Podcast: The Future of Ad Spend Attribution

Written By Sherilee Maxcy

Episode Summary

In this episode of the Free to Grow CFO podcast, Jon Blair and Michael True discuss the intricacies of scaling DTC brands with a focus on marketing measurement. They explore the transition from traditional click-based attribution to marketing mix modeling, emphasizing the importance of understanding the relationships between various marketing channels. Michael shares his journey from the music industry to founding Prescient AI, a platform designed to provide insights into marketing performance across multiple channels. The conversation also touches on the challenges of measuring retail performance, the impact of iOS 14 on digital advertising, and the role of AI in enhancing marketing strategies.

Key Takeaways

  • MMM is essential for brands expanding into omnichannel marketing with significant top-of-funnel spend.

  • Tools like Prescient AI can help scale profitably by uncovering relationships between channels that traditional attribution methods miss.

  • The future of marketing measurement lies in combining MMM with other tools to create a full picture of brand performance.

Meet Mike True

Mike True is the co-founder and CEO of Prescient AI, which provides AI-driven marketing mix modeling solutions for omnichannel brands. Prior to starting Prescient, Mike was responsible for helping clients of App Annie, IBM, and Oracle generate millions of dollars in revenue through the implementation of various artificial intelligence and analytics solutions.

Transcript

~~~

00:00 Introduction

05:48 Understanding Marketing Mix Modeling

11:58 Navigating Post-iOS 14 Challenges

18:11 Insights on Retail and Delayed Sales Data

25:49 Understanding Diminishing Returns in Ad Spend

32:02 Integrating Prescient AI into the Marketing Tech Stack

38:07 Closing Thoughts

Jon Blair (00:01)

Yo, what's happening everyone? Welcome back to another episode of the Free to Grow CFO podcast where we're diving deep into conversations about scaling a DTC brand with a profit focused mindset. I'm your host, Jon Blair, founder of Free to Grow CFO. We're the go-to outsource finance and accounting firm for eight and nine figure DTC brands. Today I'm here with my buddy, Michael True, co-founder and CEO of Prescient AI. Michael, what's happening, man?

Michael True (00:25)

What's up, Jon? Good to see you, brother.

Jon Blair (00:27)

Yeah dude, I don't know if you guys are feeling the same way but we're like kind of in this sprint right now between Thanksgiving and and like the Christmas season in terms of like I don't know if you guys are seeing the same thing but in our world everyone has gotten through Black Friday Cyber Monday so they're starting to think about 2025 what improvements they they may need to make for us on the accounting and finance front I'm sure there's people thinking about how to measure their marketing better but are you guys

Are you guys kind of a race to the finish line before people start disappearing for Christmas?

Michael True (00:58)

It's interesting this year where, you know, we had, there was some brands, I would say that pre Black Friday Cyber Monday, we're starting to think about tech stacks for 2025. And then we've got some, some in bounds and some re-engagements post Black Friday Cyber Monday, really digging into 2025. We made it a pretty clear priority or a pretty clear like requirement. You know, during certain weeks, there was no outreach to anybody that was not a customer. Just let people, you

Jon Blair (01:09)

Mm-hmm.

Michael True (01:25)

stay heads down on one of their busier times of the season. So starting to pick back up now for sure.

Jon Blair (01:29)

For sure,

for sure man. Well, I'm stoked to have you on the show. Today we're gonna be talking about, well, marketing measurement, I'll say in general, but more specifically, getting more specific about marketing mix modeling, which you guys do over at Prescient AI, versus what I'll call the more traditional click-based attribution, which is what I would say the vast majority of founders in the DTC world that we operate in.

are used to using to measure marketing performance. But before we dig into that on a more detailed level, I'd love to hear a little bit about your background and your journey to founding Prescient AI.

Michael True (02:10)

Yeah, so my background has always been in big tech. was largely in the database world at Oracle IBM, then moved into the analytics world over at IBM and then into their Watson team, which was, I don't know if the folks might remember, is this little AI thing that beat Ken Jenner on Jeopardy, long time ago.

Jon Blair (02:25)

yeah.

Yep, definitely remember that.

