Henry Innis - MMM & The Mutintex Growth Story - S5 Ep14

Speaker A: Hello. Uh, bonjour. Ni hao comestas. Welcome to Champagne Strategy. Listen to this episode if you dare, but you've been warned. There's no going back. We're here with Henry Innes, founder of Mutinex. Welcome to the podcast.

Speaker B: Thanks, man.

Speaker A: Good to be here. The first time we met, it was like a random LinkedIn connection or something. I remember being in Melbourne at the time. I think we met up at like, Crown or something. When you're there.

Speaker B: It was near Crown. It was. I remember it sickly. At the time we were working at Freshwater Place and there was this pub down the road that I'd gotten a whole bunch of people to sing. So my co founder is a guy called Matt Ferrugio.

Speaker A: Yes.

Speaker B: And I. I bought an entire bar around to sing the song Hallelujah. But instead of Hallelujah, it was mad faroogia. And so, yeah, so I saw you at that bar again and I think I was a little dusty as well, because we'd been out the night before, done that, and it's still on my Instagram. It's one of the, like, funniest moments

Speaker A: of my life, I think. We came to contact and then I just saw this rise, very consistent rise from this idea to a full fledgling, massive business. I just thought it'd be really interesting to hear from the horse's mouth where you started. And then I found from a lot of people, these businesses kind of go through these choke points or like bits where you kind of have to come across this massive problem, then have to change something quite fundamentally or overcome a really big problem. And so you get to the next phase and then sometimes you need to bring new people in or fire people and all those kind of changes that go along. And I think a lot of people think that's quite linear and just, you know, you hire more people and get bigger. But for me, it's in stages almost. So I just want to go from the beginning. How did you come up with the idea?

Speaker B: It was Matt and I who came up with the idea. We were both working at YNR at the time, which was a division of wpp. I think when you're around holding companies and things like that, one of the central questions that, uh. Because what holding companies are doing is they're kind of bringing together a lot of the capabilities that exist around a customer to try to create a complete view of execution around a customer. Right. And that's the central promise of those businesses and. And that's why they work in kind of global accounts as well. Quite a lot. But I, I think one of the challenges that we face there when coming into those sorts of groups, because that was a big part of our job, was, well, when you aggregate all the execution, the one thing the customer wants to know is what's changing in the overall roi. Are we getting more effective as a result? And I think like increasingly the conversation was moving from actually, are we able to cut cost out of the kind of, you know, the head hours models and things like that, the retainer models. And I think in 2017 was like the zenith of that procurement push. It's still there but it's, it's less so. Like, I think it's a lot more acknowledged on both sides that the race to the bottom has not generally been good for advertising. And so increasingly we were getting asked, well, does this new operating model deliver, uh, a more efficient bang for buck in our execution in market as it relates to influencing consumers? And I think that was a really interesting question. It was one that was solved at the time relatively archaically. So if you looked at market mixed modeling solutions around that period, there'd been a really large focus on, ah, kind of market research style solutions, stuff like that, or I mean actual market mix modeling studies. Right. Like mmm studies was still a big business then. Yeah, um, they have been around for about 30 years. Much bigger in the States than Australia at the time. But it was there. And I think what was fascinating about looking at those solutions is the media market was starting to move at pace. It was easier than ever to change your message, your price, how you distributed to the consumer. And that meant that the market dictated your effectiveness as much as you did.

Speaker A: Right.

Speaker B: And so from there we kind of thought that actually you needed this data to be coming through in relative real time. And so we set out to found that business and kind of, you know, we were regarding it as a mutiny against measurement.

Speaker A: Hence the name.

Speaker B: Yeah, And I thought we started in late 2018. We actually had two other people who were joining us at the time, Chuk Chang and a lady called Harriet Ray. Harriet, um, was ex PwC. Chuk Chang was former CEO of Omicom Media Group. Didn't work out for various reasons, but I think Matt and I still felt the idea was, was really, really valid and pushed through. I think it was very hard at the time. Like we were all slightly conflicted about what we wanted the business to be. Um, and you know, I think at the time you had, there was kind of two diverging schools of thought. One was let's build a SaaS business and go after that. And the other was let's build a SaaS enabled media agency and then maybe do a roll up of other media agencies. Which I understand the logic behind both, but I think it's very, very hard to build purity of thought, vision and clarity when you have multiple revenue lines built in at the start. I just think that's a very hard business to build. And for me it spoke to a lack of clarity amongst us as a group. Chewy kind of left the business and kind of went to non exec. Um, Matt and I had to buy the shares back which was a pretty hard and challenging thing at the time. We were all really happy with each other and he was great about it as well. So he wasn't, he wasn't difficult about it at all. He was, he was a gentleman about it, to be honest with you. And uh, and then we just continued to kind of focus on actually how do we start to build this thing. We didn't know what venture capital was at the time, which I think everyone would find quite ironic now, but we actually had no clue that was an option. And so I remember quite distinctly like as we were building a business out, we were trying to finance developers.

Speaker A: So at the time we're like 150k.

