Marketing Superintelligence Unlocked - Amin Mrini & Henry Innis

0:00Hey, welcome to the CMO Whisper Show. I'm your host, Steve Olensky. Part marketing practitioner, part ad agency veteran, part journalist. I was a writer for Forbes for 10 years. I've had so many insightful conversations over the years with business leaders, to athletes, to celebrities, to, of course, CMOs. The only difference now is instead of sharing those insights through written form, I'm doing it this way. My guests today are two leaders who are reshaping how marketing makes decisions in the age of AI. Henry, or his close friends, and pretty much everybody calls him Hen Innes, is the co-founder and CEO of Mutanex.

0:40The company behind Theseus AI, I hope I said that right, and you'll tell me, it's built on a simple belief that measurement should be continuous, transparent, and tied to real business decisions, not quarterly reporting cycles and guesswork decks. My other guest is Amin Marini, I hope I said that right, who is the Chief Digital Officer for Lions and Wark, where he connects the dots between strategy, media, and the fastest-moving part of modern marketing, the moment where brand, product, and purchase collapse into a single action. Together, they are building what they are calling, quote, marketing super intelligence, combining global benchmarks with real-time modeling to give the C-suite something marketing has been chasing for decades, a shared truth.

1:35Wow, what a novel concept. I cannot wait to get into this. It's the kind finance believes the boardroom uses and marketing can act on tomorrow morning, not in six months. So let's start right at the highest altitude, the state of AI and measurement today. And now we're going to dig into what marketing super intelligence unlocks and how it changes the relationship between marketing, finance, and decision-making inside business. Hen and Amin, welcome to the CMO Whisperer Show. Thank you for having us, Steve. Thanks very much, Ryan. How do I do on the pronunciations? Is it Marini, Amin? Yeah, it's a tough one.

2:15What's the English pronunciation of a Moroccan name when my first language is French? So, look, anything works. It's as close as you're going to get anyway. So, good. I got it. And, Hen, is it Thesis AI? Thesis. Thesis. Damn, I was close. We have an obsession with Greek mythology in our business. So every single project is codenamed after Greek mythology internally. So code base is littered with those sorts of names. It's a quirk of our engineering department, really. Got it. All right. So let's just jump in with the state of AI and measurement. And I got to start with responsible AI. And I'll start with you, Hen.

3:09Look, everybody's embracing AI quickly. That's not breaking news, but not always thoughtfully, right? So first question is, what does responsible and rigorous AI in marketing actually look like today? Well, I think two things. I think we need to talk about the reliability of AI generally. So I think most of the AI tools generally are quite unconstrained in how they can use tooling and core functions and all of that kind of fun technical stuff, which means that effectively they've got a very high probability of making stuff up and being you know less reliable than they should be in many situations and I think the second thing as well is making sure that AI is you know effect like being used in a way that is effective in amplifying human capability I think is fundamentally quite important because I think that a lot of the time it's tempting to replace humans in the loop and what you get is a lot of slop from that.

4:17I think on the reliability point though, you know, I don't think we should be afraid of hallucinations necessarily. I mean, I was talking to someone in my organization, which was a bit of a light bulb moment for me, who said, if we just talk about hallucinations as the percentage that people make stuff up, he said, I think a 4 % rate of people making stuff up is better than most meetings that we walk into with humans. I thought it was actually quite a well-articulated point because, you know, you don't really characterize hallucinations like that. We look at more like a software metric. But, yeah, so I think those are the two areas I think are very important.

4:58I think the future that ICAI being applied to is a huge magnifier for productivity and a huge magnifier for getting the economy back to a productive state when we have really been in, if you look at productivity over the past five years, been in a huge slump on productivity for the past five years. And so I think that's where we're going to get huge advantages from that. So, I mean, to kind of pivot from that, like, again, that term responsible AI, right? Yeah. I mean, do you envision a world where we're going to have these rules and regulations in place at any point? Look, I don't know about rules and regulations.

