103. Why commonly accepted "science" could cost your brand millions with Felipe Thomaz

Felipe Thomaz
00:00:00

We are too afraid to tell our clients to do something different because they want to hear reach and that's all they want to see. If all I need to do is have this reach always on, talk to everybody.

If that's sufficient, all the other things will be true and we've proven that to be wrong. So that idea is just false.

Dom Hawes
00:00:18

Right.

Felipe Thomaz
00:00:18

So it's just immediately disproven. It's rubbish. It's not going to get you the thing that you think it's going to get you.

This campaign, the single campaign, is going to create awareness with the right people. It's going to create associations with those other people and I'm going to get sales and it's going to go viral.

It's actually going to turn me into an iconic brand and then Santa Claus is going to come back with my entire budget back. Like, it's just not going to happen.

Dom Hawes
00:00:52

If you're a marketer or you are interested in marketing effectiveness, this could be both the most controversial but also the most important hour you spend on anything this month. No, this year. Because Today I have Dr. Felipe Thomaz.

He is the associate professor professor of marketing at the said business school in Oxford, among other things. He's going to introduce himself in a minute and he has literally just published a research paper in one of the best marketing journals in the world.

This is a peer reviewed paper on why reach on its own is not enough. That's what this podcast is about.

He will talk to you in detail about the findings of his research and why some of the other academic institutions published papers might not be all that. That's coming up right now.

Felipe Thomaz
00:01:44

Thanks for having me. Excited to be here.

Dom Hawes
00:01:45

We're talking about something really important today. We're talking about science, how it's properly.

Felipe Thomaz
00:01:50

Done and how it can help us.

Dom Hawes
00:01:51

Because everyone always says marketing is a mixture between art and science. We talk about the art a lot. We don't always talk about the science. And often when we talk about the science, it's not actually very scientific.

Felipe Thomaz
00:02:01

I would say, I believe with that split, right, there's aspect of judgment, there's aspect of risk taking, which is that arts side of can I stretch things or the creativity of it. Although there's a science behind the creativity as well.

But I think science, when it's done well, when you do a good job of it, gives you a map so you can understand why things are happening and you can predict what your actions are going to do. That's my role.

Dom Hawes
00:02:26

And today we're talking about effectiveness specifically.

Felipe Thomaz
00:02:28

Right yeah, at times it will verge into some the efficiency side of it as well. But yeah, two sides of that coin and how to, to spend our budget, how to get media to do their job.

Dom Hawes
00:02:43

The reason that I was so keen to get you in the studio today is I saw some hullabaloo online about a paper that you are about to produce and Ritson, the Godfather in movie terms of marketing, made some reasonably strong comments too. We might come to those later. But I saw your paper and instantly loved it and so I wanted to get you here.

Maybe you could start by introducing you, telling us a little bit about you and then introducing the paper and why you wanted to look into that particular area.

Felipe Thomaz
00:03:16

I'm a professor of marketing at Oxford. The foundation of my work has usually been on marketing finance interface.

So I'm very, very interested in making the best use of our capital and justifying marketing activity in terms of how does this help the business. So I tend to operate a lot at company level analysis. I understudy or pay less attention at behavioral effects. So what is the consumer doing?

What is psychology? So I tend to operate mostly on the economic, mathematics competitive landscape, which I'm really passionate about.

And that's kind of where I started and what motivates a lot of my thinking, it's. I was in industry before academia and I saw a lot of waste. Okay. And in parts, that's how I got into academia as well.

It's just like that, you know, we spent 5 million on this campaign that had zero lift. Oh no. Right. So like how do I correct that? That brought me to the side of, of the science fence and I've been working on that ever since.

Publishing in mathematics, publishing in marketing, publishing in conservation science, where it's needed. But so it's this weird combination of skill sets to try to address some really, really important questions.

Dom Hawes
00:04:33

And what was the question, your current paper, what was the question that you had originally that made you want to research this area?

Felipe Thomaz
00:04:40

This started from a series of other papers and works that explore the combination of channels. So if you just go back to the very, very initial thought of why we create campaigns, right.

Is this idea that there's some combinatorial power that if I put two things together they might work better. And if you look at the literature as it exists, everybody had this ability to measure two channels, right.

So there's a lot of the papers were what happens if I turn on Facebook and TV at the same time or TV and Twitter and I'm have joint consumption and it's this Duality and this dyadic relationships dominated the literature that just said I can measure two channels and they interact, they have cross effects, all of those. And over time the literature started dismissing this idea of complementarity and said everything is just synergies.

The simplifying things of it just works well together. And that was a worry because it started to move away from the full picture of how channels are actually interacting and combining.

And because we have a series of partners in industry that actually had access to really good data.

So in this project we had Kantar with tons of data on cross media effects and lift studies we decided to start addressing of what if you didn't oversimplify what does the combination of channels look like? Is there an optimum, is there recommendations we can make?

Can we fight against this channels and channel combinations and integrated marketing comms as pure synergy and have more fine grained controls. That was the impetus for the start of this.

Dom Hawes
00:06:24

And talk to me about the research because you looked at a lot of campaigns with Kanto.

Felipe Thomaz
00:06:28

Yeah. So in this one I had access to a bit over a thousand campaigns. I think it's like 1081 or something like that. And these are very large campaigns.

So as you can imagine it's a sort of company that. So these are very well known global brands.

I can't say who their clients are and all those things but as you can imagine some brands that would spend multi million in media buy alone for these campaigns and would be buying then additional services to understand allocation of budgets and those things. So it's. These are heavy lifters, very heavy advertisers that we have. I think the average campaign spent about $12 million for that run.

So it's, it's pretty significant investment and we're trying to understand then how much are you getting from that investment? Is it proportionally, is it well allocated and is there a better way to allocate and can we start understanding how the channels combine?

Dom Hawes
00:07:23

So and that's, I mean that's pretty fundamental stuff for marketers because it, because as you, as you say your interest is in where marketing and finance reach and asset allocation is one of the biggest challenges of every cmo.