Speaker B: Yeah, I mean we definitely couldn't afford that. But you know, we found, we found some amazing engineering talent. One of whom did not work out, um, but the majority actually worked out with us for quite some time. One of our first engineers, um, he's still with us today as was our first designer who's now like he's a senior GTM lead in the business. Like he runs. He was the guy who set up a US office. He's now come back to set up another division in the business. Like he's gone from being a designer to being someone who's incredibly talented and has built his career with us, which has been great to watch. Um, and so I think, you know, we kind of just went through this period where we were just doing a lot of consulting work, um, probably working 18 hour days to try to finance the development of this product in the company. It was called Warchest at the time and originally we thought it would actually be a five product suite. We've just never moved on from the first product.

Speaker A: How many products do you have now though?

Speaker B: One.

Speaker A: Oh, one. Okay.

Speaker B: Yeah. Which is growth os. Then you've got the supporting product in data OS which cleans all the um, unstructured data for us. But that was really. Yeah, it was it was a really, really hard slog and I think it's, it's extremely humbling when you're kind of working with clients who, who believe in you a hell of a lot and believe in what you're doing a hell of a lot. To the point where we had some clients who, they were really helping us to kind of craft the vision of uh, Warchesterly. They were giving product feedback, you know, they effectively acted as co creators of the product.

Speaker A: That's the best way to do it though, I think like instead of people, because there's two schools of thought, you know, you created a sort of MVP and you sort of develop it with the feedback loop or you go and you spend years doing it like Figma did and then release the market. But you know, you run the risk of like that not being, reflecting the needs of the market. Right. If it's too detached.

Speaker B: Yeah. And I mean we certainly got really lucky. We had three very involved founding clients who are pretty instrumental in helping us get it right, gave us lots of feedback, iterated the product and they're actually all still with us today incredibly, which is, you know, hugely appreciative of that. And I think, you know, for us the vision is quite simple. You know, MMM was and is an archaic solution. If we can build an end to end solution where you can collect, organize, process and then present your data at scale to users in a way where they can use it to make the next bet to get the best answer um, to their next growth decision, that's a really clear proposition. I think the marketer uh, does not need more data, uh, does not need analysis products. They don't need more kind of exhaustion from the space. And I think if you talk to any MMM vendor in the space and you know, shout out to the Recast crew who, who I, I deeply respect and the analytic partners crew who I also deeply respect, I think they're both great operators. Tom Vladek from Recast showed me around to New York Steakhouses when I first arrived, which is very kind of him and he's an absolutely top individual. And I think any of us, when we talk we all kind of, we'll share the same kind of challenges, which is you want people using the data to make the next best decision, you want to be able to prove out growth, um, and you want to be able to link growth to what actually grows a business quite tangibly. And so I think we're pursuing that vision maniacally. We see ourselves as the solution to that by being an answers company, not A data company. And by structuring the MMM in software workflows that generate answers, whether that be through. You know, I think the future in the space is obviously language models, but also I think general workflow. You, uh, can do a lot with workflow in a SAS tool that goes beyond just presenting a, you know, a bar or something. Yeah. And I think that's really important to the future of the space.

Speaker A: Interesting. Yeah. Ah, like queries, real time answers.

Speaker B: Yeah, queries, real time answers, even down to user experience. Right. Describing things in terms of less. So this, uh, is a response curve. Because realistically if I'm to present a response curve to 90% of the market, what's the question which are actually trying to answer what's the right level of investment for my spend? And so, and there are two options within that. One is I can look at a response curve from a marginal ROI perspective or I can look at it from a yield perspective. Yeah. And they're two very different things. Yes, you can build a much more efficient business spending 1 million and making 5, but the actual business most brands want to build is like the 10 million making 40.

Speaker A: Right.

Speaker B: And so it's actually starting to describe things in the way that users want them and the way that they're seeking answers is very, very important, I think, to build patterns in the product where you're actually deeply thinking about what is this person trying to answer with these data points. And then how do I actually integrate the data points really well in a workflow that delivers that answer?

Speaker A: Yeah, because, um, we're having a discussion the other day, uh, about CFO speaking the language of the CFO or the executive, like non marketing executives. And obviously there's a big gulf sometimes between the two. I find people who probably can't communicate well with them are the ones who may be steeped in that, those volume or vanity metrics, sometimes they're lumped into, and then the others sort uh, of bridge that divide. They talk about incrementality, they talk about sort of diminishing returns and yields and that kind of thing. Very different kind of language. So do you find like this tool helps bridge that gap and pull clients towards that language?

Speaker B: Yeah, I think you have to think really deeply about user, uh, experience to make the most out of this space. Like, I think that it's very, very tempting to go, let's build the best model. And I genuinely think we are building the best model. And you know, I'm, I'm in big agreement with some of the other vendors that the right Way to assess that is things like holdout testing, parameter recovery, environmental error correlation is something we look at a lot, um, which is a very important aspect to stop variables being thrown in for no reason and creating error within the, uh, ROI plausibility. So analyzing the plausibility of ROIs against how they kind of look elsewhere, the convergence of models is something that's really important because I think not enough people look at their models from a stability perspective and kind of go, okay, if I rerun this model with this data, uh, or if I change input slightly, am I going to see a dramatic change in output? And that's a really important technical aspect of models to look at because, again, it's very easy to build unstable models. It's very hard to build stable ones. And the stable ones are the ones that you want to work with because they're able to use the data in a far more consistent way.

Speaker A: Yeah, M. You can rely on them a bit more as well. You trust them.

