Latest / Shiny New Object - a Marketing Podcast / Avoid data paralysis & embrace performance AI - Evergreen Finance's Dan Saunders
Transcript
- 0:00Dan Saunders: A lot of people will say, Yes, we're data
- 0:02driven. Yes, we're data centric, but then they'll have data and
- 0:05data stacked up in a storage cupboard somewhere.
- 0:13Unknown: Are you a fitness loving foodie? Then defy snacks
- 0:16is for you. Defy offers real high quality chocolate you love
- 0:19combined with functional ingredients, giving you as much
- 0:21as 26 grammes of protein per bag. So you can be both a little
- 0:25bit bad and a little bit good at the same time as a female owned
- 0:28brand, defy dedicates a portion of its profits to other women
- 0:32owned businesses currently sold in the US. You can find us on
- 0:35defy with an i snacks.com and follow us on social at defy
- 0:39snacks.
- 0:46Tom Ollerton: Hello and welcome to the shiny new object podcast.
- 0:49My name is Tom Ollerton. I'm the founder of automated creative,
- 0:53the creative effectiveness ad tech platform, and this is a
- 0:56weekly podcast about the future of data driven marketing. Every
- 0:59week I have the pleasure and the privilege of interviewing one of
- 1:02our industry's leaders, and this week is no different. I'm on a
- 1:05call with Dan Saunders, who is head of performance marketing at
- 1:08Evergreen finance London. So Dan, for anyone who doesn't know
- 1:12who you are and what you do, could you give us a bit of
- 1:15Dan Saunders: Yeah, definitely. Thank you very much for that
- 1:15awesome introduction, mate. So a lot better than I'm going to
- 1:15give you now. But yeah, yeah, the I've been working in. My
- 1:15name is Dan Saunders. I've been working in marketing for the
- 1:15better part of 20 years now. Everything I do is very, very
- 1:20data driven. I spend a lot of time speaking to C suites or
- 1:24exco members, depending which way you you like to refer to us.
- 1:24background?
- 1:27So my job is basically showing how whatever we're doing is
- 1:32making money. I've got very good making other people more money,
- 1:37and that's what I'm good
- 1:48Tom Ollerton: Right. So, Dan, are you a marketing book reader?
- 1:51Is there a particular book that you find yourself talking about
- 1:55or not?
- 1:56Dan Saunders: Yeah, you know what? Actually, when it comes to
- 1:58these things, the probably the best book for marketing I've
- 2:03ever come across is actually an old sales book. It's gonna be a
- 2:07lot of sighs and a lot of eye rolls like this. But the best
- 2:11marketing book I've ever read is the way the wolf by Jordan
- 2:15belfont. You know the the guy from Wolf of Wall Street.
- 2:18Tom Ollerton: I do. I've not read it, and not met anyone as
- 2:22far as I know that's read it. So what are the main themes of that
- 2:24Dan Saunders: Basically, it's trying to get across the art of
- 2:25book?
- 2:28selling. You know, everyone always wants to show you how to
- 2:31be a better person at selling. My key driver is that everybody
- 2:36that works for a business that's not their own, even in some
- 2:40cases, businesses that are their own, their main job is sales. No
- 2:44matter what they do, it could be Dev. It could be trying to think
- 2:48So now, getting a bit more specific, have you got a
- 2:48of other non sales specifically. It could be ops. Everything they
- 2:52do is sales related. And this is why I think this book is so so
- 2:57useful for for marketers as well, because it's showing how
- 3:00you can get across better in life as a better salesperson. So
- 3:04these key main elements are that you need to show you are
- 3:08mentally agile, fast on your feet. You won't waste people's
- 3:12time so sharp as a tack is the phrase they use, enthusiastic as
- 3:13marketing tip to help people become better data driven
- 3:16hell. So you know that your product is the best thing offer
- 3:20your in some cases, a product for people like myself is
- 3:24themselves so enthusiastic as hell about your products, and
- 3:25marketers that you find yourself sharing?
- 3:28then an expert in your field is where the third main point is as
- 3:32well. So if you can establish those three things to whether
- 3:36it's your bosses, to your clients, to your customers,
- 3:39they're going to buy from you, to continually buy from you as
- 3:43well. So there's a lot of life lessons I draw from a very, very
- 3:47old sales book.
- 3:58Yeah, you know what? It's one that's been said on your show
- 4:01before, but it just rings true no matter where you say, Don't
- 4:06get stuck by data paralysis. People in our positions, we want
- 4:10as much data as possible. But often you look at it and you're
- 4:13like, well, where the hell do I start with this? If I do this,
- 4:16if I do X, Y, it's gonna make Z happen. But if I don't do A, B,
- 4:20C, is not going to happen, and that's going to interrupt that
- 4:23flow as well. So a lot of people will say, Yes, we're data
- 4:26driven. Yes, we're data centric. But then they'll have data and
- 4:29data stacked up in a storage cupboard somewhere, and they
- 4:32won't be using it as too frozen to deal with it. So get stuck
- 4:36in. Find a starting point. Find your own starting thread, using
- 4:40your own not just your own KPIs, but your own goals, and follow
- 4:44it along and really dig in from there.