5:37I think there's probably a role for governments to step in at some point, and hopefully we're not too far from that. When you talk about responsible, you can talk about transparency, accountability, making sure we're not – I mean, the models are biased, right? In nature, by nature, are they trained? You talk about data security, I think there's a – for me, the take that I have on responsibility is responsibility to society and staff when it comes to uh ai adoption uh at the company level responsibility to tell the story and the value story of ai in a way that is not scaremongering all right so i hear too much on cost efficiencies and cost savings and yes we can't be naive to the economic reality of of of the sector uh but to me the and and henry touched on that it's the multiplier effect that matters It's augmenting humans.

6:32That matters. And I don't think we've been doing ourselves a favor in the media, primarily, in the way that we have been talking about AI. So to me, this notion of responsibilities, the notion is that in the sense of leaders being responsible in the value story and the way they position the role of AI in the future of society and companies. And then there's also, I'm curious, what do you guys, what do you both think of as we put our consumer hats on? right and and not just even a b2c but a b2b but mostly i would think of consumers um and i'm always asking cmos about what do they think consumers uh what because we're all consumers right it's hard for us to separate because we're in this world but as a consumer to me there's there's got to be that sense of responsibility if i'm if i'm engaging with a brand right to to know where that line is drawn between real and what's not real?

7:29Well, I think that how much of the internet has been real for some time, I think is, you know, I don't think anyone would characterize the internet as a misinformation-free place, free AI. So I think we were already trending in that direction somewhat. I think it's just accelerating some of the natural human foibles. But, you know, I think it's very, very important for brands and ecosystems to be building a world where we have a principled approach to AI. You know, one is, to Amin's point, a responsible story around AI that's about the amplification of productivity and the building of industries rather than the narrative which seems to be put out there, which is tearing industries down, which I don't think makes much sense.

8:16And I don't think there's actually that much evidence for. The second kind of component is, well, okay, how do we start to also ensure that AI is trained in a responsible and ethical manner? One of the things that I think was really important, for example, in Australia, where I come from, is that Australia has been to protect copyright of people producing content and things like that, which I think is a really, really important step. like, you know, AI companies should not be free to train and, you know, nick copyright and things like that off publishers just because it's in the interest of training their models, for example.

8:54And I think we need to rethink some of those elements. Like, I mean, when you talk about the government stepping in, I mean, I think the government really needs to step in on copyright massively around AI. But I think the benefits of AI is that we are starting to get to a position where general intelligence can interact and act as an interface between humans and quite technical infrastructure that we just didn't have before. Like, you know, the fact that designers and product people and people with ideas can now interface with technical systems, the fact that business people who don't know SQL can now interface with quite, you know, technical model specs and things like that and databases, that's a force multiplier across so many industries.

9:38And I think it's going to be a huge breaker of silos to some degree in its current state. I think the business that has the best ideas for me about AI at the moment is Anthropic because they're focusing so much more on the code to English interface as opposed to anything else versus, you know, content generation or image generation. Like I question why some companies are going so hard to create highly photorealistic AI that doesn't really feel like it can go in a really great direction to me in many respects. Yeah, it's anything but realistic in a lot of ways. I want to pivot to measurement. And I mean, look, marketing measurement, we all know, and it's called it, the measurement has lagged behind.

10:25I'm sure we could all tell stories. But my first question, Amit, is why do you think it fell so far behind? I think the blind spot for me on measurement, I think it's fallen behind giving marketers an accurate picture of effectiveness because of over-reliance on short-term metrics. I think we've been obsessed with what's easily measurable. I think we've been obsessed with what's immediately visible. And while nothing about the fundamentals has changed, right? So tech, digital media, social media, now AI, the reality is that the fundamentals of marketing cannot change and they will not change. But I think we've forgotten that.

11:07And so when you layer on top the rapid and complex evolution of the landscape, fragmentation of audiences, data, channels, and so on, and then losing sight of what matters and how the fundamentals and how marketing works, I think it gives you a very dangerous cocktail. So I see way too many brands in the doom loop of optimizing for what's easy to measure as opposed to measuring what truly impacts business. So to me, there's a giant gap in there and an obsession over what is easily measurable that's probably at the heart of the longest of issues that the sector is facing right now in measurement.

11:48But Henry's probably a better place to end on that. Yeah, and I was going to ask you about the always-on modeling, right? That's really become non-negotiable now. Yeah, so I think there's a few things. I think we are living in a world where information has sped up and with the speeding up of information, I think there has been a speed up in the cycles. Like there used to be that old rule of like, you know, you have eight years of good times and one's two years of recession. It more feels like two years of good times and then a quarter of recession to today. Like, and I think that's a function of information cycle speeding up.