Felipe Thomaz
00:07:35

Yep.

Dom Hawes
00:07:35

Understanding how to spend that money better. It's not quite going to answer the wanna make a principle a paradox. Sorry the one but it's going to come some way.

Felipe Thomaz
00:07:43

Yeah, but so that's the stuff that we all like and excuse me for injecting myself into the seat of the practitioners as well but like that's. Everybody struggles with this. Right. It's spend it or lose it. Everybody's looking at marketing as a cost center. So there's no expectation of return.

So it's just toss money away, which lowers the overall value of the function and the CMO in practice. So it's like this really weird, endogenous, complicated system where we are undervalued because we don't understand our stuff.

And that's kind of that problem. If you don't understand what you're doing with it, then you can't ask for more budget. Right. So I've.

Back in the day when I was in practice, I went to finance and I said, what's my required rate of return on my budget? Right. You've given. Personally, myself, as Felipe, I had 5 million to spend a year on my brand.

Dom Hawes
00:08:36

Yeah.

Felipe Thomaz
00:08:37

And I wouldn't need approval, you know, if I needed more that required approval. But I could toss away $5 million. What's my required rate of return? How much do you need to see this back? Because the company has a cost of capital.

Let's say they had a 5% cost to borrow that money. I expect that I'd need to show more than 5% ROI. Right. Obviously, they look at me as like, you're a marketing guy.

What the hell are you doing in my office? Get out. Right. Like, it's just they didn't understand why marketing wanted to have that information.

Dom Hawes
00:09:06

Yeah.

Felipe Thomaz
00:09:07

Right now, this is now a couple decades ago, but still, that shouldn't be a controversial question.

Dom Hawes
00:09:13

Yeah.

Felipe Thomaz
00:09:14

And I should then just say, look, every dollar, every pound, ever, whatever you give me, I give X back. And through this, in the short term, into the long term, into this reduction of risk, exposure, all of these things I'm accomplishing by my spend.

But right now, it's not managed that way. Or actually a lot of it is managed that way.

But the way people talk about it and if you just listened to the wisdom of the crowd, you would think that this is science fiction.

Dom Hawes
00:09:42

What I thought was really interesting also, when you and I met before this recording, I wasn't aware that at Oxford there is this thing called the Future Marketing Initiative, which is where academia and commerce kind of come together to look at the future of marketing. Talk to me about that initiative a little bit and your role in it.

Felipe Thomaz
00:09:59

I'm now co director of the Oxford Future Marketing Initiative. It predates me in Oxford by one year. So it was created by my colleague Andrew Stephen, and it's one of the best mechanisms in Oxford.

It's one of the main reasons I Came to Oxford and stayed in Oxford. That might sound weird, but it is like it's a fundamental drive to research.

One of the biggest threats to, I think marketing researchers anywhere, especially from an academic background or an effort threat to academics, is that we can just sit in a room by ourselves, read journals, come up with questions, often just by reading the back of the paper that says future research questions and answering those questions and moving on with our lives.

If I just did that, I might generate ideas and solutions that have absolutely zero value to you or they're going to be very incremental, very minor, and don't address the reality of lived practice of marketing as it is today.

The Oxford Future Marketing initiative is this interface with practice and we have a lot of really good companies that serve as our council essentially.

So I won't name them all, but this is the Google's a partner Meta L'Oreal, Diageo, Kantar, hence the joint partnership on this project, WPP, et cetera, et cetera. Like, there's an amazing set of companies where leadership comes in and we have these private conversations around what can they not solve.

They are all full of immensely bright people. They don't lack for data scientists, they don't lack for data, they don't lack for capabilities.

But there are things that are wicked problems, things that they can't solve because they have a narrow view of the marketplace.

Dom Hawes
00:11:49

Yeah.

Felipe Thomaz
00:11:49

And we sit in the middle as a non commercial partner with everybody having the understanding that we are going to publish whatever the truth is and they have to accept the risk. And that explains, you know, the difficulty of a partner joining.

Because if I have something bad to say about this part of the industry, they're going to have to deal with the consequence. Right.

It's then a triangulation of visibility of agencies, platforms, brand strategy that give rise to the questions and then the ability to actually answer because they have the data or they pose the question says, look, nobody knows how to do marketing. Mix models with creative embedded in a practical good way. Let's solve that. How do I do White space forecast? Well, let's do that. Right.

So they provide the challenges and as long as it moves the science forward and it's not just pure consultancy, but it also has a commercial value, therefore it has a relevance to practice, then we have a good project.

Dom Hawes
00:12:54

Because today I think what we've called this podcast reach is not enough.

And I think one of the dragons we're going to slay is this sort of widely held belief that success in advertising specifically is about Reaching as wide as an audience as possible. And the paper you're publishing shows that it's not actually just reach at all.

Can you explain to me maybe why just aiming for maximum reach isn't the most effective strategy?

Felipe Thomaz
00:13:21

A little bit of background of how this started, because as you, as we started our conversation, this wasn't a part of the project. Right. So the initial question was how do channels work together?

Dom Hawes
00:13:32

Yeah.

Felipe Thomaz
00:13:33

And not reach. Allocation, breadth of targeting, like none of that was part of the core question.

But in the process of peer review, the other professors that are challenging all the findings came to a question which was prove to us that this is relevant to practice.

Dom Hawes
00:13:56

Okay.

Felipe Thomaz
00:13:56

Right. They're like, prove to me that they don't already do this.

So it's like, can you show me in concrete, rigorous terms that this is not just widely known, widely done, and you're wasting our time.

So that's where that study on reach came, which was what you alluded to is a stochastic frontier analysis, which is if you abstract away the idea of media channels as investment opportunities.

So I'm saying I'm going to invest in tv, I'm going to invest a little bit into Facebook, I'm going to invest a little bit into Twitter or X or whatever its name now. And I'm going to invest into all these various channels and I expect some sort of return from that investment by ways of lift or whatever.