Michael True (02:31)

So my focus is predominantly on like financial services and the healthcare system, know, deploy, flying around and deploying software across at that time, physical servers, like an x86 server. And then started to see the migration to the cloud. Yeah, it was just nothing to do with DTC for a long time until, until recently. I've met my co-founder, Cody. I had been dabbling around in some ideas in like the AI space and found myself into the music space, trying to predict tours for record labels. And,

COVID hit and touring went away. Got really lucky that that happened honestly, because the shift had, the focus had shifted for labels. When COVID hit everybody's inside. Well, what are people going to do when they're inside? They're not going to go to a tour. They're going to stream music. And so there's attribution in the music industry had not really existed because there's no Google analytics behind Spotify and Apple. So there's no last click. yes, I met my co-founder, Cody.

Jon Blair (03:10)

Mm-hmm.

Yeah, for sure.

Interesting.

Michael True (03:26)

I'd like to consider him the Michael Jordan of research science in this field. He's a very competitive researcher. We met and we started building, he started building models and I started selling models to a major label. And they did a really good job of being able to measure, forecast and optimize marketing spend for Cardi B and some really cool artists during their early 2020, 2021.

Jon Blair (03:52)

That's awesome, So I actually was a touring heavy metal musician for a number of years. And so I've, yeah, this is real. That's probably a whole nother podcast episode, but the music business is interesting in a number of different ways. you know, I'd love to, so how did you go from that kind of like marketing analytics in the like label world to, you know,

Michael True (03:58)

No way, is this for real? It's awesome.

Jon Blair (04:20)

kind of this focus on e-commerce, digital advertising in more of the DTC or like omni-channel e-comm space.

Michael True (04:27)

Yeah, great question. It was February of March of 2021. And we predicted Cardi B's number one single up at 96.3% accurate by recommending the label to ship from X channel to X channel to X channel X channel and then predict how many streams she'd get compared to their existing spend. Predicted at 96.3% accurate was like a big win for us. And we weren't too sure what we're going to do next with the technology.

But I'm sure everybody on here will remember April 26th of 2021, the day when Apple announced iOS 14.5. And my co-founder called me and was like, hey, which industry do you want to go to? He's like, I think we should go raise capital and I think we should build a SaaS platform. This model can work across any dollar spent online, like which vertical do you want to start with? So we talked to like healthcare and financial services and ultimately made the bet on e-commerce at the time.

Jon Blair (05:02)

for

Michael True (05:22)

because just like streaming, e-commerce was exploding and we had made a bet that the technologies that were being applied to solve for iOS 14.5 at the time was the right call, right? was Meta, people were still pumping on Meta Google and a pixel was really, could really solve for that. We had some Prescient, which means having knowledge of foresight and would have meant before it occurs that.

Jon Blair (05:37)

Yep. For sure.

Michael True (05:48)

brands would start to want to scale into top of funnel. They would want to go more omnichannel. And we're going to have about a two year head start on the research of these models. Because everything that matters is the math, right? All that matters is the math. And we have the math infrastructure. We had a head start. And so yeah, we started in DTC, shipping to Amazon. We're coming out with a retail model here in Q1 as well. Yeah.

Jon Blair (06:12)

Nice.

Dude, that's awesome. I love that. So I want to start with just like the fundamentals, because I think that, you know, we work with dozens of DTC brands and I think still, well no, I mean, it's very clear that still, you know, kind of the go-to is Triple Whale, which has a number of different issues in my personal opinion. But I think they're starting to

the vast majority of brands are starting to get exposed to this concept of marketing mix modeling. So like at a basic level, what is it and how does it differ from traditional kind of click-based attribution?

Michael True (06:56)

Yeah, for sure. first and foremost, it's like, the MMM is not the source of truth. The MTA is not the source of truth. Incrementality is maybe the closest thing at a point in time you're going to get to source of truth. Source of truth is the marketer. There's just a variety of different ways to measure something that are trying to give you the ability to do data like measurement triangulation. I'd say a good example, we send Triple Whale I don't know, two to five deals every week for sure.

Jon Blair (07:06)

Sure.

Mm-hmm.

Michael True (07:22)

and so I guess the difference of a brand that would come to us and say, Hey, we're on, we're on Shopify, we're on Meta and Google. They don't need an MMM. They're going to be onboard to our platform and then leave our platform because it's just, they don't need that sort of measurement at the time. Right. But when you're a brand that is starting to go into YouTube, TikTok, podcast, linear TV, new channels like app loving really hot top of funnel. If they have a DTC plus Amazon presence, even better fit for an MMM model. And so.