12:30And I think that's why, why many marketers now and businesses feel like we're kind of living in a constant state of turbulence to some degree. Yeah. And I think that is what is driving a demand for, well, we need to understand how to navigate turbulence more quickly. You know, if I'm, you know, driving on a straight road for, you know, 200 miles, then, you know, normally I don't need a GPS. However, if I'm driving down, you know, a very complex suburban suburb with, you know, 50 different paths and things like that, and there's lots and lots of different things I can do and lots and lots of different choices I can take, well, then I start to actually need a GPS, something that's helping to nudge me and tell me where to go and that I can kind of input my destination and all kind of go, have you thought about going here, here and here?

13:20And so I think that's been a fundamental shift from the principles. I think it's as a why isn't the measurement landscape? kept up. I think two things. One, it's really hard to organize the data quickly and at scale, and I don't think that's been anybody's fault. I just think that the organization of data from getting it organized, modeled, and then drawing an insight out of that has taken an incredibly long time. That process naturally has taken a very, very long time prior to AI. I don't think AI fundamentally changes the principles of modeling and measurement and those aspects. But I certainly think it can help compress those timelines to help give that GPS.

14:07I think as a second point to that, I think one of the other reasons I think measurement is incredible is, and I'm saying this as an MMM vendor, I think MMM is too inward looking. You know, I think most businesses, when they look only at the MMM results, will only look inside their business. They will never be able to source external context. And whenever you talk to someone about what should I do about a decision, you know, say we see a low ROI on a channel, how do I interpret that? I should look at both my own data, but then I actually should be trying to also piece together what else exists out there in the world to understand where could I go.

14:51And I think that question of where could I go and piecing together data about where could I go, it's been incredibly difficult to step together. Like I just go back to my time doing this at agencies. You know, I pulled together the MMM results that would be, you know, a week to kind of get that together. I then sense check the narrative with everybody. Maybe that's a week to two weeks. Then I'd probably do some desktop research and start to pull together what are the relevant kind of pieces of content, information about effectiveness I can find out there in the market. Call that another week and then integrate those stories.

15:28Call that another week. Now, for me to get to an answer to a question, that's a six-week time lag to answer one question. Now, I'm betting that the average CMO probably gets asked 10 questions a week by their other C-suite executives. And so being in that position where it's costing six weeks of internal time to answer question by question means that, A, you start answering less questions because you've got to be selective about the ones that you do answer and invest time in. and B, you often take shortcuts on the questions. And so, you know, you don't answer the questions in a fulsome way. So you might just look at the MMM data and things like that.

16:11And I think that's why it's lagged behind because it's only had a very internal and, you know, I think MMM can sometimes have a myopic view, frankly. It doesn't have enough of the world context to really deliver that broader picture of effectiveness and that broader context which will put the numbers in context. And so I think that's critically where I think we've not just lagged behind, but lagged in potential of what could be done. You mentioned the six-week thing, and we all know, at least in the U.S., and I assume it's relatively similar in other parts of the world, the quote-unquote average length of a CMO is somewhere between three, three and a half years.

16:51That time goes by really quickly. And when you're talking six weeks per question in true time, right, that's a great analogy that there's just not time to do all these things. Correct. You know, if I'm looking at six weeks per question, average CMO tenure is around 42 weeks. That means fundamentally they can probably answer 28 strategic questions in their tenure, right? pretty tough. Like if you're in a job where you have hundreds and hundreds of decisions to make, and you can only answer 28 decisions credibly with data without overloading your team and your agencies and things like that, I think no wonder tenure is hard to retain.

17:33Like if in a major business, I could only answer 28 strategic questions throughout my tenure, I'm going to find it very, very hard to be responsive. Contrast that with what the capabilities that finance teams naturally have with how well organized their data, how much internal muscle that they have around answering these questions quickly. And that's why the CFO has power. CFO has power fundamentally because they have an ability to answer and respond to questions with data quickly. And that capability is embedded into their organization. I think CMOs have, from their measurement toolkit, have lacked that ability to process that quickly.