Speaker B: Correct. It doesn't make sense for me to change an input by 1% and then the outputs change by, you know, 50%. Right. Um, and I'm not saying that we haven't had that problem in the past. I mean, if I look in 22, 23, we probably battled the stability challenge a little bit as we started to hit scale. And that was kind of a key. You know, going back to the key challenges, I felt that was a really key one for us, was understanding that we had to build a lot more structure into our model and a lot more causal structure, like a causal graph into our model, essentially, to actually drive stability and an opinionated structure into it, which I think sits at the heart of most models. We believe that we're getting it right because we're testing for generalized dynamics, and we're testing. Okay. Does this feature work well across all? Um, rather than just experimenting with one, which is what most do. And I don't think you can.

Speaker A: You need critical mass to do that, I suppose.

Speaker B: Yeah. And I think that's why it was probably hard for us to solve the stability problem pre 22, but it was much easier to solve it at scale. And it's one of those ones when you see kind of new MMM vendors popping up. There are obviously new SaaS players coming up. I think there'll be new ones every few months as the media agencies, like a lot of the independent media agency CEOs, uh, they kind of look at this space and go, oh, I could build that. And they hire a data scientist to kind of think that they can do it. And that's, and that's totally cool. Where they will struggle is it's very easy to say, you know, you're building a neural network architecture when you have no clients. You know, it's a bit of a self defeating purpose because if you're building a neural network architecture that relies on number of samples, if you have no clients, where are your samples?

Speaker A: Exactly.

Speaker B: Um, it doesn't make sense. An oxymoron.

Speaker A: So maybe good for raising some money initially but.

Speaker B: Well, you know, I think, I think raising money is easy. I think building products and, and great businesses and great selling them and. Yeah, and just making sure they're used as well is very important. I mean the number one metric I look at in our business is usage of the product. That's the thing that tells me everything I need to know about my business. In my business I look at maus probably every second day across the customer base. I can tell you at any given time which customers need more education based on the MAUS and based on the usage patterns. And we're looking for four quite distinct usage patterns that tell us how well embedded the product is to a customer. So we're not really looking for an individual relationship with them as well. Which are like, I know a lot of vendors would just look purely at like, is the customer saying they're happy? We want to actually see usage of the product across all of our customers, big and small because that tells us that we're doing a really good job of like is this being used in a repeatable way that's extracting value for them?

Speaker A: Yeah.

Speaker B: And I think too many vendors, it's too tempting to sit there and go, are we sitting there having lots of good meetings and smiling versus am I seeing our average um, session time per month is something like 90 minutes per month out on average per year.

Speaker A: That's pretty good.

Speaker B: Yeah, it's pretty high considering a lot

Speaker A: of reporting is done like monthly anyway because you know, sometimes these campaigns run for a long period of time.

Speaker B: So that's. Yeah, so we think that's a pretty good indicator of like if you're getting kind of, you know, anywhere between four to 10 users in on average per client per month who are all spending on average 90 minutes across each of those users. We think it's a, probably a pretty good indicator that ah, that we're in the right area.

Speaker A: So you start with four people, bought one out. There's uh, three of you.

Speaker B: So Harry left as well at the top.

Speaker A: Okay, so back to two and Then you created the M, you'd call it. It got some key clients on board which are still there supporting business, which I really like because that kind of takes away a lot of the variability of income and kind of plan it. But what was the next sort of major problem? You sort of.

Speaker B: I think we went into product thing or, you know, how do we scale the product? I think, um, and for that period, the scalability challenges were basically two areas. I think data ingestion is quite clearly one of the central challenges. Everyone talks about this and data ingestion and standardization.

Speaker A: I think in general, like, not even just marketing. I've done BI dashboards before. It's always the issue.

Speaker B: Yeah. And I think for us, I think the conventional wisdom at the time, and you know, and we, we brought in some execs in that period. Um, you know, we brought in some pretty senior people who worked with us for a period to get to work through this phase. I think there were three things that we encountered. So I'm kind of going back to 21, 22 now and kind of towards the end of 22, we're really running up against three things. The first thing was we had to raise capital. Um, now if, if people probably don't remember, like, that's because we were running out. We needed to scale, um, to, to get more customers on boarded and we needed to scale the team to handle that onboarding. But we didn't necessarily have the capital to do that in April. So we had to make a decision, are we going to try to win more customers or kind of grow a bit slower? Um, and we, we took a judgment call to try to go out for capital and kept the business going hell for leather whilst we did it, which was. And I think at the time it's very easy to Forget, but around April 22, this was when we did our seed round. The markets collapsed.

Speaker A: Yeah.

Speaker B: Um, and so we were kind of had. We're kind of, you know, pretty up against the wall and you know, this is a long time ago. We've got a very stable and big business now. And so I think, you know, we're pretty lucky like that. But that was a really tough time that, that first round. And then we had a term sheet from a venture capital group that we were trying to negotiate and we hadn't signed it or anything like that, but, uh, but I thought we would get it done. And um, and it got pulled and that was, that was very, very tough and that was very, very, very hard. And I think it was a really humbling moment. And I think we, I think honestly, like, I, I, I had a bit of mental breakdown at the time. I was spending about, I spent about four hours crying. I think it was very, very hard.

Speaker A: But all that time and effort that goes into just negotiating and, you know,

Speaker B: and we'd met every vc.

Speaker A: Yeah.