But if I look across all of those campaigns, those are not stable lifts, there's variance. So it's an investment at risk. Right.

So in that way you can abstract this idea of a financial instrument that you invest into it and it's going to come with some variance. If you do that, then you have a portfolio campaign as a portfolio. Right.

So I can organize my various instruments to maximize my lift and minimize my risk exposure so I get more money with the lowest risk of that money not coming back. And then you just have a risk trade off preference of I will accept more risk for a higher chance of pay off.

And that's the frontier, the curve that you would accept a fair trade off. So I did the analysis on the sort of outcome that brands were getting. One that commercial lift, brand metric, lift or just reach.

Are they actually getting the reach that they want with low risk? And are they getting the sales, the motivation, the salience, all of those things without risk.

And what we saw is like let's say 80% of all campaigns, it's a bit higher than that. But let's say just, just to play with round numbers, 80% of all campaigns have as high a reach as they can get, right?

So clearly shows that there is a intentional effort for reach to be optimized, maximized. But if you look at the other graph and if you look at the other efficiency trade off on lifts, none of them actually approach the optimum.

And there's actually a normal distribution well away from the optimum. Meaning you kind of have a little bit of a random allocation of outcomes despite maximizing reach.

So you have these different sort of consequences of everybody's doing what they're told, which is go maximize reach, but they're not realizing the actual goal because I don't think anybody here or anywhere has reaches the financial KPI, right? Like they actually want some, some monetary gain and they might have some intermediate intermediary measures in terms of brand metrics.

So the goal is not just to talk to people, the goal is to get money in your pocket. And what we're seeing is that you're maximizing the reach, but you're not getting the so what on the end. And that was the huge concern.

There is quite a few reasons why that can start happening. Part of that is you by overcasting the net, right? By going too broad, what you get is leakage, which is an older term in integrated marketing comms.

So we're going back to the 60s here, right? Imagine early days of media buy and you had to buy the whole distribution of a magazine, right?

You put your ad in that magazine and you had no choice but to ride that wave. And you're going to show that to everybody that bought the magazine.

Now you don't want to talk to everybody that buys the magazine, but for whatever reason you want to talk to a subset, so you've overpaid for that exposure to talk to people that are never going to matter to you, to talk to the people that matter. And that inefficiency, that leakage is part of that loss inefficiency. You're just talking to people that don't care.

You're talking to people that don't matter. You're talking to people that will never matter.

The instances that I've talked to anybody where there's a strategic reason to talk to non consumers is small to none, right?

And one good example of like those counterintuitive that I have found interesting is like heavy luxury brands where the value of the brand is for other people to have the perception so they're not going to buy it, but they're going to be jealous and I need them to be jealous for you to see the value in the brand. Therefore I need to talk to somebody that has a non commercial. It's a strategic but non commercial value.

Yeah, on that immediate but that network effect is a value.

The other arguments that I'm sure you bring up is this like, well, just in case, I don't know when you're going to be in the diaper market, maybe never, maybe you will or maybe you're going to recommend it to a friend. So I'm going to do it anyhow. That's what leads to this.

Dom Hawes
00:19:04

Yeah. So two things strike me immediately in business marketing.

We have the same effects in luxury brands in that we need to influence people who are probably not going to be consuming the product, particularly in things like software where the risk is very high and everyone wants to buy safe.

So you find software companies marketing messages about trust and longevity and robustness and redundancy to people who are not customers because they're part of an endorsing decision. And the other point is we are being force fed this thing called the 95.5rule at the moment. As though 5% of people are in market at any one time.

95% of people are not. And that's being used and applied to, well, you must do brand marketing to the 95%. You must go whole market in what you're doing.

And I think when I read the paper, what it sold for me is some of the received wisdom that we have that you have to go whole market, that it's all reach, it's all about exposure. And that's just patently not true, is it?

Felipe Thomaz
00:20:05

There's one immediate thing this paper does is we labeled this as in this reach sufficiency hypothesis, right? So it says if all I need to do is have this reach always on talk to everybody, if that's sufficient, all the other things will be true.

And we've proven that to be wrong. So that hypothesis does not hold. So that idea is just false. Right. So it's just immediately disproven. It's rubbish.

It's not going to get you the thing that you think it's going to get you. It has a lot of other implications along with it.

But I do worry about those broad generalizations and percentages and allocations that are perhaps untrue and not deceiving, but dangerous for people to take at face value and ignore other things that we've known, but we try to simplify away from, right?

So never in those conversations do I hear available addressable markets, time out of category and purchase journeys and purchase patterns in your category.

Because if you're an auto salesman or auto brand owner and your time out of category is, I don't know, three to five years, to six years, to 10 years, it's very different from me in fast moving goods that has 16 choices a year or more. Right. So like those timing frameworks will lead to completely different strategies.

And if you're going through like, oh, I must reach the whole available market.

So my definition of growth is this whole available market share of the entire category, then you fall into a mass market strategy which dictates lower price premiums, which dictates your entire cost structure in some of your company, which is true for one company that is going to be a cost competitor on undifferentiated goods, but it's not true for anybody else.

So you have this really weird decomposition of this complex strategy into something that is true for one person and says go ahead and do this thing because you worked for them.

Dom Hawes
00:22:13

And this is where the more controversial side, controversial to some of the paper you're publishing now comes out because it's, it's basically disproving something that is a very widely held belief in the market about how brands grow.

Felipe Thomaz
00:22:29

Right.

Dom Hawes
00:22:30

What was it Ritson said?

Felipe Thomaz
00:22:33

I believe he said, if you're going to take a shot at the king, don't miss.

Dom Hawes
00:22:36

Okay.

Felipe Thomaz
00:22:38

Which is very interesting. I haven't seen a king in science yet.

Dom Hawes
00:22:41

But he's softening, he's changing. Over the last few weeks we were saying just before we started recording that he seems to be coming back towards differentiation.

Felipe Thomaz
00:22:49

Yeah.

