Wisdom Led Marketing & AI Audience Agents - with T&P's Mark Cochrane

It's really easy to get caught up in the deluge of data that we're all surrounded by. Mark Cochrane, Head of Media Strategy at T&P, has worked out a four-tiered hierarchy to move us from raw data to wisdom that generates actionable insights.

As Mark puts it, "we're often mesmerised by spreadsheets and dashboards... but the problem is we focus on the data that's easiest to get hold of, and not necessarily by the data that's most important."
Instead, his advice is to identify four tiers that connect data to wisdom:

1. Raw data, which is "a bit like crude oil, not powerful until it's processed and refined."
2. Information coming from organising the data so it has context
3. Knowledge, resulting from applying strategic expertise to understand "why what happened, happened."
4. Wisdom, i.e. thinking about how you're going to influence consumer behaviour and what you should do differently in the future to build the brand for the longer term.

One thing Mark has adopted as his shiny new object, in line with using refined data to reach marketing wisdom, is AI audience agents. Remember the boring customer personas? They're no longer flat and abstract, if you can train an AI model to generate synthetic customers that you can brainstorm with, analyse data, and use to prepare and optimise real-world audience interviews.

Learn more in the full podcast.

The following gives you a good idea of what was said, but it’s not 100% accurate.

Mark Cochrane 0:00
Wisdom isn't about data, it's about the human in it. Uncover human truths that will inform our strategy.

Speaker 0:07
Are you a fitness loving foodie? Then defy snacks is for you. Defy offers real high quality chocolate you love, combined with functional ingredients, giving you as much as 26 grammes of protein per bag. So you can be both a little bit bad and a little bit good at the same time as a female owned brand, defy dedicates a portion of its profits to other women owned businesses currently sold in the US. You can find us on defy with an i snacks.com and follow us on social at defisnacks.

Tom Ollerton 0:46
Hello and welcome to the shiny new object podcast. My name is Tom Ollerton. I'm the founder of automated creative, the creative effectiveness ad tech platform, and this is a weekly show about the future of data driven marketing. Every week or so I have the pleasure and the privilege of interviewing one of our industry's leaders. And this week's no different. I'm on a call with Mark Cochrane, who is head of media strategy at T&P. So Mark, for anyone who doesn't know who you are and what you do, could you give us a bit of background?

Mark Cochrane 1:14
Hi, Tom, thanks very much for having me on. Yeah. So I'm, I'm a marketing strategist, media strategist, which basically means I help develop the master plan for how some of the biggest companies in the world talk to customers, talk to consumers, to drive the growth of their business. And I guess my sort of route to becoming that is 20 years in the in the 20 years long really. But I started off my career working in creative agencies and account management, and then a couple of couple years that found myself working for dunhumby, right in the very early days of that business. Who are you know were one of the first organisations to really use customer data really effectively, and that was a springboard for me, and led me to working in media agencies ever since my stint there. And really what my job is now, and has always been since dunhumby, is really about trying to connect businesses with the humans out there in the world that they need to connect with to drive their objectives. So that's really my job. Is just understanding people. I think grandly, call it human insight. But you know, it's about understanding people and helping that, helping that businesses, you know, connect, connect with that, to meet, meet the needs of those consumers and be successful as businesses and brands.

Tom Ollerton 2:34
So are you a marketing book reader? And if so, do you have a title that you recommend to people most often.