Speaker B: We were coming off this period where we were just way too, like everyone back then was like, raising a stupid valuation. So we're all kind of coming in a bit cocky as well.

Speaker A: Yeah.

Speaker B: And so we pissed a lot of the VCs off. Like, you know, I, amazing VC at air Tree, Alicia McDonald, who's lovely, lovely person, great VC and you know, like, probably burned my relationship with her by just coming like in, in a bit too hot. Hard and hot.

Speaker A: Like Silicon Valley. Like, um, you know, have you seen that scene on the boardroom table, like the HBO show?

Speaker B: I, I can imagine it would have been.

Speaker A: This is like taking to the nth degree, like even more. But I can, I can see it working negotiations because you come across confident, like, you know, you're a growing company.

Speaker B: But then, yeah, to be honest, like, I just found that that style does not suit me. So, you know, we got, uh, and we very nearly got, we, we very nearly got a deal away with EVP as well at the time, but they kind of said no, just based on the early stage of the SAS renewal cycle.

Speaker A: Okay.

Speaker B: And that was tough. And, and so, you know, we basically, I went to a few people in the industry and you said, what do you think of this idea? Does it have legs? And universally, everybody spoke to our clients, they came back and said, look, we'll put the money in. So I had a bunch of people from the industry put 202.4 million, um, in at the time at a 10.5 pre money valuation. So, so that took us up to a 12.9 post valuation.

Speaker A: So there's like seed sort of funding still. Yeah, there's no official round kind of seed stage. Okay. Yeah, yeah. So then that solved that problem, at least temporarily.

Speaker B: And then the next phase was kind of starting to scale up and, and really solve two problems. One, we had to reduce a lot of the technical debt in the platform. Whilst the platform was very good at the time, we built it very quickly. And so we had to spend a lot of time slowing down to speed up. Having to change the kind of technical architecture took us some time and we needed to bring on very good engineers to do that, a lot of whom are still with us today.

Speaker A: Okay.

Speaker B: The second problem was solving the data Ingestion at scale. Um, and that was, that was a very hard problem because there were kind of two conventional schools of thought at the time that felt like they, to us, that they didn't, they didn't quite make sense. The first school of thought was, well, let's have someone who's really close with the customer, who consults with them, who'll kind of do all of the work bespoke for them and coach on their site. The problem with that is you essentially get bogged into a consulting project every time to collect the data. And so it actually doesn't create a repeatable process. And, and I think human to human generally doesn't create repeatable processes very easily just because there's too much variables to

Speaker A: go wrong, too many people and the politics and everything, navigating all that, it's pretty complex.

Speaker B: Yeah. Um, and so I think that's. That was very hard. And then the other kind of route that people kind of went was, well, okay, have every single one of our people make sure that they have a very clean warehouse on their side, you know, hold them accountable for the data and things like that, which inevitably. I think that's nice in theory. I'm not sure how well it works in practice. And I think that is still where the industry is today. You know, our perspective was quite different. We saw that data, in our opinion, was a workflow problem and all of the data looked slightly different, but was actually principally the same. And so the job of a data platform was really to collect and organize the data at scale and extract the patterned information off the user, uh, and organize it into an MMM M ready format so that the user would be able to kind of put unstructured data into the platform as they had it. Right.

Speaker A: So, you know, this ingestive sort of upload.

Speaker B: Everybody has a weekly sales report somewhere in their business, like everyone has a weekly pricing report somewhere.

Speaker A: Yeah.

Speaker B: The question is, is can you take that weekly thing that currently exists and get it uploaded it in? That was a really hard problem to solve because if you think about like the volume of schemas you have to look through, building a universal set of schemas that would work and generalize across an industry was a massive undertaking. And everyone told us we couldn't do

Speaker A: it because the variability and everyone's different systems and stuff.

Speaker B: Yeah. And I mean, we just took a contrarian view that we didn't think there was variability. We think the variability was surface level. We didn't think it was underlying, we didn't think there was Variability in the principles. And we were right. And so data OS became our, uh, primary mechanism for collecting and organizing data, uh, quality data at scale. And that was a really big bet at the time. Because if you think in 2022, if you were to say that the MMM M vendor will primarily focus on collecting unstructured data through a platform and then plugging it into a universal data schema, that was mad. Nobody would have.

Speaker A: How do you monetize that?

Speaker B: No one would have thought that through. No monetization plan, nothing. But we took the bet and that's paid off massively today. Because if you think about the average business, the reason that MMM costs so much with a large market research vendor, uh, or, you know, with many of our competitors as well, is that process. Is that process. Is that process. Organizing and maintaining that is really expensive months. And for them, any adjustments to the schema takes forever because you've got to

Speaker A: redo the whole thing.

Speaker B: Exactly. And so by solving that problem at scale, we've reduced the cost significantly to actually service the mmm. And I think that's been a very, very important part of trying to think about how do you open up the market? How do you think? User first. And I'm probably a much bigger proponent of, as you can tell, thinking user first, customer first. How do you understand deeply what the customer is thinking in every single mom. And, you know, in our landscape of customers, it's not just, you know, it's not just the customers that we sell to, you know, it's all the stakeholders and conglomerate of agents, the media agencies they work with. Yeah. And then, you know, we've always spent a lot of time, and I spend a lot of my time personally building, building the model. So the model's always been a really important part of what we do. I think, you know, I don't want to go too far into it, other than. And I know a lot of other people would go very transparent on this, but I do think the model governance is something we've invested a, uh, lot in. And model governance has started to translate into tr. And I mean trust at scale.