I saw two or three videos of him talking about how marketers had forgotten about differentiation and it matters which is off the back of the some of the work I've shown a can like maybe two years back now that actually differentiation is a massive driver of abnormal stock returns. And salience and awareness and distinctiveness of assets don't feature like.

So a differentiation drives about 17% of abnormal stock returns, distinctiveness drives zero.

Dom Hawes
00:23:21

That's pretty cut and dried.

Felipe Thomaz
00:23:23

It's what it is, right. You just go look. Because if you look, what is an abnormal stock return? Right.

So getting more into the finance side, like those are unexpected growth or if you can look at downside, could be unexpected decrease as well. And that's why I've in trying to explain why that's the case is I don't see.

And with apologies because I'm now driving back to that lower level of analysis, not to company, but psychology and consumer decision making, growth and Change in penetration. Like why does somebody. How do you acquire a new customer? Right. How do you get the next person? Is winning a moment of choice. Right.

So you have to convince somebody to either pick your thing or to go from a thing that they picked before to a new thing that they're going to pick. So you're either poaching somebody or you're the point of entry into the category or whatever it might be. And you have to win on that basis.

If distinctiveness in that sense is the alternative brand elements. So you're looking low level construction of the brand which is the visual representation of the brand.

And this is again a very old world on the shelf mentality. Still most purchases tend to be physical, non digital. But gets harder when you just have an empty search bar and you have to enter something. Right.

But I challenge you to find a time that you bought something because the box was blue. Right. Did that element actually convince you to buy this relative to something else? What it does do is it defends your market position from.

From a purchase that has already happened, which is a function of brand. Right.

Brand is supposed to win the moment of choice and keep people locked into that choice so that when they're back into their next purchase occasion, they don't seep into a competitor, they just recognize something that they had at home and they just rebuy. And you don't want them to revisit that moment of choice.

So distinctiveness in terms of the brand is a protective defensive mechanism that maintains your share so it doesn't seep into the other one. Therefore it will never show an abnormal growth because it's in the maintenance expectation.

Dom Hawes
00:25:34

Yeah.

Felipe Thomaz
00:25:35

It's in the perpetuity assumption. It is not in the growth. Differentiation which convinces somebody to reevaluate their choices is what drives choice.

It's what gets you the next customer.

Dom Hawes
00:25:46

Distinctiveness is a defensive mechanism, effectively we're saying to protect market share or to protect choice. But differentiation is what leads to that choice in the first place.

Felipe Thomaz
00:25:58

One drives growth and the other one keeps the growth that you've attained and hopefully maintains your existing user base. Otherwise you're just constantly fighting for higher and higher churn inside of that.

So if you talk about customer retention and those the functionality of brand to maintain somebody is how do you keep them in your wall like within your.

Dom Hawes
00:26:19

Yeah.

Felipe Thomaz
00:26:19

So defending your share. That's why like people thought I was being kind of cheeky when I said like this whole idea is not how you grow but how you stay big.

It's not just A reframing, it's just like it's a defensive strategy. It doesn't work. If you're a new entrant in the market and you have unique, just distinctive assets, nobody cares.

You have to have another reason for choice. After you have the choice made, then you can start playing off of that mechanism so that maintains your installed user base.

I don't see any logic that would make somebody go, yeah, I'll pick the other one.

The other thing that does happen, which is an old result, which is why people start using market share to answer every sort of thing is if you know nothing, like it's a social proof sort of mechanism. Right? If you know nothing about something.

So if you go into a country and you have a headache and you've never been in that country, you don't speak the language, you can actually even read the Alphabet, let's say, right? But somebody says, here's all the payments, good luck. And one product has 10 facings and the other product has two facings. Which one do you pick?

Dom Hawes
00:27:26

10 facings.

Felipe Thomaz
00:27:27

Exactly. Right. This is a known thing. So aspect of facing category management, distribution, that hasn't gone out the window.

Yeah, but if you're managing your whole multi billion whatever brand off of just a social proof effect, I feel bad for you.

Dom Hawes
00:27:43

I might buy the two facings. If it was priced high enough and someone could tell me it was that because it was more effective, I might do price then comes into play as well.

Felipe Thomaz
00:27:50

Right, but then you're using like heuristics for search, right? Search is costly. We're lazy human beings that we just want to use.

If a million other people have made this choice, then it must be a good choice for some reason. And if everything is equal otherwise then this actually gives me enough information to make a choice.

But it's like that effect hasn't gone out the window, right? Like, yeah, that's true. That's why it's really hard to get distribution and you know, go get facings. I worked in industries where facings mattered.

Every time we had a new product, something had to go out. We didn't get another one for free. And we were a massive brand.

Dom Hawes
00:28:25

I used to run a distributor and getting distribution is not easy. Right? Getting people to list and categorize for you is not easy.

Felipe Thomaz
00:28:32

So that's one of my other fun things. Right, so what's. Especially with the ongoing wisdom and oh, just go get distribution.

All right, so tell me, in your experience, what drove the choice of who got distribution? Why would you give somebody distribution it.

Dom Hawes
00:28:49

Was really complicated, actually. It's about their own reach. It's about margin structures, it's about capacity. It's about. Actually, it's quite a.

It's often a financial decision because you're.

Felipe Thomaz
00:29:00

Carrying a lot of risk as a distributor, you're carrying inventory risk and all of those issues. But really, is that, will it sell?

Dom Hawes
00:29:08

Have they got the ability to deliver me volume?

Felipe Thomaz
00:29:10

Your distribution capability is a function of your ability to sell that product.

Dom Hawes
00:29:15

Yeah.

Felipe Thomaz
00:29:15

So they will only distribute. They'll only give you that physical distribution if you already have sales.

Dom Hawes
00:29:21

Yes.

Felipe Thomaz
00:29:22

And the more you sell, the more distribution you're going to get. It's not an independent gift that if you're nothing, they're just going to like, oh, it's an easily achievable thing. Just go get more of it.

It's this. It's an indogenous system. It all depends on the other things.