Mark Cochrane 2:42
I don't. I do. I do read marketing books. I don't probably read as many as I should, but and the one that I recommend most is, I think, one of the first ones I read, actually, and I always recommend it primarily because, actually, it's not a marketing book at all. Instead, it's, it's probably the best book I've ever read that that goes some way to explaining how the human mind actually works, and that, that's the book written by Daniel Kahneman, who's a Nobel Prize winning psychologist, and it's called Thinking Fast and Slow. So I'm sure lots of your listeners will have, will have heard of that and probably read that book already, but certainly if you are in the field of marketing and you have a recommend, haven't read it, I would highly recommend reading it. I mean, what it does is, you know, as I said, it explains how we as humans really begin to think. And you know, if our job as marketers is to influence people, we have to understand why people do what they do, you know. And if you know, we've got all these marketing theories, but they're pretty useless unless you know, we actually understand how people make decisions, you know. So Daniel Kahneman, if people aren't familiar, was one of the sort of fathers of this field of behavioural economics, and what he does in the book is explain that our... the way our brain works, really, and this idea that it has sort of two operating systems, system one and system two. And that idea that, you know, system one is this very fast, sort of emotional part of our brains that makes these sort of intuitive, you know, almost, you know, subconscious decisions very, very quickly, you know, using lots of shortcuts and biases and rules of thumb to do it. And then system two is the slow and much more methodical, logical kind of analytical part of our brains, and that, you know, is much more resource intensive uses our sort of rational, conscious thinking. And actually, you know, whilst you might think that all the decision making feels like system two. We use that for far fewer decisions than system one. And, you know, really avoid using that sort of that system two thinking wherever possible. And I think what's amazing about this book is it really illustrates how those two systems. Um, work with one another, the kind of constant interplay between them and the conflict between, I guess, that emotional, intuitive nature, and our rational, controlled mind. So everyone is listening, all of us, we feel that every day, you know, we will often have make an emotional decision about something. I'd quite like that. I quite like to own that, or, you know, eat that, or try that, and then we our control, our rational, controlled mind jumps in and says, Well, are you sure about that? Can you afford it? And that's this, how these two systems work. And in the book, Kahneman defines a number of the sort of mental shortcuts and the biases that brains make to help with the decision making we have, which is many of those have been used to create hugely impactful marketing. And I think the one that I always find most interesting in that is what we call anchoring. So there's this sort of tendency to be really influenced, overly influenced by that first piece of information that we get. So that idea of, you know, for example, if you see an estate agent shows you a house for sale, the list price of that home is the anchor. And you know that that that very much affects what we think the value of that property will be, whether we, you know, are aware of it or not. You know. And anyone who shops on Amazon over Prime Day, you know, needs to be really aware of that as well. You know, if you've been told that price has been discounted, and it was this, and it is that, now you know that anchor of those pricing is a hugely powerful effect, even if we are very much aware of, you know, what game is being played by, by those brands and those those businesses.

Tom Ollerton 6:39
And so how are you how are you using that day to day? Is it one of those things which is like a lovely thing to have, or are you specifically using anchoring as a word or a term, or are you going through Thinking Fast and Slow to when you pitch ideas or looking to get a client on board?

Mark Cochrane 6:54
Yeah. I mean, I think, you know, as marketers, and you know, obviously I work with with lots of big brands and on their behalf, you know, I think it's just at a fundamental level it's so important we understand the way that people make decisions, because that's how advertising works. And, you know, we know that, you know, in the long term, certainly the world's best advertising does its work with that system 1 thinking, you know, it uses colours, shapes, jingles, all the sorts of shortcuts that grab the attention of that fast moving animal-like brain, that the advertising like that creates these emotional associations that make us feel really good about our brand. And they really aim to build imprint in people's brains that build that mental availability, as you often call it, so that when a buying need does arise. You know, the that brand that's been advertised is the one that sort of instinctively and effortlessly comes to mind. And then, sort of conversely, we know that, you know, advertising that has loads of sort of rational product features and reasons to buy are very reliant on system two thinking that much slower moving, more rational part of the brain, and actually more often than not, that the brain is too busy doing other things to notice and you know what? Hence why, you know, a lot of the time I find myself in in rooms with clients being the one to remind people, you know, often slightly more politely than this, that people just don't give a shit. You know, consumers really don't care. But we often have to advertise. And it's kind of this Thinking Fast and Slow that really helped me begin to understand how we can, how we can change that, you know, that sort of revealing, that really fundamental truth of what we do in our profession, in marketing, is that we're not talking to like a rational actor in a spreadsheet we're trying to connect with, you know, enormously, beautifully complex and often irrational humans as I think that kind of foundational understanding for me is, you know, has really been where, where, where I begin most of my work, you know, we're responding to any, any sort of brief, working on any sort of particular marketing challenge, you know, recognising that you can't just say it rationally, that won't, yeah, it simply won't connect.

Tom Ollerton 9:18
So you've given us a book. Now, what I'm looking for is a very specific bit of advice. What is your top tip when it comes to advising someone to be a better data driven marketer?