18:14So I don't think that CMOs should be talking. Someone asked me, how would you KPI an analytics function? And I said, what's the cost of answering your question? That's the simple truth of how you should KPI your analytics function and marketing effectiveness. Because if you're KPIing it any other way, based on actually, did we not do these four deliverables or these four reports, you're not talking about the core unit of value that matters most, which is how many questions can we answer and what costs do we incur to answer them? Exactly. I mentioned in the intro that the work and MutantX are collaborating on what's called the marketing super intelligence.

18:56I mean, let's break that down in layman's terms. What does it really mean? So our mission at WALK is to power the industry by giving them the evidence, expertise and guidance to make marketers more effective. right in other words to use language that henry likes to help them make better growth decisions right so ultimately that's what marketing is about when i was talking about the fundamentals i think too often right now we're talking about outputs when marketing teams are here to deliver value and then not growth so nowadays uh if you're lucky enough to be uh to be a mute next client our data the walk data so the universe of best practice the uh the world winning campaigns all that advice, all that goal is available inside the MutantX platform, as in where investment decisions are made, right?

19:43So when you're talking about your information serving the purpose of making media dollars work harder and be more effective, powering decisions that unlock growth, marrying our data with MMM models and the AI layer that enables you to interrogate the data and compress the cycle between analysis and decision time, it is a dream combination for us to super serve our mission and partnering with Henry. And from your perspective, what does marketing super intelligence mean? The frustration for me as a user of these products is, you know, I remember, and sorry, I mean, I'm probably going to tell a personal story of how we came up with this, you know, because Amin and I had been, you know, friendly for some time.

20:30I'd been friendly with Paul as well, you know, just sharing ideas because we'd met a number of times. And We were sitting in a pub in New York and I showed Amin a first early iteration of our AI. And I remembered sickly Amin saying, we could power this with the world's best effectiveness content. And it was a light bulb moment because Walk has the world's best body of research and contribution, hands down, and effectiveness. There is no other place that you should be going. If you're a media planner, a researcher, like someone who's interested in effectiveness, you should be on walk. Like, and so to my mind, when you think about why is best practice so divorced at times from the reality of making decisions, it's because when we're looking at the data and those very granular data points that we're using to inform the decision, Marrying that exact data point up to that exact piece of content takes an incredible amount of time, right?

21:33And it's those weeks that I spoke about just before. So for me, the crazy thing here is that in like 90 seconds, you could scan your entire data set to answer a question in MMM. Then you could go in and you could literally scan Walk's content and things like that and understand, okay, this is the information that the model's telling me, but this is also what's out there to answer this question, the wider world of effectiveness as well, and bring those two things together like that. Like, to me, that was as someone who's done marketing effectiveness strategy, someone who's had to pull together these papers, like, that is the dream.

22:18Like that is literally the moment where I go from having to answer a question in two, three, four, five, six weeks to I can do it in sub 90 seconds. And if I can bring together that information in sub 90 seconds, guess what? We have more time to think about strategic planning than ever before. And marketers get empowered to spend less time trying to find the information and more time thinking about the information. And that to me is the mission critical thing. You know, marketers, we need to reclaim time. What do we want super intelligent tools to do? We want them to be able to navigate our questions, source the information for us, and reclaim our time to think again.

22:58And I think every CMO deserves to have the privilege of thinking again. They deserve to have the time to think about strategic decision-making again. They deserve to have the world's best effectiveness resources paired with the best marketing intelligence humanly possible and have it all linked to business outcomes. And we have achieved that today. And to me, that's just a pinch me moment. And sorry, I'm getting very excited here. No. The level of passion for effectiveness data is unmatched. Unmatched. Yeah. Yeah. But, you know, it is. The passion is coming through, guys. I love it. This is the Nirvana product that, to me, means that we can start to get back to that fundamental question of how do you make every marketer effective?

23:48By radically dropping the cost of answering effectiveness questions. And if we can radically do that, that is what we intend to do with this product, with these products more globally. And that's the problem to solve. And I think that's the problem that fundamentally will reorientate and transform how marketers can be effective in today's age. I can't even imagine, and I mean, you can just, or whoever, like what's possible when like global benchmarks meet real-time modeling? Like that's crazy. We're talking about shorter, faster, much better planning cycles, instant access to our global database, the ability to benchmark and marry MMM data with the best investment advice that is out there and validating each and every investment decision with both ROI modeling, Henry's smarts and the data science behind it, and our qualitative and quantitative information.