Jon Blair (07:44)

Mm-hmm.

Michael True (07:51)

What an MMM is not trying to do is trying to say, Jon Blair saw this ad, click this ad, click this ad and have a deterministic customer journey. What an MMM is trying to do, I always say like an MTA is going to tell you a lot about a little. So it's going to tell you a lot about some of the transactions that came through deterministically. Traditionally, a media mix model would tell you a little about a lot. So it's going to tell you a little bit about the relationships between all of your channels and your revenue on kind of a time series path.

Right. So one is a statistical model and then one is a deterministic model. So what is that causal relationship to, Hey, I'm increasing my spend. I'm sure a lot of the brands are seeing this. I'm increasing my Meta spend on my YouTube spend, but I'm starting to see some of my Amazon sales go up or I'm starting to see some of my retail sales go up or I'm starting to see, you know, our top of funnel for just our DTC store. Right. I'm starting to see like our last click is starting to take a whole bunch of credit where maybe it didn't deserve to have that credit in first place.

Jon Blair (08:50)

For sure, so am I correct in saying that like when we're talking about a brand where a tool like yours and just marketing mix modeling in general can start to become really valuable is when they have heavy top funnel spend and they have more than one channel where sales are potentially converting. That the more a top of.

funnel spending you have, the more sales channels you have, the more potentially valuable marketing mix modeling can become. Is that correct?

Michael True (09:23)

Yeah,

I wouldn't isolate it to being a requirement to be an omnichannel sales brand. But if you're a brand that's like four, we like to see five channels and above. call it like our sweet spot. Four to five is considered a good fit for us. Three to four gets kind of dicey. Anything less than three is just not a good fit for this. But you know, we'll have some brands. And then there's a sense of like, we have brands that are, you know, on Shopify and they're Meta Google YouTube. And we'll talk to them saying, well, we're planning to go into Tatari.

Jon Blair (09:42)

for sure.

Michael True (09:53)

you know, next year we're planning to have you in the podcast. We'll onboard them to the platform so our models can start learning their data from the channels they were spending on. And then as they start to layer on these new channels, we're able to pick them up, you know, at a much higher, much faster, if you will, to start being able to give them a pulse on how that new channel is performing.

Jon Blair (10:14)

So if you have a brand that is in that sweet spot, we'll call it five channels, right? Heavy top of funnel spend. If they are sticking with one of these more traditional click-based attribution tools like Triple Whale, what are some of the things that common pitfalls that you see that might get brands into trouble if they're just using.

something like that, and they're not considering using marketing mix modeling for measurement.

Michael True (10:45)

Well, the pixel or an MTA is not designed to measure a statistical relationship between, you know, views and, in revenue. It's really looking for that click. so when you get more biased towards bottom of funnel, it's not going to allow you to feel as confident in the measurement of, how is this heavily view based channel like connect the TV or YouTube actually driving our sales? And so

Jon Blair (11:01)

Mm-hmm.

Sure.

Michael True (11:11)

When we, quick aside on this, like when we first came in to decide to start with e-commerce, I interviewed close to a hundred DTC marketers. And the thing that I found out was, is I was trying to get to their why of like, measurement? Like, what is it? What is this why? A lot of people, like, I just got married. I want to buy a house. I'm paying off student loans. I have all these bills to pay and I'm coming into work.

and I'm being tasked to scale this business, I know that I can't do it anymore on just these two channels. I need to go into these top and funnel channels. And I just want to feel confident every day that like, I'm going to make the right bet, right? And so without leveraging an MMM, which they're designed to do, you know, that level of confidence is, you know, is not as strong when you're making those spending decisions.

Jon Blair (11:44)

for sure.

Totally.

Yeah, yeah, no, mean for sure. It's, in the post iOS 14 world, I'm always hesitant to say post iOS 14 world just because it's such a cliche at this point, right? But I've been in e-commerce for long enough to know what it was like then and what it's like now and how different things are from, in terms of having that confidence of like, I'm spending a dollar here and it's driving this. I feel confident in the connection to the revenue it's driving, right?