And if you just believe, I'm going to go and get this without the reason, the good marketing, often if it's a new entrant, you need to convince them that you're going to drive people to the shelf or you're going to drive them to the store or there's a reason for me to carry you, you're not going to get that asset. It's an important relationship with channel distributors and all those. But you can't just manufacture out of thin air. It's dependent on the sale.

So it's a weird sort of belief and recommendation, which inversely, to sell more, you just need to sell more. It's what it often comes back to. If you sell more, you will sell more tanks. Like, yeah, sure, it's tautological, it's circular.

Yeah, there's inertia, there's all of those things. But tell me what actually I need to change to sell more. Don't just tell me to sell more because if I sell more, I will sell more. Got it? Okay.

Let's say, yeah, the more you sell, the more you sell. I think we're on the same page. What do I need to do in my context, in my industry, in my life stage to achieve that marginal sale?

And I think that is completely lacking in most of the conversation right now because it's been reduced to things like this reach argument, which as we've shown, is not true.

Dom Hawes
00:30:53

Yeah. Okay. So talking of selling more, your paper and your research found that different channels have their own unique strengths.

I wonder maybe you could give some examples of how channels impact consumers differently to kind of give some ideas to our listeners of where they might depending on what they want to achieve, what they might want to be considering.

Felipe Thomaz
00:31:15

There's a little bit of triangulation here. So there's only so far one paper can go.

So I've actually been around and I've shared results from a second paper that's not published yet that dives into specifics. So just to warn folks that might go into this paper and said, oh, where exactly is the functionality of Facebook?

You will see some of that here, but not a complete breakdown of cross category regional strengths and relevances for each specific channel that is forthcoming, let's put it that way. So I have a really nice project with Wavemaker at the moment and they gave me a million customer journeys. Wow. Yeah.

So you know data nerd in the candy store, right. Like it's. So imagine a million purchases with all of 72 touch points. Wow.

Dom Hawes
00:32:11

Okay.

Felipe Thomaz
00:32:12

Right.

So like all of that richness to identify essentially the power of each one of those touch points to influence and change that navigation towards purchase from change consideration and then change purchase inside of that ecosystem across categories. So I can start to say actually in beauty this channel works better than this too for purchase intent. But that's not in this paper.

Maybe that's a future conversation.

Dom Hawes
00:32:46

Sorry, but the fourth thing you just said for beauty in this channel, because so, and I'm rowing back slightly, I should have covered this probably already, but. But you talk about archetypes.

Felipe Thomaz
00:32:55

Yes.

Dom Hawes
00:32:55

Can you explain the concept of archetypes and maybe what they are?

Felipe Thomaz
00:32:58

An archetype is a bit like a recipe. Okay, right. So it's a combination of here in the papers, it's a combination of probabilities.

So it's like what's the likelihood that your campaign should use tv? And everybody always use TV on every sort of campaign.

So it's going to have a somewhat high probability that you should use it, but it'll show actually what's the likelihood then that point of sale will help you in your category. What's likely that radio actually should feature and you're going to see either very high probabilities or very low probabilities.

So they start giving you kind of the sense of indication of strength or frequency in which those channels have featured and combinations that have been put in a way that drives a given KPI. So that's part of the mapping exercise that we did.

And I think it's the most useful to think of in terms of a mapping of there's a huge surface of variations that you can say more tv less TV blend of old media, new media and different platforms and different kind of consumption occasions. All of those things are in this complex space of all the possible alternatives that you can use. Right.

So if you think with 11 channels, I think that we had all the possible combinations of on and off. If they were just pure switches, it's like several thousand. It's a couple thousand of different ways that you can flip those switches on or off.

This just gives you a few clusters that we identified that lead to best performance.

Dom Hawes
00:34:39

Is that for specific outcomes or for.

Felipe Thomaz
00:34:42

Specific problems or both for specific brand mindset metrics.

So what we had is essentially salience divided in aided and unaided sort of awareness because different brands and different business problems required those into different amounts. Right. So if you're on the shelf in a very physical setting, then recognition actually is very valuable to you.

But if you're a very digital first, you don't have a visual cue most of the time when you're shopping, you need to actually elicit that from top of mind without recognition. So the other KPI is more valuable for you. So we try to split on those association in terms of can you connect the brand to concepts.

So again, mental networks and those aspects of brand and motivation or intent to buy, purchase intent. So it's like those are sort of the KPIs that we kept in mind and that's mapping of different.

What was interesting then as a result, which we set out to prove was would one combination actually meet all of those requirements?

Dom Hawes
00:35:54

Right.

Felipe Thomaz
00:35:55

So that was the first thing like, so the initial name of this project was actually Silver Bullet or no silver Bullet.

Dom Hawes
00:36:00

Is there an optimal mix that everybody should use?

Felipe Thomaz
00:36:03

Because if imagine that, that would be ideal world, right? I've said X percent this, Y percent that, Z percent this other thing and everybody's good would have been delightful, but it's not true.

Dom Hawes
00:36:14

Yeah.

Felipe Thomaz
00:36:14

Okay, so there's different combinations help different people to achieve different things. And that's the wildest sort of thing.

It's scary, I guess, if you are a planner on some sense, because there's no, oh, I just hit the same button over and over again and I'm going to get. There's a best for whatever you're trying to do. There's just no universal best.

Regardless of who you are and what you're trying to do, there's not a dominant strategy. The issue that we found that was on the scary side is we talk about dominant and dominated strategies.

So a dominant is like always best dominated is never. There's always something Better.

Dom Hawes
00:36:58

Yeah. Okay. Right.

Felipe Thomaz
00:36:59

So he mentions like completely dominated strategy is you do this and there was always a better choice.

Dom Hawes
00:37:05

Okay.

Felipe Thomaz
00:37:07

Right. So imagine like that's, that's a hurting moment. We see a lot of those in, in the wild. In the wild and in our. Really.

Dom Hawes
00:37:15

Okay.

