going to be interesting. And then what
- 24:18happened at like this guy had an ad, he's just like a good
- 24:21looking bloke in a jumper and a bunch of USPs or whatever. And
- 24:24that was his best performing ad, static, whatever. And it's been
- 24:27replaced by a grand granny, wearing a, wearing a jumper. And
- 24:31it was obviously A.I. slop, you could spot it a mile off. And
- 24:34then it had been taken out and it hit their community. You go,
- 24:37why is it, why is a- This is for a sexy young Gen Z-ers or
- 24:41whatever, and all this, you know. And it's easy to, you
- 24:45know, like dump shade on, you know, a bit tech that goes wrong
- 24:50once or whatever. But there is, it's still a hallucinating
- 24:54thing. Like, yes, you get an answer out of ChatGPT and it
- 24:56sounds so coherent, but how often do you go like, is this
- 25:00is, does this look good or is it good?
- 25:02Des Bateman: So at the moment. I think it's a bit like when you
- 25:04read a newspaper. So this is like... So when you read a
- 25:06newspaper and you trust all the articles in the newspaper, don't
- 25:09you? You're like, you read you know, you read up an article and
- 25:11it says, you know, the sports writer's telling you what's
- 25:14going on at Chelsea and you're like, wow, that sounds like
- 25:16that's everything that's happening at Chelsea. And then
- 25:18you read another article and it's about the tobacco industry,
- 25:21and you go, well, that sounds exactly right. So what's
- 25:22happenning in the tobacco industry? And then you read
- 25:24about your own industry and when you read about your own
- 25:26industry, you go, well, that's not how that works. That's
- 25:29not... That's not right, right? But you've trusted every single
- 25:33other bit of news in that paper. You haven't, but you start to
- 25:36question the one that's in your area, but you don't, you don't
- 25:39read the other bits of the newspaper going, well, is it
- 25:42right? Does that person really have that knowledge? That sounds
- 25:44a bit made up because you don't have that knowledge. And I think
- 25:46with A.I., it's a bit like that, whereas I think if you have that
- 25:50knowledge and you have skills, it's at that point where you can
- 25:52go, oh, well, no, that's, that's not right. It's kind of there,
- 25:57but it's not, it's not correct. And you can see it, you know,
- 26:00when they done test it with legal cases where they've got a
- 26:03real solicitor versus A.I. solicitor. And to your point,
- 26:06the cost goes down using the A.I. solicitor but the more
- 26:09likely person to win the case currently would actually still
- 26:12be the solicitor you paid loads of money to. Over time it's
- 26:16going to get better.
- 26:17Tom Ollerton: Well, you say, you say this, right. But I think the
- 26:20question to ask is if I, if this came slowly and it was
- 26:23expensive, would I be into it?
- 26:25Des Bateman: That's a good question.
- 26:26Tom Ollerton: Do you know what I mean? Like, if I handed my three
- 26:27weeks for this and it cost me ten grand for this, like, you
- 26:30know, page of copy or whatever... No, no. And I think
- 26:36that the whole A.I. bubble is built on that premise of...
- 26:39Yeah, but, yeah, but yeah, but you know, like and then, then
- 26:43OpenAI go and raise however many trillions and you know, Google
- 26:47do whatever... Everyone does the same and everyone thinks right,
- 26:50if you put money in the top, quality will eventually go up.
- 26:52Des Bateman: There's definitely an A.I. bubble. I mean there's
- 26:54too many people talking about it. The most interesting thing I
- 26:56read that I, I think is fascinating is there is only so
- 26:59much data that can be ingested and at some point there's no
- 27:02more data. So then you're just learning of yourself and your
- 27:04own data, which again is, is quite interesting. Sort of go,
- 27:06okay, there's a bit of me that also laughs at the thing that it
- 27:10learns of is us, and let's face it, we're not that bright. So I
- 27:13find that quite funny that this thing is learning off from us.
- 27:16Tom Ollerton: The other angle I heard on this is that, was it...
- 27:18I’m going to say, was it Yann LeCun? And I think he was a, he
- 27:21was Head of A.I. at Meta, and he did..., and his whole thing now
- 27:27is training A.I. on video. Right? So if, if I train an A.I.
- 27:34on text, there's obviously so much subjectivity involved in
- 27:39that, right, whereas if you have a video, a thousand videos,
- 27:4210,000 videos of me knocking a glass of water off the table and
- 27:45landing on the floor and it's smashing.
- 27:48Des Bateman: Yeah.
- 27:48Tom Ollerton: And then you, you did a prompt. Shy, ginger bloke
- 27:51with beard, knocking down a glass at the table and smashing
- 27:54it on the floor, would be trained on reality to a degree,
- 27:58whereas you can have 10,000 descriptions of a glass being
- 28:03knocked off the table. The glass elegantly flew through the sky-
- 28:05Do you know what I mean? Language is incredibly powerful
- 28:08but is hugely limited at the same time.