Mark Cochrane 9:35
I mean, I always love this question, and I think my advice, my answer, at least might be a bit controversial for a podcast about data driven marketing, but my sort of best advice here is to stop obsessing over the data. And I know that sounds weird, but let me sort of explain. You know, we are not short on data, and you know, far from it, we're we're in a sort of data deluge, aren't we, and most of us, and certainly lots of the brands and organisations I work with, are kind of treading water just to try and keep their heads above the water. And we're often really mesmerised by spreadsheets and dashboards and all of this kind of, you know, hugely clever bits of technology. But the problem is often we focus on the data that's easiest to get hold of, and not necessarily the data that's most important. And so first of all, we get this over emphasis on lower funnel performance, clicks, conversions, cost per acquisition. And that's because we can get it and it's really immediate. But actually what I find is, if you want to become a better data driven marketer, first, it's important to understand, you know, the difference between data and, you know, ultimately, wisdom. You know, it's wisdom is the thing that that makes you a better marketeer. Ultimately, and recognise this kind of there's this hierarchy that connects data to wisdom. It's a sort of four layer hierarchy, I guess. And so at the bottom of that layer you have data, so, you know, raw data, raw noise, just clicks, impressions, open rates, you know, and relatively, on its own, mostly useless. But because of the sheer volume of it, we're kind of, we're often obsessed with it, but it's not really, until we begin to move up that hierarchy that its values realised. And you know, back in my days at dunnhumby, Clive Humby, who found dunhumby with his partner, Edwina, he coined this phrase, data is the new oil. Again, I'm sure lots of you would have heard that, but a bit like raw crude oil, raw data isn't powerful until it's processed and refined. In its raw state, like crude oil, it's pretty costly and risky to manage, so if we're in our oil refinery and we're moving up our hierarchy, so I'm conflating analogies here, but the next layer up is information. So organise your data. So actually not just clicks, but these are clicks from our summer campaign on this particular platform, you know. So that that data now has got context, but it's but it's still just telling us what happened and and often that's where a lot of data driven marketing kind of get, get stuck. So again, you need to move up the next level, the third, the third level up, which is, you know, knowledge. So when we apply strategic expertise to information, you know, further refine it, and we under we then want to understand why has what happened happened. So if we're talking about our summer campaign, we can see, for example, our summer campaign drove higher engagement because the creative tapped into a powerful human insight about a desire for connection, for example. So, you know, that's then a genuine learning. That's real knowledge about understanding. So we've gone from just about clicks on a platform to big, sort of a real understanding of something. But actually the real goal, the sort of top of the hierarchy of Pinnacle, is wisdom. So wisdom takes that knowledge and refines it, you know, to its fine, you know, finest level, through applying foresight, so thinking about how we're going to influence future behaviour. So not what happened in the past, but you know, what should we be doing differently in the future to build our brand for long term, for example? So wisdom isn't about data. It's about the human in it. It's about saying this insight about our summer campaign that showed us as this human desire for connection, for example, is, you know, powerfully important piece of information. So why don't we build our brand around that in future, to create mental availability, for example? So you know that that kind of that, that movement from just simple raw data all the way to something that really influences what we go to do in the future. You know, that for me, would be my kind of advice there. So, you know, move away from just counting clicks to actually trying to understand people. Use it. You know, it doesn't have to be huge amounts of data. It can just be a small piece, but uncover human truths that will inform our strategy. So, you know, don't be data driven. Strive to be kind of wisdom led, you know, use data to not just create spreadsheets, but understand human insights so that you can connect with people much more effectively. And I think that is, you know, is more true now than it's ever been, given, you know, in a world where AI there is a concern, at least AI is eating jobs and is going to be doing an awful lot of things for us. You know, AI can crunch an almost endless amount of data and turn it to information and provide us with knowledge. But actually creating wisdom needs humans and so, you know, my recommendation would certainly be, you know, focus, focus there.

Tom Ollerton 14:39
Yeah, the kind of automated creative version of that is, I always say everyone's got the three D's, right, which is decks, data and dashboards. But actually what any brand needs or agency as well is inspiration, right? However, that data comes back and as is processed into wisdom, it should be like the label should we go on? Oh, brilliant. Oh, let's go and do that. Then. Right? If you're not inspired into action, then that it is, as you say, it's a lower order thing. I was and I was first, I think I was first talked to about this by, I think it was Perla Bloom, so, Perla can have this, even if it wasn't you. But I interviewed her for a book I'm just about to publish called Using creativity and data in marketing. And her view as a strategist was that you have to, you have to talk about the data in a way that the creative person goes, Oh, brilliant. I now know what to do. And this inspiration is lacking. I find data and decks and dashboards everywhere. Inspiration a lot harder to find, I feel.