Michael True (12:08)

You

Jon Blair (12:28)

in the post iOS 14 world and I'll just say it's actually not iOS 14 in my opinion. It's that e-commerce is becoming more saturated, right? It's still growing. It's still growing at a nice clip, but there is a low barrier to entry to just fire up a Shopify store and fire up an Amazon seller account and start spending ad dollars. It doesn't mean that every brand is

you know, really sound with their marketing fundamentals or has really great products or even has a great customer experience, but there is a lot of competition, right? And so because of that, it's, it's driving brands to have to be really meticulous about understanding how profitable their spend is in any, any channel. And I would even venture to say that in many cases, the average brand is having to expand into more than just a Shopify channel a lot quicker.

than they used to have to say seven years ago. Seven years ago, you'd still could be like a first mover in a product category, DTC. you, and I mean like at Guardian Bikes, when I was at Guardian, we brought on a lot of people who left Tuft & Needle after the, they merged with Sealy Simmons. And they scaled to nine figures in revenue with very little top of funnel. A lot of it was Google.

bottom of, like, because they were the first mattress company to sell mattresses people were already searching it. That's a way different game than like, I need to find the next new set of eyeballs that I need to get to, right? I need to keep leaning into top of funnel and I think I need to get into a new channel and I don't have a lot of confidence of how that's gonna impact the bottom line. So it is, it is a stressful, stressful thing. I would say the one that we see most commonly and I've talked with Will Holtz who works with you quite a bit about this.

is understanding the difference between or how to think about how Shopify versus Amazon is performing when you're spending heavily on a top of funnel channel like Meta or YouTube. We see that all the time with the brands that we work with, And what I can tell you from a CFO's perspective is that there's no doubt that they're connected, right? But measuring the degree to which they're connected, right, is a different story.

What kind of advice do you have or maybe not even just advice, like how can a tool like Prescient help with understanding? You're spending heavily on top of funnel channels. You've got an Amazon store and a Shopify store. How can you guys help a brand discern how that top of funnel spend is driving each of those storefronts?

Michael True (15:09)

Yeah, and again, it comes down to the triangulation of, what are you seeing within your Amazon store, right? That's very deterministically based off the conversions that are taking place in there. With with Prescient and typically MMMs, you would see an MMM run once a quarter. And this is a big education thing we've been working people on. It's like once a quarter, maybe once a month, it's at the channel level and it's giving you a cross channel media measurement and maybe some recommendations on how to spend.

which unique to our platform, it allows people to test and iterate very quickly. So we have what we call halo effects, which essentially is we're taking credit from the bottom of funnel. So search across Amazon search across, you know, Bing and Google, and we're redistributing that credit up to the top of the funnel where we have the highest confidence that the likelihood that the awareness from this campaign is actually what drove conversions over onto Amazon. for us,

We do this at the individual campaign level. The models will run every single day. And so as you start to shift and test new spends based off of what you're seeing with our halo effects. a good use case of this would be, you know, a large cookware brand was spending on Mountain. And when they were looking inside of Mountain, the performance was not that strong, but we were showing incredibly strong performance. Well, because Mountain was only measuring over to their DTC store.

Right. And so when we started to say, actually, there's a lot of people that are sitting on, you know, their TV, probably on their laptops, they see your ad and then they go search, can we find a prime deal on Amazon and go search that way? So the granularity of what we're allowed them to do is being able to make very rapid and dynamic decisions on how to reshift their spend for non Amazon ads and then look at that impact and quantify that impact over time compared to, you know, how they were previously spending before. So.

Jon Blair (16:36)

Mm. Yep.

Yeah,

that's really interesting. Do you have any other examples that come to mind of like how you've helped one of your clients like really unlock but make decisions based off of the data in your platform to really enhance? I mean, obviously scale, but at the end of the day, profitable scale.

Michael True (16:58)

Thanks.

There was a brand, it's about a hundred million dollar beauty brand and they had Tatari turned on. He's become a good buddy of mine. Had Tatari turned on, actually down in Austin too. And he turned it off because they needed above a three row ass. And so it was like, I they have 2.67. So when they onboarded, they onboarded the Tatari and they can go back a year. And so he's like, I think that there were these two campaigns that were actually pumping that weren't getting credit. So we onboarded him.

We had all of like what the platform reported was being shown and we hadn't run the model yet. Bing was showing like a 22 ROAS and he's like, dude, that's way too high. I'm like, so what do think it is when we run the model? He's like, I think it's a six. So we turned on the model, ran a couple of days later, the insights come back and we had Bing at a 4.65. So the Delta between 22 and 4.65, we're gonna be redistributing that credit.