Speaker A: I have heard some worrying things in, you know, around this, like incentives that go on to make certain media platforms look better than they are. Yeah. So, I mean, commissions and stuff get floating around.

Speaker B: Yes. What do you mean in the MMM space?

Speaker A: Well, I have heard one or, uh, two. This Eric Weinstein quote, you know, if you're controlling objective measurement. Right. And there's trust around objective measurement, then you can't take the client for ride as much Anymore, you know what I mean? So there's some complex dynamics that sometimes.

Speaker B: Yeah, well, look, I mean, like, I mean I, I sit in this space, right. So I can kind of explain the dynamics pretty openly. Like, like there's obviously a convergence of interests, Right. With people who are receiving money from advertisers versus the advertisers themselves. And often, you know, groups who've been receiving the money have actually provided to some degree measurement services.

Speaker A: Right.

Speaker B: And that's been going on for ages. It's nothing new. Uh, I think that the publishers by and large pioneered a lot of measurement and I think they pioneered a lot of measurement to move funds in their favor. Right.

Speaker A: Well, why wouldn't you? It's business models.

Speaker B: Exactly. So the dynamics that I see at play and whether these universally apply is that increasingly publishers are asked to work and, and contribute to some degree to the MMM in two forms. Like if there's a large degree of spend with that publisher, like that publisher should be, uh, contributing to independent measurement. That's quite a common ask.

Speaker A: Yes.

Speaker B: Right. And so that's, that's a relationship I do see. But the measurement is kept as independent, which is important to note. That's what I have seen. Um, which is, which that, that's a relatively new development, but that's definitely one that's been happening behind, behind the scenes.

Speaker A: Yeah.

Speaker B: What you are seeing a lot though is like the publishers trying to push best in class innovation on mmm, which is kind of making sure that their channels are presented in the right way with the data that maximizes the visibility of their channel in a model.

Speaker A: So this is like a selling point now.

Speaker B: It's become, well, if you look at the open source.

Speaker A: Right.

Speaker B: What are the open source MMM models trying to do? Um, what they're trying to do is, you know, it's no shock to me that Robin can be calibrated a lot through GEO lift tests. Why? Because META is really, really good at running GEO lift tests. Yes.

Speaker A: Yeah.

Speaker B: So surprise, surprise, if you use Robin and you're calibrating a lot on geography, hey, you're going to be using a hell of a lot of META to calibrate that model really, really well.

Speaker A: Right.

Speaker B: And so there's nothing wrong with that on the face of it. It's just, I think it's really, really important to understand the incentives, why these open source models are being maintained. Um, and you know, but I do think that there has been a real push by all publishers to focus on independent effectiveness measurement, particularly as the publishers are learning more and more that they are slight, somewhat dependent on each other, um, to perform in the boardroom. If everyone. If one. If a marketer goes all in on one channel alone, chances are the marketing performance will tank and the budget for everybody will be lower. So there's actually an overall incentive, I

Speaker A: think, to kind of get along a bit. Yeah.

Speaker B: To start to get along a little bit more, particularly as the market has started to come down.

Speaker A: Yes.

Speaker B: And so I think that's one of the things that publishers have gotten a little bit savvier on lately.

Speaker A: So a bit less combative around their competition and a bit more.

Speaker B: They all want to talk a lot more about synergistic effects of. You've seen. You've seen Think Box come out with a lot of that. That research.

Speaker A: Yeah.

Speaker B: You know, you've seen, you know.

Speaker A: Well, this makes sense. I mean, who is just only consuming one channel? Like, no one. Everyone's consuming lots of different mediums all the time. So, like, kind of makes sense. It's like, to pretend that you are, like, somehow better than everybody else and exclusive. The thing is unrealistic.

Speaker B: Totally. And I think that that's like, you know, one of the. One of the things that, um. You know, I think that that has been really, really good about the measurement conversations lately is that, you know, you are seeing the publishers lean in a lot more to that sort of conversation. Now. Are there kind of, you know, other arrangements and things like that? I don't know. Mucinex has never taken one of those that has not been disclosed to client. So for anything that we take, whether it be agencies who refer us to Deal, which is about 10% of our business, roughly, I think there's a. There's a conception out there that a large part of our business is actually built through agency. That's actually untrue. It's about 10% and about another 5% comes through kind of what we would call publisher referrals. And in both cases, you have the publisher referrals, you know, you general. They generally are helping to kind of recommend.

Speaker A: Yeah.

Speaker B: Um, and then probably when the client

Speaker A: goes, hey, we need an mmm. We're talking about it. Who do you know in the market? And they'll be like, oh, we know Mutinex.

Speaker B: Yeah. And they'll generally put up multiple options, too.

Speaker A: Yeah.

Speaker B: Yeah.

Speaker A: Well, you kind of have to.

Speaker B: There's multiple options that go up.

Speaker A: Yeah.

Speaker B: And then there's. And then on the other side of it, you know, on the agencies, it's often like, you know, there'll be fees to, uh, get the data and stuff like that from them and stuff like that that are disclosed to the client. That's, that's a very common practice.

Speaker A: Yeah. I think 85 is direct sales.

Speaker B: 85 of our business is direct. Yeah.