Tom Ollerton 15:42
This episode of the shiny new object podcast is brought to you in partnership with Madfest, whether it's live in London or streamed online to the global marketing community, you can always expect a distinctive and daring blend of fast paced content, startup innovation pitches and unconventional entertainment from Madfest events, you'll find me causing trouble on stage, recording live versions of this podcast and sharing a beer with the nicest and most influential people in marketing. Check it out at www.madfestlondon.com.

Tom Ollerton 16:13
So we're going to move on now to your shiny new object, which is AI audience agents. There's some lovely alliteration there. I didn't, I didn't notice, but I think I know what that is. Everyone's got a different version, I think. But could you explain what an AI audience agent is and why you've chosen it as your shiny new object?

Mark Cochrane 16:38
Sure. Yeah. So, I mean, what an AI audience agent does is really it moves us beyond what lots of lots of us will fondly remember and still use to a great extent, you know, these sort of PowerPoint, pen portrait views of people or segments, and it gives us instead a much more dynamic, interactive model of consumers, a synthetic consumer that we can talk to and collaborate with. So, you know? So back to my time working at dunhumby. You know, that was... we crunched huge amounts of data collected through grocery retail loyalty card schemes, so often Tesco club card, for example. You know, can they those things can tell us huge amounts about consumers, because you are what you eat. And so like, whilst the what we uncovered was really transformational in many instances, you know, often the sort of personas, or the customer segments and the pen portraits that we would create that we then use to help marketing teams understand those people, you know, had to be pretty simple and, you know, and they really lack detail. They're often a bit sort of simplistic, and there is a risk of sort of falling into stereotypes. And actually that, you know, that was 15 years ago, and I don't think our approach has really changed much since. So we often work with these, you know, overly simplistic, static pictures of consumers, and that that's not great for marketing effectiveness. And what's exciting about audience agents is that rather than just analysing what customers said in the past, you can interview a model, a synthetic consumer right now. So you can test ideas and hypotheses, and they work just like any other large language model. So chat, GPT, etc, you know, you interact with it with a, with a, you know, normally a, you know, you type in interface. But instead of getting sort of generic responses to the questions that you ask, you hear really, this sort of real, really specific perspective of a particular type of person, which means we can ask questions and get responses to ideas in real time. So whether that's, you know, you're talking to a synthetic consumer for, you know, an automotive manufacturer, you could ask them what they're, you know, what, why they use a particular type of car. You know, why are you considering buying an SUV? And you'll hear back from that agent if it's been built with the right data. We know what you know, what their what their view is, and why they do what they do. And equally, you can then test, you know, creative work and lines and all sorts of things against with these audiences in real time. So it's, you know, it really allows us to avoid creating campaigns that are sort of based on flawed or incomplete understanding of the audience, and really revolutionises how we bring audience insight into that, the sort of marketing and advertising campaign development.

Tom Ollerton 19:23
So I'd never thought about AI audiences in terms of the like the pen portrait one pager that always have those, like, unbelievably annoying names that, you know, there's always, there's always two words to them aren't there. They're like, hungry, obsessesers or whatever. And it always turned my stomach because it's...

Mark Cochrane 19:42
Just yeah, a random set of...

Tom Ollerton 19:45
And there would be like a stock image or a Gen AI image, which sounds a bit more recently, of like someone in pink jeans sat on a bench on their phone or something like that, right? And I was always like, what am I supposed to do with this? Right? But they seemed that was the best possible thing you could use, whereas, compared against that, this is like a ginormous leap forward. But how are you comforting a sceptic who's like the idea of calling something synthetic data? I don't know what the posh word is, but it doesn't make any sense, right? You can't have synthetic data because that data is tracks the thing that happens, so it didn't happen, then it's not data. It's synthetic stuff is closer, I think, so how are you challenging a sort of sorry answering a sceptic, when they come back saying, Well, you've just made this stuff up?