- 28:09Des Bateman: But yeah, because also it's interesting. I don't
- 28:11know who, I don't know how, I've got a feeling it might well be
- 28:13Rory Sutherland that sort of said they thought- It might be
- 28:16Dave Trott, but it was to do with what words- When you read
- 28:21words, the reason why books are so good for everybody is because
- 28:25the swing about how language communicates, so something about
- 28:27how humans talk, right, is... Is something ridiculously low like
- 28:304% of the words we use, right, because we all take different
- 28:34meanings out of the words depending on our own life
- 28:36experiences or our own memories of what's being described of
- 28:39what that would mean for us. 7% of kind of that communication
- 28:44understanding is something like, the tonality. So the actual
- 28:48pitch of the words, if you said it. So I think there's a comedic
- 28:51sketch on, you know, I think it's Bill Burr, but it's like in
- 28:54a court of law, you know, when they're discussing a case, how
- 28:57somebody says a sentence really can change whether or not that
- 29:00was a good thing, you know, like, like someone says, "Do you
- 29:02want to do something?" "Leave me alone." Right, very clear that
- 29:06it's leave me alone. If you had, "leave me alone," like suddenly
- 29:10the words are exactly the same, but your meanings change with
- 29:13your tone. And then the other is completely, apparently, I think
- 29:16it's quite ridiculous... above 30% is just how you're standing
- 29:22and the other things that you're doing and the pace of your
- 29:24breathing that we're not necessarily aware of
- 29:27particularly. But that's creating a meaning as to how you
- 29:30feel about the situation. The words just, just on the basis of
- 29:33how we as humans understand words and why books become
- 29:36amazing, because we fill in all of these blanks and all of the
- 29:39stuff around it. Probably makes it quite a limiting thing in
- 29:42descriptions because it's why I think ChatGPT has made people
- 29:45oddly, not oddly, but better briefers because if you brief
- 29:51somebody like in our industry and they come back, it's not
- 29:54what you want. You're like, well, obviously they're the
- 29:56problem, right? Because they didn't understand my brief,
- 29:59right? If you brief a large language learning model and it
- 30:01comes back something that isn't right, it's got nothing to do
- 30:04with a large language learning model. It's because you have
- 30:06written it right, and that's why all these people are telling you
- 30:08how to prompt, what to write to get the right results. And so
- 30:12you've got this sort of interesting thing, which is
- 30:13like, actually, it makes people better briefers because they
- 30:16don't have to go, oh yeah, you're right. I didn't really
- 30:18state that clearly because it hasn't been stated clearly
- 30:21because I've left vagueness and things there.
- 30:24Tom Ollerton: Yeah, it doesn't sound like a $10 trillion
- 30:27business to me. But you know...
- 30:29Des Bateman: I don't think it's a $10 trillion business, I
- 30:31think, where do you want people to actually focus that will make
- 30:34a difference that tangibly makes that difference because if, to
- 30:38your point, if everyone's using ChatGPT, right, and so you can
- 30:42say ask for a media plan, you can ask for a marketing plan.
- 30:44Suddenly they all become the same because the data will tell
- 30:47you because it's collecting all of that they're all the same. As
- 30:50soon as you're accepting that you're losing competitive
- 30:52advantage. And I think the only way we'll win are where people
- 30:57go, oh, no, no, actually, we have this but we have this,
- 31:00which is the, the competitive advantage is that we do it this
- 31:03way or we also, to your point, we use, you know, we get actual
- 31:06creative directors, we get people who got opinion, we get
- 31:08skilled copywriters or tech people or directors. And, and
- 31:12adds that extra layer, and the extra layer becomes more
- 31:15valuable because it's this sort of weird thing, almost like, I
- 31:19don't know if, I might be quoting Rory Sutherland again,
- 31:21embarrassingly, but there's a weird example of like, do you
- 31:24prefer machines or do you prefer real people? And I think it's
- 31:27what car washes, you know, there are machines that you can go and
- 31:31get your car washed by a machine and you put your money and it
- 31:34does all the things and it cleans your car. Or you can get
- 31:36a group of people to go and clean it by hand and weirdly, a
- 31:40lot of people still prefer to go and get a clean by hand despite
- 31:42the fact that there’s a machine.
- 31:44Tom Ollerton: Des, this the longest podcast I've ever done.
- 31:4833:31, very enjoyable. Thank you for your time, but I've got to
- 31:52go and help run Automated Creative and do some good
- 31:57things, but thank you. Really, just a really funny chat and
- 32:02good luck with your new job.
- 32:03Des Bateman: Thank you, and good luck with everything.
- 32:05Tom Ollerton: I will be thinking about you at half three on
- 32:09Wednesday morning formatting the deck.
- 32:14Des Bateman: Oh yeah, yeah, yeah, yeah. I'll be- you can
- 32:16call me anytime, I'll be around.
- 32:19Tom Ollerton: Des, if someone's wanting to get in touch with
- 32:20you, what's a good place to do that and what makes a message
- 32:22you'll reply to? Quick answer.
- 32:24Des Bateman: LinkedIn. Probably the easiest way because you
- 32:27could email me at Desmond.Bateman@HavasMediaNetwork.com.
- 32:33But yeah, LinkedIn and just you know, I like meeting interesting
- 32:36people and I love having chats like this one.
- 32:40Tom Ollerton: Clearly, clearly you do and if you get any wins
- 32:43off the back of this podcast, I'm on commission, yeah?
- 32:45Des Bateman: Yeah, yeah. I'll make sure they come your way.
- 32:49Tom Ollerton: Des, thanks for your time.
- 32:50Des Bateman: Take care. All the best. Bye!