Speaker A: So outbound making relationships.

Speaker B: It's actually mostly, it's mostly inbound and it's mostly from customer referrals.

Speaker A: Yeah. Okay. So people CMO goes, I use that. Uh, and they're talking to another CMO

Speaker B: and they go, hey, yeah, we think so. We think so.

Speaker A: That's certainly how it works for these kind of.

Speaker B: Yeah, like we, we think it's mostly that. And you know, um, we're obviously building a relatively big business in the US at the moment. Australian market's been a great market to us. We've. We're pretty lucky in Australia. We've kind of worked with the same customer base for a long period of time. Okay. Um, so the customer base is very consistent here.

Speaker A: Yes.

Speaker B: Um, the U.S. it's, we're trying to see if, if the same trends will play out. Um, but I think that's. Yeah, it's been a fascinating kind of journey.

Speaker A: Yeah, I mean it's like 14x the straight market and uh, very state fragmented as well. Like I think the, the eastern build states here are a bit more sort of homogenous in a way.

Speaker B: So that was one of the really. So, you know, I think the next kind of big challenge in our business, the one that we're going through at the moment is expanding to the US market. I work US hours at the moment and Australian.

Speaker A: Okay.

Speaker B: And it's. Yeah, it's hard. But I think that, you know, what we're really excited about is, I mean I've got a saying in models that granularity is clarity. And I think that's the same for life for me. You know, like details matter, details matter, details matter. Like if I've got the most beautiful brand ad in the world on a six second bumper, it's never going to work. Like in. Sometimes we're too comfortable to gloss over. The detail and granularity in Australia has been important. But my gosh, in the US does that set you apart? And I think that's been really, really exciting to see. Has been. Yeah. The difference is even state to state, you know, the taxes. The taxes, for example, on a state to state basis will actually affect the value of a customer, which affects, you know, what your media ROI and your returns are.

Speaker A: Wow.

Speaker B: Uh, at a customer level.

Speaker A: And you put it onto your model 100.

Speaker B: Yeah. So we have varying CLVs, we have had that for ages. Right. So in varying silvas are really cool to have. Like if you can have of varying customer lifetime value over time, what you can see is okay, if we go into slightly more effective media choices and we're able to attract a higher value of customer, do we then see that pay off in a, in kind of a disproportionate way um, to then just like acquiring mass market volume who might be much lower values. So and the same applies for states as well in the U.S. so one of the really important things in the U.S. is ah, okay if the customer m lifetime value varies by state or even down to dma. Right. Like you can go all dma.

Speaker A: So good.

Speaker B: It's awesome. And so when you think about that problem it's just so interesting because it's like actually my media optimization is not just dictated by you know, my media optimization choice is not just dictated by how much volume I'm getting from people. It's dictated by how much value I can like extract and the type of customer that I might attract through different media choices. So that's just been a fascinating kind of exploration for us. It's something that's been really effective in the US and seems to have taken off.

Speaker A: Yeah, no, it's interesting. I remember um, buying DMA sort of targeting in meta ads like many years ago in America. Yeah. The granular you can go into crazy like with uh, direct mail systems like a DMA level. It's crazy.

Speaker B: Yeah, yeah. And I think you know that for me is really, really exciting. You know I think long term what do we want Mutex to be? And you know, and this is something I think about a lot is and we are not an MMM platform long term. We're a growth co pilot and we're a co pilot that helps customers make better pricing decisions, better marketing decisions, better distribution decisions. All of these things that really matter to actually optimizing and delivering on growth back to a business. And I think traditionally we have had really, really really large teams that have not been able to act quickly to provide that information and those answers to uh, two people at scale. And I think MMM is a really good vehicle for that. But I don't think it's the end state vision. I think the end state vision for us is to move, move m beyond mmm. And I think that's a really, really important thing. We think about a lot.

Speaker A: Almost like a decision business decision making engine or something like that.

Speaker B: Yeah, we used to call it uh, a uh, recommendation engineering. Um, but we think the growth co pilot more neatly encapsulates it. And I think copilot's an important word because it, it speaks to the fact that we see this very much as the AI is an assistant. Yeah.

Speaker A: And you're driving it for your own gain.

Speaker B: That's right. So you know, rather than replacing it. Yeah. It's sitting there as the co pilot rather than the pilot. And I think that's very uh, much where I feel AI is going. That's very much where like it seems to make a lot of sense to me and I think that's what, that's what's really, really exciting because we've never had effective growth co pilots to process the reams of data. If you speak to anyone in a media agency, if you speak to anyone in client side, if you speak to any kind of people using this stuff on a day to day basis, they are drowning in information. Drowning.

Speaker A: And different types of reports and different channels. You got to amalgamate them all together. It's hard.

Speaker B: 100%. It's really hard. The ah, great unspoken um, truth of marketing is we've 10x the workload and halved the resources.

Speaker A: Yeah. And the wages are stagnated or gone down.

Speaker B: Exactly. And so I think we have to have a structurally better system to solve that problem. And I think that's where I get really, really excited about the potential here.

Speaker A: I see the problem with that every time. I mean even, even if something simple as like let's pretend they don't have an MM or any of that, they're getting multiple reports and multiple uh, digital agency over here, other agencies here. And the poor CMO has to kind of make sense of all that somehow present that up in a report up to the board and go, this is what's happening. Like that's a nightmare within itself.