up to campaigns at the top of the funnel. Now they are DTC plus Amazon brand and their ROAS was actually just under 4.5. And so they turned Tatari back on and when they turned it on, they were able to go back and look at some of these like saturation plots, right? We'll be able to show them like, hey, where's your sweet spot of spending? know, you're oversea, if you spend two dollars and make $10, you can't expect to spend $2 million and make $10 million. Eventually there's going to be some diminishing returns. And so.

Jon Blair (18:38)

For sure. Yep.

Michael True (18:40)

our team would work with them and we started very carefully scaling and testing the Tatari channel after they onboarded. we'll be coming out with a pretty cool use case here in the coming weeks of how they're able to successfully scale that channel and continue to scale that to this day, continuously optimize the spend. And so there was a retroactive play of saying, hey, I onboarded to the tool. I was able to go investigate and confirm some of my hypotheses before with your MMM.

We turned the TV back on, worked with our team to scale it and we're seeing great problems and results from there.

Jon Blair (19:15)

So I'm actually curious, and I think probably maybe specifically in the retail channel, like physical retail, but there's probably some other top of funnel channels that I'm not thinking of that probably have the same issue. How do you guys deal with, we'll use retail as an example, you've got a channel where the sales...

the actual bottom of the funnel like sales conversion signal might be delayed depending on what retailer you're using and how quickly that data is available. How does that play into your models being able to provide insights about a channel where it may take a little bit to understand really what sell through is?

Michael True (19:58)

You're saying the example would be you have a brand that's selling in Target, Walmart, Costco, and that retail data might come in at certain granularity. You might get some DMA level data that comes in weekly versus just total sales that comes in monthly. Is that what you're asking?

Jon Blair (20:05)

for sure.

For sure. Sure.

Yeah, yeah. Like how do you guys deal, how do you guys advise brands on working with these data sets that come in? It's not like plugging into a Shopify store, right? Or an Amazon store. What is some advice or how do you guys work with some of those channels to get as timely insights as possible?

Michael True (20:36)

Yeah, so we leverage all of their existing media channels. So the data that when they onboard to our DTC chip platform, it's 12 to 16 minutes point and click and just all of that, you all of your paid social, your GA and your Amazon DTC data. For retail, it really comes down to, we have some retail clients that own their own POS and own brick and mortar stores. And that just data comes into a database. We pull it from there and we can run the model. but there's other examples where, you know, you're getting Costco data once a month.

Jon Blair (20:55)

Yeah, that's nice.

Michael True (21:04)

We've come out with what we call, it's like a configuration offering. It's more of a, you have a dedicated research scientist, right? You have a dedicated engineer and our team is working with you on just getting exports of that sales data. We ingest it, we run it through our models and then we actually come back with like a customized retail readout, which is going to show them the exact same thing as you would see with your DTC, within our DTC SaaS platform. So.

Jon Blair (21:17)

Nice.

Michael True (21:30)

Where are you overspending? Where are you underspending? What is that optimal budget? How accurately we back testing against the last 12 weeks of your sales? The last read we just did was at 98.2 % accurate, which was remarkable to see as researchers in this field. But it's a very similar output. It's just a human walking you through based off of whatever cadence you're going to be getting those sales data from. So it's a little bit more customized.

Jon Blair (21:47)

Nice.

That stuff's super important because we've worked with several brands that are in the higher eight figures pushing into nine figures and physical retail becomes kind of a like a must at that point, right? Like in this day and age, it's not to say that there aren't e-comm brands that reach nine figures. are, but I see them as the exception, not the rule. It's not like back in the day, Tuft & Needle could get to 150, 200 million and not have to get into physical retail where that ship has sailed.

But so, we're advising a lot of brands that are on the upper end of eight figures and pushing nine, like how to think about, how to think about the fact that these are all heavy top of funnel digital advertising spenders. Meta, YouTube, maybe even Connected TV, and that absolutely drives the physical retail channel to some degree as well, right? It's not a completely separate channel.

Michael True (22:54)

question.