Felipe Thomaz
00:37:15

And that's kind of scary because like I said, I have massive multimillion campaigns.

Dom Hawes
00:37:21

Yeah.

Felipe Thomaz
00:37:22

Using dominated strategies. Meaning that there was always money left on the table.

Dom Hawes
00:37:28

Huge wasted, huge wastage.

Felipe Thomaz
00:37:30

Right.

And that connects to our earlier conversation of they might have been using what in the papers described as this incorrect lay theory so that that reach is sufficient. So if they believed they probably optimized to scale but didn't hit any of their targets.

So they could have done a number of things to improve their performance and maintain scale. But they didn't. So they're using that dominated strategy. There are contextual best.

So we found a really interesting specific archetype that works really well for CPGs across all criteria. So if you are a CPG, there's actually a really, really combination that will drive salience, association motivation.

Everybody else, multiple alternatives are going to be adding those combinations. So just using lessons from. Oh, but you know, I don't know, pick, pick a brand that everybody loves like oh, Coca Cola does this.

I'm going to do what they do. Oops. No. The other thing just to throw it out there because it's very common is hedging your bet gets you towards dominated.

Dom Hawes
00:38:45

Dominated.

Felipe Thomaz
00:38:45

So if you are like, oh, you know, but I'm just going to start throwing stuff on. I'm going to do all the things is the lowest performing architecture.

Dom Hawes
00:38:55

Okay.

Felipe Thomaz
00:38:56

So if you just go, oh, I have my campaign. It's well structured. It's, it's close to an archetype that drives my outcome.

But I'm going to actually, I have extra budget, I'm going to throw this other things on and you move away from the archetype. You lost money.

Dom Hawes
00:39:10

How does a marketer, how does a CMO make sure that their strategy is dominant, not dominated? What's the, what's the magic?

Felipe Thomaz
00:39:20

So one thing is for them to recognize. So our advantage here and the reason why we're able to do this is that unlike a brand, we had data for every brand and across category.

So we have variants that brands would never see.

Dom Hawes
00:39:40

Okay.

Felipe Thomaz
00:39:41

Right. So they had blinders on of just historical information of we always do this and experimentation is difficult for brands and costly. Right.

Go ask CMO to turn off tv.

Dom Hawes
00:39:52

Yeah.

Felipe Thomaz
00:39:53

And see how quickly you get run out the door. Right. Like nobody's going to turn that off for any reason. But it's. So it's really hard for you to prove what the right levels and those things are.

So for them they'll be looking at their own personal, blindfolded, sort of historical data and see validation because they're MMMs are saying the same thing because they're just in that inertia mode. So the reason why we can say different things is, but because we had variants that nobody else had access to.

So that's one of the reasons why my view might be different from their internal view. Right. Because of those sort of blinders for them is a few things, recognition of their own context.

So they need to understand their category and location in a sense and strategy of the brand, which is life stage kind of the path dependencies. What have you done up to this point that would prevent you from doing something else or requires you to keep doing?

So you have to respect those constraints, right? So you're going to be constrained and you're going to have budget constraints like everybody else in the planet. Right.

You have to operate within those, but within those boundaries. That's going to give you a narrow range within what these, what our model suggests. Right.

And it's going to be within a category specific domain and requirements, whatever that might be. So if it's a tech company, it's going to have a different set of weights or a different best goalpost or guideline.

And then once you have your boundaries, is identifying your for a specific outing into the marketplace, a campaign, something that you're going to turn on. What is the goal? Right. So what are you trying to do with this? Are you just trying to generate short term immediate sales?

Are you trying to create some brand connections, associations, create differentiation?

Are you at a stage that you actually do need a heavy investment in salience because you're entering the market or you're, you're trying to compete with somebody that's really loud at the moment and you're trying to do those. So you have to understand what are you trying to do. So it's conditional on goal, right?

So two conditions is who am I and what are my operating conditions? What's my market, what am I trying to accomplish?

Given those there is a optimal allocation and that's going to be a specific archetype, might be number four, number six, number seven, whatever. We didn't give them names, we're not trying to be cute.

But there's, there's a map and says go to this best allocation that's going to give you a set percentages of channel, channels that you can use and kind of a suggested budgetary allocation. As you recognize by media, there's discontinuities, TV is expensive, you need to put a certain amount of money.

You're never going to be quite as clean as what we can be because we have this pure mathematical model off of the data where everything is beautifully continuous and you don't have thresholds of purchase. So you're going to be close to whatever that is. Right.

Within that point, you would be at your theoretical best outcomes for whatever it is that you're trying to do. But it might not be feasible and it might not scale well enough. Right. But right now you're just in the channel function construction.

Now I said, okay, how do I scale this bad boy? Right. So I need to talk. This is when Reach enters the room. Okay, okay.

So he says, I know what I need to do, I know what I need to turn on to get that thing done. Now how do I get the most of my intended audience to see this?

And it might be a slightly different allocation within those channels that gives you the right audience construction. So just recognize that you're going to be moving away from a optimal outcome to have a optimum scaled outcome.

So you're moving away from the best lift or best return to get that lift multiplied by the largest number of people doesn't mean you go to a whole different neighborhood to create a whole different audience. You're taking steps away, seeing that you're. Think of it that your margin per person is going down, but your number of people is going up. Right.

So you're traveling away in that local neighborhood. You're not switching planets to a different sort of strategy until you find a happy dual optimization.

I'm still very close to my optimum lift with my precise audience construction.

Dom Hawes
00:44:29

So in this case, the optimal lift is the archetype, for example, that's the reference data. And then the real world comes in.

So you then have to start making some changes to the archetype based on your ability to spend, buy and or physically deliver.

Felipe Thomaz
00:44:46

Yeah, what can I afford? Where can I buy this? Do I have deals? Do I have this? Those are constraints that.

Dom Hawes
00:44:52

And by following the archetype, statistically certainly or scientifically, they're looking at more effective outcomes.