- 4:53Tom Ollerton: So we're now going to come on to your shiny new
- 4:55object, which is performance AI. So a lot of people listening to
- 5:00this will think they know what that means. But what is your
- 5:02definition of performance AI?
- 5:04Dan Saunders: So the best thing I could always describe
- 5:07performance AI is things that used to be, they used to be
- 5:10subjective, you know, like artwork or content or just
- 5:14general, you know, even like UX perspectives, wonder, the new
- 5:18age of AI. These things have all become quantifiable, and as soon
- 5:22as they become quantifiable, they become performance metrics
- 5:26as well. So, you know, one of the things that I use in
- 5:30particularly performance and AI, is measuring the quality of
- 5:34content that we're putting out there. You know, it's using
- 5:38different factors. There's an agency actually, that use it
- 5:42really, really well. I've tried to take, I've tried to take my
- 5:46own version of this and really make it more for me and for my
- 5:50Tom Ollerton: humans wouldn't agree on but a human has to
- 5:50company. But, you know, it's looking at the actual quality of
- 5:54the content and being able to score, you know, using things
- 5:58like the Fisher Klein and scale the technical quality of what's
- 5:59provide the training data for that, AI, or am I looking at it
- 6:02there. Is it satisfactory? Is it for users? You know, what age is
- 6:06it? Does the age that it's written for match the average
- 6:10age of your of your persona users, you know, trying to match
- 6:14these simple, simple things, but things that we don't tend to
- 6:18take for... the things that we used to take for granted as
- 6:22subjective, but now that it's become more objective, as I say,
- 6:26then becomes more quantifiable. If it's quantifiable, then it
- 6:30falls on the realm of performance.
- 7:01in the wrong way?
- 7:08Dan Saunders: No, no, it's exactly that. And if you look
- 7:11into things like Adobe's adaptation, sorry, not
- 7:15adaptation, their acquisition of SEMrush or SEM rush, depending
- 7:18on the years that you've been in SEO, that is a key driving fact,
- 7:25that performance AI is becoming more and more of an element
- 7:28where Adobe, that was used a lot for imagery, has now come
- 7:34through. They've obviously gone through the the different
- 7:37channels in terms of different CMSs, different art platforms,
- 7:44etc, etc, but they're now using the SEMrush to then put a
- 7:50quantifiable score onto their own products. And this is, you
- 7:54know, this is the thing where we can see these multinational
- 7:56corporations are taking more notice of quantities. And this
- 8:01is the big thing that we know every time that you have to
- 8:03speak to a stakeholder, if you're speaking to a CMO CEO
- 8:07CFO, they're going to care one of three things. They're going
- 8:10to care about, time, money and resources. And if you're doing
- 8:14your job right as a performance expert, the thing that they'll
- 8:17care most about is actual time and resource. Money will be the
- 8:21Well, shouldn't always be the last thing, but it'll be, if you
- 8:23can show they invest one pound, that you're going to get 10
- 8:27pounds on the other other end of it, this is the key element. And
- 8:30now the fact that we can bring this into into the subjective
- 8:36areas, like, I say, art, content, etc, etc, this is the
- 8:39key element. And to answer your question, it's about taking, not
- 8:44just using tools in isolation, like, you know, Gemini, chatGPT,
- 8:49etc, etc. It's about taking a wider, wider scope of that. So
- 8:54if I say we look at my site, for example, our site, we know our
- 9:03our you know, we know. We know we were in a really, really
- 9:06brilliant position, the fact that we know everything about
- 9:08our customers, because they we have to get all the data right
- 9:11to make sure we're the right product for them, because we
- 9:15definitely don't want to be selling our products to someone
- 9:24that doesn't need it, or in fact doesn't actually want it. So
- 9:24it's about making sure that we deal with all these different
- 9:24personas types, and once you build with these different
- 9:24personas types, again, factoring into quantifiability.
- 9:29Quantifiability, the quantity of looking into that word that I
- 9:33just made up is, is the real depth and the real art of
- 9:39performance marketing at the moment.
- 9:47Tom Ollerton: this episode of the shiny new object podcast is
- 9:50brought to you in partnership with Mad fest, whether it's live
- 9:53in London or streamed online to the global marketing community,
- 9:57you can always expect a distinctive and daring blend of.
- 10:00Fast paced content, startup innovation pitches and
- 10:03unconventional entertainment from Mad fest events. You'll
- 10:05find me causing trouble on stage, recording live versions
- 10:08of this podcast and sharing a beer with the nicest and most
- 10:11influential people in marketing. Check it out at www dot madfest
- 10:16london.com.