Jon Blair (22:55)

So what, for you guys to able to provide some analytics about that is like, I think incredibly, incredibly important because in the past, I'd say like seven to 10 years ago, retail was kind of this black hole if you're a DTC brand expanding into physical retail. And I was like, hey, I'm sure we are driving some demand, but measuring exactly what we're driving is really, really hard. I would even venture to say, and I'd be interested to know if you have any,

kind of anecdotal kind of evidence of this, but like I've actually even seen if you can prove out, if you can start proving out a connection between your spend and sell through inside of a physical retailer, you can actually really start influencing the buyer's decisions on future purchase orders showing like, hey, you're getting free advertising spend from our efforts on Facebook and YouTube.

And here we've got this great model that Prescient AI put together for us where we can actually show you that. Like to me, I think that's a huge game changer when it comes to negotiating with buyers at Target and some of the big retailers.

Michael True (24:07)

Yeah, so we'll 100 % right. Because with the models, it's going to forecast out. And if they get in data, we have a mattress company. If you get data by the DMA level, we can actually start forecasting out what we think those predicted sales are going to be over the next quarter by DMA and start forecasting out what we think the change of CAC or new customers or ROAS is going to be by DMA or product category or SKU by DMA. And so it gives them a lot more ammunition to being able to go have those conversations.

Another interesting thing, think, just for folks and they're thinking about the omnichannel kind of retail spaces, we spoke with both Meta and Google and, know, Google Meta has now coming out with like a kind of think about it like an advantage plus shopping, like that sort of, you know, programmatic model, if you will, but doing it and being able to tie sales from their digital ads into retail stores. So the same thing with Google being able to do that. then, you know, if they're able to track your

Jon Blair (24:54)

Yeah.

Yeah

Michael True (25:04)

phone and have the IP address seeing a conversion in that location of that store. Now they're starting to trying to stitch together that relationship just as we're doing. We're just doing it across all channels, but Meta and Google are now starting to try to optimize for that as well.

Jon Blair (25:16)

Interesting.

That's interesting, man. So I'm curious, one, because, you know, Free To Grow, we work specifically with fast-growing, profit-focused ecom brands. They're not, they don't have a bunch of venture backing, right? So like, every dollar they, every incremental dollar they spend on ad spend, they need to have some confidence that it's driving incremental profit, right? Or at least incremental contribution margin dollars. One conversation we're always having with clients is, you know, to expect,

and being able to absorb diminishing returns over time because as you're scaling, we see diminishing returns. I'm curious, like, your models, do diminishing, when you talk about forecasting, do a lot of the diminishing returns get factored in based on, like, statistical models showing those diminishing returns in the historical data? Are there other methods you guys use? I'm just kind of curious about that.

Michael True (26:20)

Yeah, we'll plot out. It's one of our biggest, I would say, points is a quick aside on that is typically saturation plots have to use linear regression models. Everybody's seen the linear regression models where they're trying to show what is that plot. We figured out a way using some proprietary machine learning models to do non-linear regression models. so instead of trying to, what does that mean is a linear regression model is going to assume that connected TV and Meta.

Jon Blair (26:32)

Yeah, for sure.

Michael True (26:47)

have a similar shape of saturation, right? Because it's trying to fit that line to the historical data. What we try to do is we try to find the shape of the data so our saturation plots can look like a camel's back. They can look really wonky, right? What it allows us to do is really pinpoint historically what is that sweet spot of where you should be spending, right? But we also allow brands to do is run simulations. And so on the fly, right, you can type in, hey, well, what if I increase my spend to X amount, right?

Jon Blair (26:51)

Mm. Yep.

Michael True (27:17)

It's going to show you on the saturation plots where you end up and then what is the forecast over a certain time period compared to your existing spend. it's kind of like showing what is the incremental growth based off of adjusting my spend from here to here. So we allow the users to manually do that, but now our models just scan every single campaign factors in things like seasonality of the business, the buying cycles of the products, and it'll tell them exactly how much to spend on each campaign.

and then forecast out with a certain level of confidence what we think that predicted CAC will be. What do we think that predicted top line revenue? What do we think that ROAS is going to be? So it's a much more dynamic ability to look at saturation plots and then kind of tinker around and pull levers to see what will happen on the forecast.

Jon Blair (28:03)

That's interesting, man. That's super interesting. And I mean, it makes sense. Like the thing about linear, I've seen guys in their content on, you know, LinkedIn and Twitter talk about like these linear regression models to basically forecast the diminishing returns of your ROAS or MER. And like, the thing is they're just overly simplistic in my opinion. And such that like, these are real dollars that people are spending that they're nervous to spend. You can't have...