Felipe Thomaz
00:44:58

Yes. So as long as you avoid the other error, which is I'm going to do all the outcomes at once.

Dom Hawes
00:45:05

Yeah, right.

Felipe Thomaz
00:45:06

Because that's the other thing. You know what I'm going to do this, this campaign the single campaign is going to create awareness with the right people.

It's going to create associations with those other people and I'm going to get sales and it's going to go viral and then it's actually going to turn me into an iconic brand.

Dom Hawes
00:45:23

And by the way, I want to increase my margin so it's going to support my price rise.

Felipe Thomaz
00:45:26

Yeah, yeah. And you know, and then Santa Claus is going to come back with my entire budget back. Like it's just not going to happen. Right.

But we see that over and over because they over measure, they measure everything. They try to get everything to shift with once.

Dom Hawes
00:45:38

Yeah.

Felipe Thomaz
00:45:38

And it's just like look, that again gets you to the. It's the similar dominated strategy. If you need to do everything on one outing, it's going to be very low returns.

So you can think of it as if this were a financial product. If you are the always on everywhere, doing everything all at once. It's kind of like the bonds of financial market.

It's probably going to do something for you and it's probably going to be a 1% return or lift and it's going to be very safe because you will hit somebody that cares. You will get some clicks, you will get some sales stuff is going to shift, but in tiny amounts. Right.

And then the person with a better strategy that has a portfolio of activities of right now we're going to do this and then we're going to do that. Is the person that's going to be hitting the 10% lifts, 15, 60% lifts that we see in some of the optimizations. Right. Like these are large leaps.

Like in one of the things that my colleagues did sort of modeled what if statements with the real Data is like 750% gain in outcomes from reorganizing. That's on the extreme side. And this was like with aided awareness I think he calculated. So it's like massive shifts.

Like some of it will come in with risk and that might be for a specific category and on a specific thing. So it's not like everybody gets that sort of outcomes.

But what we did observe in that initial motivation was that yes, the average lift that those reach maximized campaigns were getting was about a 1%, 1 1/2 percent. And that's, that's why I keep saying like that's really sad.

Like you know, if that's, if you go to market, spend £10 million and you see mindset metrics shift 1%, you should cry inside. Right. Like this is not, this is not what I wanted for me.

And if you get those people that disagree, I think what we see often is those well differentiated, targeted brands that are small so they don't do the overreach because they don't have the budget to overreach. They actually do better and they grow faster and they start eating away at the share of this big behemoths that are doing the safe bet.

And like, I don't understand how they're doing so much with so little.

Dom Hawes
00:48:12

Yeah, very, very targeted, very focused. Yeah, yeah.

What are some of the common mistakes that you've seen when, when it comes to channel selection among marketers other than like dialing them all up?

Felipe Thomaz
00:48:22

So hedging is common. Doing.

So hedging, meaning I'll turn on everything because I can or I'll toss, I have extra budget and I must spend everything or I'm going to lose my budget. So I'm going to start turning on things that won't help me in the least. My personal experience that led me to academia was something like that.

It was almost 100% leakage. They found a specific channel that had Almost, I think 2% of the user base was of value to my brand, but they still invested 5 million to that.

I was just like, wow, nobody wants to use this product. Why are we advertising here? So it's like, so those are significant. There's a over reliance on historical data.

So this is really hard to overcome because you're doing data driven marketing, right? It's like, oh, I used yesterday's values and I'm going to get these different things, I'm going to grow. It's very common to look at.

Well, that's how it was done, that's how it's going to get done.

And we always do this because we have those relationships, we have those things and it's the amount of inertia that I see in that is really, really high.

And then the, frankly, the other thing is not listening or ignoring advice in industry to our conversation on the recommendations and lay theories that exist in the world right now, or even theories that are suggested to be scientific.

People in media buy, people in agencies repeatedly come back and said, we know this is not how it works, but our clients, we are too afraid to tell our clients to do something different because they want to hear, reach and that's all they want to see.

So there's this perpetuity of bad advice from people that do know better because they have a commercial need to satisfy their clients requests which just want to see that number go up and at which point you don't need a media buyer because you just do a simple if then statement. Just buy more eyeballs.

Dom Hawes
00:50:43

It's pretty scary that it is because.

Felipe Thomaz
00:50:45

Honestly, again my sampling is biased because my view is fairly well known as the at this point. So people that come to talk to me tend to be non believers of the traditional practice.

But they come to me and said, you know, I've known this for 10 years, I've known this for 20 years or whatever, like at least 10. Right. And but I can't tell this to my client because they must see reach go up and that's what I'm being measured on.

Dom Hawes
00:51:13

And we know specifically now, specifically scientifically that reach is not enough.

Felipe Thomaz
00:51:17

Yeah. According to a peer reviewed paper published in one of the best journals in the field in the Journal of Marketing.

Dom Hawes
00:51:25

How can people get hold of a copy of your paper then?

Felipe Thomaz
00:51:27

There's a couple places we've kept throughout the review process. The paper has been kept on SSRN Social Sciences Research Network. So that link should stay available throughout.

If you go to the Journal of Marketing website, they host accepted paper. So right now it will be under the accepted and then there's a full copy there.

Dom Hawes
00:51:52

Yep.

Felipe Thomaz
00:51:52

As we speak it might still be behind a paywall. But one more thing to be proud of. So Oxford bears the cost of destroying the paywall. So Oxford is amazing in that way.

They pay all of the fees for publishers to remove the paywall. So Oxford papers are always open science.

Dom Hawes
00:52:16

Wow.

Felipe Thomaz
00:52:16

So if you see a subscription requirement when you go try to get that paper, just wait a little bit because it should be free any day now.

Dom Hawes
00:52:27

My experience is journals can be really expensive. Oh yeah, that's good news. So we're going to put some links to both of those sites on the show Notes for this episode.

So if you're listening and you want to find it, look in the notes below and you'll find links to those papers.