Speaker B: Yeah, for sure. For sure. Probably rooted in the way our business came about. Like by being very close to customers early, having that almost, that almost those customers built into the DNA of the business at the start kind of makes us culturally orientate a lot more to that customer empathy and that customer obsession than I think most other businesses would.

Speaker A: And who is your customer? Is it head of marketing, CMO or

Speaker B: is it slightly different CMO M heads of media, um, heads of business intelligence around ah, the media space. Um, occasionally finance, sometimes CRO or sales.

Speaker A: Depends.

Speaker B: It depends. We have had a few CRO and sales customers. I wouldn't say they're our best. I think, you know, we are at our best when we're helping The CMO champion and be their best, um, and present in the best possible way the best financial case to make the best decisions to grow and to grow incrementally. When we confuse ourselves from that identity and being that champion for the cmo, the head of media, for the marketing department to really champion the financial case up, I think it's very easy for us to confuse ourselves with that identity. And I just don't think. I think great businesses know what they are. Um, and I think we quite clearly know what we are. We're building a growth copilot with a champion of our customers, specifically the cmo, the head of media, uh, and really helping them, and helping them nail themselves. We make life easier for them by helping them collect, structure, process and ultimately use their data to generate the right answers for their business quickly. And we are razor focused on delivering that by understanding whether our customers are adopting, using and ultimately growing through our tool set. And because we're very clear on that, um, I think it's very easy to kind of, you know, start to start to go what, what we're for and what we're not.

Speaker A: Yeah, well that's great. I mean, I always say, uh, one of the big things with Clear strategies is focus. At its sort of heart, it's like, which lane are we going, which direction are we going? And sort of ignore the rest. Still be, you know, listening to it but like know that we're going here and this is what we're correct.

Speaker B: And I think, you know, there's, there's going to be like different modeling companies that emerge for different purposes. Like, I think, you know, just to reference two businesses that I'm huge fans of, the recast business is a fantastic business. The thing I really love about them is that, you know, what they're doing in the model transparency space and pushing the boundaries of model governance and really setting the bar on that is really

Speaker A: important because these would go the other way.

Speaker B: Yeah. And I think, and I think, you know, businesses like Analytic Partners really created the category very early on.

Speaker A: Yes.

Speaker B: Um, you know, Analytic Partners has been around for 20 years. It was found 25 years now. Um, it's founded by a lady called Nancy Smith, um, the local managing director of Australia. Paul Sinkinson's a friend.

Speaker A: Oh my goodness.

Speaker B: You know, he, I went through kind of, you know, some pretty tough, tough times with executive turnover and stuff like that recently, as was well documented in the media. Um, and like, Paul was one of the first to reach out and things like that and give me some encouragement and Things like that. And I think, like, the thing I do love about the MMM industry is we do have a bunch of very respectful competitors in the space and I think, you know, I'm deep admirers of those businesses.

Speaker A: Um, who else do you talk shop with? I mean, you need to talk shop with someone who actually understands it's pretty complex.

Speaker B: Certainly the media agency bosses are very interested in the space. Google has some amazing people in the MMM space. Like, just really bright. I think that, you know, I think they have some very, very clever people. Meta Obvious has some clever people in the space as well. And we've got, you know, our marketing science director is X Meta. So he's um, and he, he's kind of pretty plugged into that community. I really love talking to other people in the machine learning Bayesian space. ML and Bayesian are kind of at odds. A lot of people don't realize that.

Speaker A: Yeah, yeah.

Speaker B: You know, machine learning people are kind of, you know, pure empiricists mostly. Whereas the Bayesian people believe that, like, we should be trying to understand the dynamics of what we're doing, the principles of what we're doing, and then, and then hard encoding models around that. And I think it's a really interesting tension that exists in the data science community at the moment.

Speaker A: So let's go back to 2022. Kind of had this issue. You raised some money, 2.4 mil. And then obviously you've got sort of, um, VC funding since then. So was that a bit of a crash course and how the VC funding world works?

Speaker B: Yeah. So, um, what happened there? So EVP actually like kept in touch with me.

Speaker A: So after. So they were in the first term sheets?

Speaker B: No, no, they weren't the first term sheet.

Speaker A: Okay, but you were talking about.

Speaker B: I was talking, yeah. And then they kind of came back to me and they were looking at the business. I was giving them kind of an update on my numbers every single month. They looked at the business and, and I basically sat down in Redfern, uh, with Justin Lippman, who is an amazing VC partner. Like, I know a lot of people say a lot of bad things about VC. I have nothing but the opposite experience with EVP.

Speaker A: He's pretty active on LinkedIn, I think as well, isn't he? On social media.

Speaker B: He's hilarious. Um, so. And I mean, he basically just gave me the nuts and bolts of a deal that he thought was fair. He gave it to me up front in an email, said, here's what I think I can get through. Are you interested? It was beyond fair as a deal. Nothing dodgy in it whatsoever.

Speaker A: No losing board seats because that's kind of where I think sometimes it's just that power dynamic imbalance that's.

Speaker B: No, no, he wasn't seeking that either. He, um.

Speaker A: Oh, great.