24:44I just, as the marketer in me and all the CMOs that I know personally, let alone know, like, they're good. Some are going to be skeptical. You guys have to know that. I think people can be skeptical all they want. I mean, I basically say to everybody, the litmus test of a product is, can you demo it? We were at our marketers and money conference on stage. It was in the press because they were all there in Australia. This is our Australian customer conference. I took the high risk, high reward approach of taking questions for the AI from the audience, which my engineering team begged us not to do.

25:25But I said, no, we're doing it anyway. And if it all goes to shit, it all goes to shit. And so we were able to do that. and it was pulling Walk's information through. It was doing it flawlessly. The other thing as well is that, you know, you're only as credible as the brands you work with. We have a number of major global brands now signed up to this. We are going to continue expanding and rolling it out. I just say to anybody, if you're sceptical, come get a demo because, you know, it's a pretty simple product. It's probably the easiest product to demo ever done. I think more broadly, though, I think in terms of skepticism around AI, I think there's always a hesitance to embrace new technologies and tools.

26:10An analogy I always give is that, you know, everyone hesitates to embrace the fancy new ad format. But it's the people who embrace the new ad formats the fastest who typically see the best payoff from them because nobody else is using them. And so I think that, you know, what's going to really set organizations apart in the AI age is agility. And it's going to be their ability to answer things quickly. It's going to be their ability to move fast in this new world. It's going to be their ability to kind of take on new information. And so I think companies are going to get, companies have always been to some degree being penalized for how quickly they move.

26:50That's why DTC brands did so well against legacy brands. But I think AI is going to amplify the penalty of companies being slow massively. There's always going to be skepticism, but I think it's also always worth interrogating. We also have a view as well that the number one way to solve skepticism is validation. To prove it, yeah. So I think every model should have open source governance, for example, and be compared to the open source MMM models to make sure that the MMMs are running solidly. And conversely as well, I think that the other best way to kind of validate whether or not these things are working is, you know, we should be able to trace every element of what the LLM was calling, how it sourced the data, how it sourced the information.

27:41One of the incredible things about working with the WALK team has been the traceability of the content that the API gives. So, you know, Walk have an amazing API. It's actually some of the, you know, my team kind of came into this thinking that this may be a more difficult project and the infrastructure on the Walk side was bloody good in terms of the API. And what that meant was our ability to kind of provide traceability, to provide kind of logs on how the information was sourced and how information was found both for ourselves, but also on the walk API integrated in is really high. And so I think that traceability and transparency is going to be key in the age of AI as well and making sure that these tools are very, very transparent and can be validated very effectively.

28:30You mentioned governance, and I want to expand on it in a second, but I want to go to a mean rule first about the boardroom conversations and how this changes the way marketing and finance make decisions together. Because, you know, my audience are primarily CMOs and marketers, and they may be curious, how does that, you know, if I'm sitting across my CFO, how does this help me? Well, I mean, fix is attribution, number one. I think the disconnect between marketing and finance is that they do not talk the same language and one has got a really easy thing to do to attribute a dollar to an action and the other one is going to work a bit harder.

29:13So working on ROI, Working on metrics that make sense to finance teams, being able to solve attribution, being able to talk in terms of breaking down performance, being able to run scenarios. And the beauty of these models is that you're going to find out whether what you're doing is working or not. And so the trust divide that I think is born out of the proliferation of measurement that ultimately didn't resonate because the KPIs didn't connect to the real world and to finance are coming to an end with tools like Henry's and augmented massively by sort of effectiveness, golden advice that we can marry now together with the data.

29:57To think that marketers and CFOs can actually speak the same language, it's mind-blowing to someone who's been doing this for 25-plus years. For me, I think one of the really interesting things is that, you know, I spoke about this in the conference quite recently, is that why has finance and marketing been divorced and kind of been on that, I suppose, that path of divorce? I believe it's because in 2008, 2009, you had the launch of Last Click Metrics. And what Last Click Metrics did is they fundamentally disconnected the media's interpretation of return from the P &L's interpretation of return.