Michael True (28:29)

Yeah.

Jon Blair (28:31)

Like now every model is a model. Like the definition of a model is it is a simplified version of reality and it's not perfect, right? But it's gotta be close enough that you're not, that you've got some confidence in using it, right? I'm just curious, like obviously there's, there's, details that are like beyond the scope of this conversation and some of this is proprietary. So like, obviously I don't expect us to get into all this, but like walk me through how AI or machine learning.

is supporting what you guys are doing and how that is really kind of a game changer or like a, you know, a tailwind for you guys for having these really, really, you know, trustworthy models.

Michael True (29:13)

I would say the traditional a lot of the MMM's you'll see in this space have leveraged existing research papers that have been used since 1962, right? And some different sort of flavor of that, but they were always designed to do more longer term planning, saying, hey, well, here's how much you should distribute at your channel level. We'll run it a quarter later. We'll do something sort of retroactive and then being able to try to learn from this last three months to try to make a better season for next month.

These machine learning models, they call it, they leverage it like a Bayesian prior. So they have sort of prior beliefs based off of prior belief, meaning what is that saturation plot? What is that measurement? What is that seasonality based off of all of the learnings from your historical data? Now those priors can get updated with new data sets that comes in every, from in our case, they come in every single day from the ad platforms. So we're seeing what you made on Shopify. We're seeing all your GA data. We're seeing

how much you spent by channel, by campaign, what is the reported attribution from the platforms. And those will hit the models and the prior beliefs can change based off of the measurement, can change the measurement and the forecast based off of that new data that's coming in. And so it's kind of an always on robot that gets smarter and smarter as you go. the key thing, and I would just encourage any folks that are using an MMM or thinking about using an MMM is they do not like

Jon Blair (30:22)

Hmm.

Yeah.

Michael True (30:38)

consistency at the same spend thresholds. They want to see as much experimentation as you can. So you can start to see these saturation plots move. last point is the unique value of ours is I do that. We use the same dummy. It's an actual client, but we anonymize it. And sometimes we'll come in and do demos and I'll see three days before the saturation plots looked very different as they started to test their spend. So it's very unique to us. can almost

know, Cam from Hex Clad says it really well is like, it's kind of weird because yours runs at the frequency of an MTA, nearly every, it runs every day, almost at the granularity of going down to the individual campaign level. And it's doing all of these forecasting optimization models and halo effect models. And so, yeah, it's just a lot of statistical math that's learning from as much data as possible and looking for as much change as possible.

Jon Blair (31:32)

That's cool. what would you, what advice would you give if you got a brand that's healthy eight figures scaling their, you know, their, they've got multiple channels that say they're at five or they're approaching five and they're interested in getting something like Prescient going and they do. But what, what does the, what does Prescient replace in the tech stack and what does it not replace? What are some other things that you think that, cause I know that that can be a misconception with attribution, right? I see it all the time with like,

Triple Whale it's gonna do this. No, no, no, it's not gonna do that. It is good at this and I'll give it credit for that. But like, what will it replace and what will it not replace? And in your opinion, what are some of the other marketing tech stack tools that still should sit alongside a tool like yours?

Michael True (32:17)

I I kind of, I'll base it off of like what I see with like, you know, the guys at Jones Road and Hex Clad of like their form of triangulation. And again, they're very sophisticated in their approaches here, you know, Cody will, hey, Cody from Jones Road, I turned on YouTube, we ran an incrementality study, right? And just to validate like the incremental of this new channel, but it's only at a point in time.

Jon Blair (32:26)

Mm-hmm.

Michael True (32:41)

Then they match that up against what our measurement is and it matches up and it's like, well, now I feel confident in what Prescient is saying. And then they would use our optimization model, right? So now they've gained confidence in YouTube. They're looking at us for like, okay, well, what should we do next? Cause incrementality is only going to tell you what's happened at a point in time, but what do we do next? And how do we start to measure that in an ongoing basis where I have to continuously running all these tests. And so they looked at our saturation plot, scaled that spend, and then he ended up running a holdout, a three cell holdout after.