Speaker B: Uh, and so we basically got a, a deal through. I then actually, I was actually in the Witch Sundays as I was doing it with my dad on my first holiday in a while. Yeah, we went on this like, we went on this like dodgy fishing boat spearfishing for like nice, um, roughly about three or four days. Oh, there was like 20 other people sleeping in bunks. Like when I say it's a dodgy boat, it's like an old racing yacht, like. But it was fantastic. So I was out of phone range for like four days and I got back in the story that we'd raised money had broken. I was like, ah, oh well that's good. And, and then the second kind of round, we were talking to a few different groups. We wanted to really expand into the U.S. yep. A big misconception about the business again is our Australian business is, is really, really strong. So it's very, very healthy. I think it's a very stable business. As I said, we've got great customers. They've been around us for a long time. We've got great growth primarily through customer referral.

Speaker A: Uh, but it's also not a huge market like Australia in general. It's like there's an end.

Speaker B: And I think the US is a really big market and it's one where you do require a bit more scale

Speaker A: to get into it and foots on the ground in America.

Speaker B: And we also wanted to, we also wanted to build it out of New York.

Speaker A: Okay.

Speaker B: So we raised about 7.5 million AUD on primary. We managed to kind of get that round away, uh, at a 75 mil post money, which is a pretty, pretty, pretty decent round at the time. And then we just raised a further 17.5 this time at 132.5 post money.

Speaker A: So this was just last week or something? Yeah, yeah, I think I saw it in the news. Yeah, yeah, congrats. That's.

Speaker B: Yeah, so that was pretty good. Sustained growth. The latest round, we really got away, uh, off the back of our U.S. progress. So we've won kind of about 10 really good U.S. enterprises over there. And they're all tier one U.S. enterprises, enterprise, which is kind of where we want to play.

Speaker A: So that gives the investors confidence that, hey, you get 10.

Speaker B: I think most importantly, like the, like three of them have already been through a renewal cycle and expanded the relationship.

Speaker A: Oh yeah.

Speaker B: Um, so that, that's a really positive signal.

Speaker A: Yeah.

Speaker B: Um, for us, like we, you know,

Speaker A: see retention's pretty strong.

Speaker B: Yeah. I think like, you know, you've got to, you've got to understand a business based on like, you know, how well do you serve and retain customers. And I think, think that's something we've spent a lot of time on. There's definitely no doubt like in the early stages of mute next we had so such chaos in trying to figure out the right account management structure beyond the founders. Right. Beyond the founders taking every call. Scaling an account management function that would be able to replicate what we did.

Speaker A: That's hard.

Speaker B: Was really difficult. And we got that wrong on, you know, ah, about three customers. Um, and we got it wrong on three fronts. One, we badly over promised on the roadmap to those customers because we were just not confident with them. Two, uh, we didn't have the data ingestion scaled and that was a huge pain point for those customers. And three, the account management turnover was um, high at the time because we didn't know what we wanted, um, and we didn't know how to scale the function. And so we just made a series of really big errors there. Um, and we lost three customers. That and which was devastating, to be honest. And you know, I think a lot more about the mistakes we made and the experience that we gave those customers and our wins, to be honest. You know, that's probably what keeps me up as a CEO at night, is trying to make, make sure that we're building, you know, a very customer centric culture with a very customer centric product that's able to deliver, uh, you know, superior growth outcomes for customers. You do have to become obsessive about those things. And I think it's really easy to say, you know, we're building, we're building the best. Mmm. Product. We're building the best this product or you know, we want to handhold every customers. I want to build, build. You know, I, I want us to just to be customer, customer, customer, customer kind of culture and then we could. And then, and then you'll understand everything else.

Speaker A: Yeah, Bezos would be proud. Yeah, that's true. It's like, I think you, otherwise you can get really caught up in the technicalities of your product and, and lose sight of like what value is actually creating for the customer. If it's not creating value, there's no point to it. You know, there might be a Point for your own satisfaction.

Speaker B: We have to understand, like, why does a customer want a good model? Right. Primarily so that they can trust the answers coming out of it. Right. Like, that's the actual outcome that they want. Right. You know, why do customers want, you know, more detail and granularity? It's so that they can understand, um, and query and answer in the way that they want to. Why do customers want to be able to forecast and predict? It's because they actually want to be able to forecast. They've got an answer to the question. They want to be able to forecast what will happen with their answer. And I think when you look at every part of mmm and describe it through a customer lens, it changes your mentality around it. It changes how you think about the opportunity. Because the opportunity is not. It is not in market mix modeling. Market mixed modeling is an industry term. It's not a term that serves customers. And ultimately, we want to be a business service of customers.

Speaker A: I love it. Well, thanks for running through your story. I really appreciate it. And we're both at south by Southwest, uh, this week. So are you. You doing something on stage? I heard as well.

Speaker B: Yeah. Tomorrow. Yeah, I'm doing something tomorrow on stage. I'm doing a little demo of, uh, of our kind of our Hendrin product. And then we're doing a kind of a. I'm doing a panel on me creativity through mmm and how you can do it, which I think is a really interesting area to play in.

Speaker A: Yeah. Great. Okay. Yeah, we were just talking about creativity before with Josh, but yeah, no, thanks again for coming on the show. Really appreciate it. And all the best for Mutinex now and in the future. And I don't think we say this in Australia enough, but congrats on your success. And it should be a good thing. It should be, uh, applauded. And thanks for telling us what went on.

Speaker B: Yeah, thanks, man. Cool. Awesome.