Let's just talk very, very quickly before we end about practical application because we talked about this to start with and it's quite hard to make that leap between academic and science into action. But some people have been making their own efforts. You told me, talk to me about your experience recently.

Felipe Thomaz
00:53:00

That's actually has been pretty amazing as I think we were chatting privately before is the role of science and practice and that interaction because we're here to help.

But it's really hard to then for me to create a product off the back of this and I'm not going to create a dashboard that's on people to absorb and create. And take this model, everything is in print, you can just go and use it today. But there's that translation step that is difficult.

As we're discussing, the last question was can you inject your own context, your own situation and then once you have those, can you bound and get good predictions from this model, good outcomes? And yeah, recently I had a few email exchanges with a guy that got in touch with me that said, hey, really like this paper.

I want to make sure that I understand and how it relates to other stuff that we do. Here's how.

I've created my own kind of dashboard in Excel that uses these values to get recommendations of channel construction by archetype for whatever they're trying to do that was to their context, like in alcohol. And it was brilliant, really nice work.

Dom Hawes
00:54:16

That's great.

Felipe Thomaz
00:54:17

And it was just this nice interaction which he came to me just for validation and just like, am I understanding this right that I miss anything? And just a few email exchanges to get him to use it and you know, he went, read the paper, deployed it in his work and it's off and running.

And for us, I mean it's amazing to see because that's what you want. You want people to not just let this sit on the shelf and like, okay, I'll keep that in mind. But the money stays wasted, right?

We want to see this turn into money into people's pockets.

Dom Hawes
00:54:46

It must be amazing for you and your team as well, because getting the paper published is what a two was, a two year or more process?

Felipe Thomaz
00:54:52

It was, yeah.

Dom Hawes
00:54:53

And so now that people are taking it out into the wilds and actually creating like real business tools with it, it's fantastic.

Felipe Thomaz
00:54:59

Part of it is there's a lot in this paper that I'm incredibly proud of and it sets up a lot of other very important questions about media functionality and stuff that addresses on even pricing of media. Because if they all do different things and they're not fungible, the value of TV for you is different than the value of TV for me.

So why are we paying the same price? Right?

Like there's, there's amazing interesting things inside of it and those conversations already happening with channel owners and all those things. So I think we are in for a very interesting development and evolution of media buying organization and all of those things.

I like that it deflates ideas that are hurtful in the marketplace. So if we had this lay hypothesis of just do this, this simple thing, just if that guidance were true, this paper wouldn't exist.

Dom Hawes
00:55:57

Yeah.

Felipe Thomaz
00:55:58

So we were able to show that. Yeah. Just death, saturation, overreach, reach sufficiency is just absolutely bonkers wrong.

So I'm delighted that we can move past that now to better precision, better flight controls for what we do with what is about a trillion dollars a year of media buy. So I much rather have more than 1% return on that. And then.

Yeah, it's to see it then get into practice and live and breathe is true to what Oxford wants to do, which is not just academic for academic sake, but academic in service of the people that do it. And that's our job.

Dom Hawes
00:56:40

And it's going to spawn the next paper which I can't wait to see the study you're doing with Wavemaker, the journeys. I mean, so it's spawning other things that for someone like me, that's, that's gold dust.

Understanding which specifically which channels have which effects. Fantastic.

Felipe Thomaz
00:56:55

Yeah. And we've, we've already gotten amazing results.

I was recently in Sydney presenting some of that work early and we shown the first passes of it actually in can this year.

Dom Hawes
00:57:07

Okay.

Felipe Thomaz
00:57:07

Okay. I'm sure we'll be back in can with even more complete results.

But it's showing some worrying things like the value of word of mouth is declining for everybody.

Dom Hawes
00:57:19

Oh, really?

Felipe Thomaz
00:57:19

Yeah. And it's the most powerful marketing kind of conduit we have. Yeah.

We've lost about 15% of power in that channel from 50% likelihood of affecting somebody's choice to 35%. And that's across over three years. Across every category. It's terrifying.

Dom Hawes
00:57:37

Are you looking at why?

Felipe Thomaz
00:57:38

That's still an open question. But because it's happening all across. It points to societal change.

Dom Hawes
00:57:43

Yeah.

Felipe Thomaz
00:57:44

And it's not being replaced by influencers. It's not being impacted by celebrities. It's. It's on its own, decaying.

So people are talking and listening and taking advice from close family and friends and family. Like a lot less.

Dom Hawes
00:57:57

Wow.

Felipe Thomaz
00:57:58

Which is mind blowing.

Dom Hawes
00:57:59

Right?

Felipe Thomaz
00:58:00

Yeah. So it's, it's one of those terrifying ones. But there's other like super interesting things like to generate sales across every several categories.

Good old print does better than influencers, really. Right. So you just like things that we're tossing away because they're old suddenly still have huge amounts of relevance. Brand owned is huge.

Like brand owned has a 36% chance of influence somebody's purchase.

Dom Hawes
00:58:27

Wow.

Felipe Thomaz
00:58:29

Right. So just your own stuff that you have control. So like. So that's the sort of work that I love.

Dom Hawes
00:58:34

Yeah.

Felipe Thomaz
00:58:34

Because you can go, what should I do? Where should I put my hard money? Like, you know, nobody's.

Dom Hawes
00:58:38

Yeah, yeah.

Felipe Thomaz
00:58:39

Has infinite budgets but these things will actually lead to better functioning markets which makes me happy.

Dom Hawes
00:58:46

Well, there you go. Thank you very much indeed to Dr. Thomaz.

I have to say, extraordinary guy and I'm very much looking forward in a minute to popping out to lunch with Felipe to find out more about his work, about Oxford and about how they are trying advanced marketing science. Now, I hope you enjoyed the show. If you did, please let me know. I'd love it.

If you would subscribe or like the show, give us a thumbs up just below and we'll see you next week. If you've got any issues, complaints, queries or just want to talk to us about this content, you can reach me on LinkedIn.

My name is Dom Hawes and Dom H A W E S and I would be pleased to connect to you. See you next time.