- 10:23So what happens in a world where this stuff works 100% of the
- 10:28time? And that's a big if, oh yeah, it's a massive if. So if
- 10:33all media optimization is all automatable, then you just need
- 10:39the machine to do it right, and then if all the subjective
- 10:42elements that you're talking about are correct, you can have
- 10:45a machine automate all of the ads. Everything's going to be
- 10:48automated, creative and media. So what will be the competitive
- 10:53advantage that brands can have, apart from volume of media
- 10:57spend?
- 10:58Dan Saunders: But the best thing about this as well is and you
- 11:01know, same thing for all AI, eventually it's going to take
- 11:03certain jobs, but it's not going to take all jobs. The human
- 11:06element is going to be still there. The key differentiating
- 11:09factor as it stands, you know, it will eventually become 100%
- 11:13like you said, it's not there now, but at the moment in time,
- 11:16it's you put the the piece in, and the piece will then be
- 11:21scored against these models. So you know, if you went on right,
- 11:25you know some of the the content Fisher Klein scores, for
- 11:29example, are on a one to five or one to 100 if you then get a
- 11:33piece of content that scores a four out of five, then you
- 11:38think, actually, that's probably the best one to go with. You
- 11:41don't really need to spend too much time trying to get a five
- 11:45out of five, making sure that you're not held in place by that
- 11:48quantifiable score.
- 11:49Tom Ollerton: So what do you think the role of the
- 11:51performance marketer will be in the next couple of years? And
- 11:53once again, making the assumption everything that can
- 11:56be automated will be, what is the what is the role of the
- 12:00performance marketer, and for those people are probably
- 12:03rightly worried about their jobs. How do they make
- 12:06themselves future proof?
- 12:07Dan Saunders: It's more about balancing of the KPIs, because
- 12:11in any given scenario, AI will only, I mean, the key thing that
- 12:15we're given here is that AI is actually what people describe as
- 12:20artificial intelligence. At the moment, it's, you know, the real
- 12:25description is LLM, and you know, large language models is,
- 12:29as you rightly said, Tom, that this is all coming from what
- 12:32people have put in, and information has come into it as
- 12:35well, until it gets that point where it's evolving on its own,
- 12:38is never truly at that stage. But the key thing that the
- 12:42performance marketer is going to have to do, and will be doing
- 12:45going forward, is trying to balance those KPIs, because KPIs
- 12:49shift on a day to day basis. You know, I've been in, you know,
- 12:53I've been in this company for four years, and in this one
- 12:59particular I've never had the same KPI for more than sort of
- 13:02three, four months. And that's, that's the same for every single
- 13:07role that I've ever been in. There's always a differentiating
- 13:10KPI. There's always one large which is always try and get as
- 13:13much money as possible. But everything around that is the
- 13:16differentiation of, oh, actually, we need to care about
- 13:19these KPIs. We need to care about this KPI, that KPI, if the
- 13:23LLM hasn't had that information before, it's not going to know
- 13:27what it is. And that's where the performance marketer comes in to
- 13:30balance these information, these parts of information.
- 13:32Tom Ollerton: And with the subjective elements being
- 13:36tracked, what are the specific things that you think people
- 13:40listening to this podcast should be doing. You talked about
- 13:42matching copy to an age group, which is interesting. Often, age
- 13:46groups can be pretty broad. So what else are the other use
- 13:49cases where you're like, Oh, this is brilliant. This is
- 13:51really moving the needle for us.
- 13:53Dan Saunders: I mean, simple things, stuff that I've been
- 13:55studying for absolute ages. So it sounds really patronising
- 13:58when I say it's simple, but it's, you know, things like
- 14:01different age ranges, different people from different
- 14:04backgrounds, people from every differentiating value has a
- 14:10different relationship to certain elements. So for
- 14:13example, if we take the colour blue, so for my age group, when
- 14:19we're marketing, we associate different shades of blue in a
- 14:23negative fashion, or negative to neutral fashion, but more
- 14:27towards the the the newer generations that coming into
- 14:33marketing is what they see in a more positive neutral site as
- 14:37well. So there's a gradual shift along these different scoring
- 14:40elements. So this is where having your data rock solid,
- 14:43right from the start, knowing even simple things like your
- 14:47colour palettes, they they can impact heavily on the sharpness,
- 14:53sharpness of image, complexity of images. You know something.
- 14:58You know these, all these things that... UX developers think
- 15:02about all the time them, rather than it being their opinion.
- 15:07It's becoming more quantifiable to say, actually our users
- 15:10prefer X, Y and Z rather than ABC.