Mark Cochrane 20:27
It's a good question. I mean that, you know, there is the risk of sort of garbage in garbage out with these things, like, if you build an agent, an audience agent, it's only as good as the data that you've built it with, and so you obviously you would the better insight that you compare it with, the more accurate the output will be. So we're not creating synthetic inputs. So to build an agent like this, we will use as much, you know, we'll get hold of as much data as possible about a particular audience, so whether that's from syndicated panel data, so you GI or you gov, GWI, etc, or if there's first party research about that particular audience, or even from transactional data, any sort of combination of those things will allow us to, you know, usually in tabular format, upload that into into an instance of a large language model. So in Gemini or chat GPT, you provide it with that audience data, you describe it as knowledge and then a really detailed set of instructions that you want that that that model to use, essentially telling it to behave as the personification of the consumer in the data that we've provided, and what and what it does is it then essentially becomes, at, you know, at its simplest, a shortcut to more complex analysis of tabular data. It just gives you kind of plain English insights that it's it's pulled from that, but obviously that's just the starting point. You can then, you know, use it in much more complex and much more sophisticated ways, but ultimately, they're built on on data, the data the same data that we've been using to write these pen portraits with, it's just analysing it in a different way, with the power to go out onto the internet and find more information and more context. So it's kind of high quality data with broader context gives you, you know, and that obviously an understanding of how of language gives us, you know, incredible outputs. So it's, I'd say to any sceptics, it's not, it's not really doing... it's not using anything we've not had previously. It's not inventing. It's not inventing things about it, about consumers. It's simply articulating it in a much smarter, much more interactive way, dynamically, kind of in real time.

Tom Ollerton 22:53
It's interesting the duality and struggle that I'm having with AI at the moment. And you can laugh at me if you want, for this, but there's that sense that you're talking to a cognizant being like and especially with the voice mode on something like chat GPT, it's like the most recent version that subtleties are so human-like that sometimes you're thinking like, am I talking to someone? No, I know I'm not. But like, there's so many of those things happening that, you it's easy to fall into that, you know, I've used it for for therapy, for myself, you know, and they and you have to start believing it's a person, or otherwise, you're just, like, just too dissonant, right? But then your point there is lovely say, like, it's just information. It's just surfaced in a different way. It's all of that first party data, or, you know, You Gov stuff, combined with some internet searching by a bot, basically, but served up in a conversational style. But it's so weird for me that those two things are both true. It's like a conversation with a real person, and yet it's just data visualisation, really, in some senses, but in a written or audio form. So I really wish I had a powerful question at the end. I don't, but you've reminded me that there's that there's an odd duality going on here, that it's not it's not sent in. It doesn't understand what you're saying. There is no cognition there at all. It's a parlour trick. It is just resurfacing information that already exists.

Mark Cochrane 24:13
I think you... that's it's always important when you know you're using these agents to recognise that you know they they do not replace talking to actual people and but what they are really brilliant for is using in advance, for example, if you're going to be running some focus groups with with actual with actual customers, do some synthetic research first. Ask the synthetic version of the consumers to tell you what questions would be most interesting to ask them. And you can use these in a huge number of ways. They're just, I certainly as a strategist, where a lot of my you know lot, a lot of the trick with strategy is trying to do, you know, a bunch of really divergent thinking, think really broadly and and then refine it down to something really simple. And actually, I. Those two those two things, those two tasks, divergent and convergent thinking, are what AI is brilliant at. You can ask it to give you 100 examples of something, and it will come up with some awful ideas, as well as some ones that you that are really great that you haven't thought of before. But you need to go through and filter it, and then you can ask it so once you've, once you've done some of the filtering yourself, you can then ask it, you know, to do further convergent thinking or build on those ideas. And so, you know, I think it's, it continues to be a tool, and I think it will always be a tool that you can't replace human in it, whether that's the actual human customers or actual human experts, but it, but it certainly for me, is and the use of these agents is really transforming how now, how we build communication strategies in the future, for sure.

Tom Ollerton 25:43
Well, as always, with this podcast, we could do another hour, but we haven't got another minute, so we're going to have to leave it there. So if someone wants to get in touch with you to talk about artificial audience agents or anything else that we've discussed, where is a good place to do that and what makes a message that you will actually respond to?

Mark Cochrane 26:02
Best place find me on LinkedIn. Mark Cochrane on LinkedIn. And I think a great message would be, you know, remember thinking fast and thinking slowly. So, you know, I'm a lazy, emotional being. So maybe think of something that stands out and connects, rather than that, sticking with the rational, perhaps.

Tom Ollerton 26:21
Brilliant Mark. Thank you so much for your time.

Mark Cochrane 26:23
Thanks very much.

Subscribe to our YouTube channel

Listen to our ‘Shiny New Object’ Podcast on Apple Podcasts, Spotify and Soundcloud.