- 15:19Tom Ollerton: And how you talk about the colour blue, for
- 15:21example, how are you isolating that as being effective in a
- 15:26performance scenario? I understand that when I on a
- 15:29website, for example, Google, very famously tested 50,000
- 15:33shades of orange for their Buy Now button. So in a paid
- 15:38context, how are you isolating that variable. So you know that
- 15:42actually, young people do like that blue, and older people
- 15:44don't?
- 15:45Dan Saunders: Going back to basics, just doing simple AB
- 15:47tests as well, trying to run them under similar scenarios
- 15:51that you know even so again, this is, this is probably gonna
- 15:55be the same for everybody else. But you know, no two months are
- 15:59ever identical in the same sense that you know, just because one
- 16:03month has 31 days, another month has 31 days, you can expect it
- 16:07to have the same results of it, as well as trying to find those
- 16:12exact for those quality of points for testing and making
- 16:15sure the testing conditions are set perfectly. So like I said,
- 16:19just getting the basic right in terms of AB testing, having
- 16:23someone that's a data expert working alongside the UX team or
- 16:27the creative team or the branding team as well, to make
- 16:31sure that every decision you've done has a link back into what
- 16:35you're trying to achieve.
- 16:37Tom Ollerton: And so how would you manage the different types
- 16:39of subjective things that people are going to want to talk about,
- 16:42right? So your fixate is for argument's sake, on colour,
- 16:45right? Whereas someone else might be fixated on tone of
- 16:48voice, like, all of a sudden, aren't you going to have lots
- 16:51more stakeholders piling in here going, well, I want to know
- 16:54about feel or I want to know about like, culture fit, or all
- 16:57of this stuff that, as you say, before was subjective, now is,
- 17:02for argument's sake, quantifiable. So do you think
- 17:05that there's going to be a broader pool of people that need
- 17:08to sign this off? Because if you can track those things, then
- 17:11surely, even within a performance ad, the brand team
- 17:15are going to have a strong opinion if certain metrics
- 17:18Dan Saunders: Yeah, and this is why I'm a big fan of the omni
- 17:18aren't tracking correctly?
- 17:21channel approach. People shouldn't be working in KPI in
- 17:25silos, especially in different across different elements as
- 17:28well. Everyone needs to be working to the same goal, just
- 17:31because, say, your product team has to has their KPI of how many
- 17:35times, how? What? You know, what's the improvement rate of a
- 17:39form being filled in doesn't mean that's going to work well
- 17:42for the finance team, because it could be something else on the
- 17:45Tom Ollerton: Well, look, Dan, unfortunately, we're gonna have
- 17:46line as well. Everybody needs to have the same KPIs and have the
- 17:50same goals going through. But as soon as you've established what
- 17:54they are, that's where the omni channel and this element becomes
- 17:56to leave it there, but thank you so much. I learned a lot there,
- 17:58a lot more predominant, because you're going into things that
- 18:01actually matter and you have the data to support it.
- 18:09and I'm looking forward to staying in touch with you and
- 18:11seeing how this space develops. If anyone wants to get in touch
- 18:14with you, where would be a good place to do that, and what's a
- 18:17good message to you that you will actually reply to?
- 18:21Dan Saunders: Something that's anything that's not Hey dear
- 18:25anymore. There's still all of it, you know, even to the point
- 18:31now where people message me and say, Hey, I really like the look
- 18:33of your dog. That's it. I've had probably about four or five
- 18:37messages this year, you know, into the new year with that
- 18:39already. So, yeah, it's something, anything that's
- 18:44original, I will always give anything that's original and has
- 18:47substance to it. Not that "we help people in your situation,
- 18:52fix x, y and z." It's "Hey, Dan, I've noticed this. Is it worth
- 18:56us talking about?"
- 18:57Tom Ollerton: Absolutely brilliant advice. Dan, thank you
- 19:00so much for your time.
- 19:01Dan Saunders: Thank you very much for having me. Tom, it's
- 19:09really appreciated.
- 19:10Tom Ollerton: Hi, just before you go, I'd really appreciate it
- 19:13if you could take the time to write a review of the shiny new
- 19:17object podcast on Apple podcasts, or iTunes, whatever
- 19:20it's called these days, or whichever podcast provider you
- 19:23use. We're at any podcast, so it would go a long way for us if
- 19:28you could just share the Word and give us a bit of a support
- 19:29on those channels. That would just be fantastic. If you
- 19:29haven't got time. That's also cool. And yeah, if you could
- 19:35tell your colleagues about the podcast, and also, if possible,
- 19:39don't forget to subscribe. And I'd love to hear your feedback.
- 19:42If you'd like to speak on the podcast or be a guest, or you
- 19:45think I'm asking the wrong questions, anything. I'd be
- 19:48super interested to hear what you think. So please email me.
- 19:51Tom at, automated creative.net, that's Tom at, I'm not gonna
- 19:58bother spelling it anyway. You're working. Out. Thanks so
- 20:00much.