and the results match perfectly. And so now he's validated YouTube with us as this channel and he feels really confident in scaling that channel. Now on the other side, a lot of these brands, pretty much every brand we onboard is using an MTA when they start using us, but they realize they need to use us to scale top of funnel. The majority are large enough and they're hungry for data and they're using this from a triangulation perspective to look at creative analytics within Side Triple Whale because we're only going down to the campaign level.

Jon Blair (33:37)

Yep.

Michael True (33:39)

But I have

Jon Blair (33:39)

Yep.

Michael True (33:39)

seen recently some of the brands that they're like, hey, I have such a good pulse on my Meta in my Google, right? Through the platforms and just running these channels for so long. Like, do I really need an MTA anymore? If I have such a good pulse, I really need to be focusing on YouTube and TV and making more of an investment and focus on the MMM side of the house. And so I've seen a blend of it across different brands. Some are taking the trifecta of incrementality, MMM plus MTA.

I foresee the future being more of an incrementality plus MMM approach.

Jon Blair (34:12)

Yeah, you know what? I I think I totally agree with you that regardless of the specific tools you're talking about that there's this triangulation approach, right? That we recommend the same thing. Brands ask us all the time, what tool should I be using? I'm like, well, I do believe that the answer is different depending on what channels you're spending in, what channels you're selling on. But either way, there's not a single source of truth, right? And generally speaking, platforms...

Michael True (34:16)

Thank you.

Jon Blair (34:40)

Triple Whale and the like can be great for getting really granular, like you said, like creative testing and measurement of how creative is doing, right? But at the same time, no matter what, and even if you layer in like Prescient AI alongside that, we still as CFOs need to say, how many contribution margin dollars are we driving every single day? Because that's the financial North Star is that like, as you're scaling up and down, right?

that we're actually producing more or less profit and we know that profit is going in the same or different direction than what these other tools are saying. And so there's a financial, there's like a true financial measurement which for us tends to be contribution margin dollars. You may get really granular with something that's not, you know, an MMM, but then have MMM alongside that and you can get kind of, I think about it as like a three dimensional view about how your marketing is performing, right? And so,

Michael True (35:31)

Nice.

Jon Blair (35:35)

You know, the point that I wanna make for the audience is that like, don't get trapped in thinking there's a single tool that's gonna do it all. It comes back to strategically understanding your brand, the channels you spend in, the channels that you sell in, and understanding what you're trying to gain and achieve out of scaling, right? Different brands have different goals, and so you should match your tech stack with

the strategy and ultimately the goals that you're trying to achieve. Because at the end of the day, all of these resources and tools, they exist as tools to help you get to where you're trying to go, right? And so, what I was really hoping we could do, which you've done a fantastic job, is get people to understand where MMM should sit in the tech stack, when it's maybe you're not a good candidate for it, when you are good candidate for it, and then,

Michael True (36:15)

Yeah.

Jon Blair (36:32)

really what else do you use alongside of it? And I think this is super helpful in that regard. What other, just like, what other, I'm assuming you've talked with tons of brands who are considering using your tool. What are some of the common misconceptions you see out there about MMM where like brand founders are regularly kind of it wrong in terms of assuming that the tool can do something that it actually can't?

Michael True (37:01)

There's a shift of focus and education that we've learned a lot from. A lot of these brands were never familiar with an MMM before this, right? They came in the 60s and when the internet came, like they went away, MTA was there. You know, you could track everybody on the internet. Taking the time to understand and work with us or any MMM is what is the model trying to tell you and how does the model work, right? And not thinking about it in a lens of an MTA.

right? Shifting that focus and saying like, what is the larger story here versus across all of our spend versus really trying to like break apart the model, right? Because there's statistical models, right? And so making sure that when you onboard one, you're in the right time to use it. And like I mentioned, like, we will not onboard in any brand that is not in the right fit for an MMM, right? And it's the right fit is going to be when

Jon Blair (37:31)

Yeah, yeah, yeah.

for sure.

Michael True (37:58)

you're aggressively looking to scale top of funnel. You're going omni-channel and, or you have, you scale top of funnel in the past and you thought it performed better and you want to go back and look at, you know, you want to look at those results from a more holistic angle.

Jon Blair (38:12)

I love it man, this is super helpful. before we land the plane though, I always like to end with a personal question. And so my question for you today is, what's a little known fact about Mike True that people might find shocking or surprising?

Michael True (38:15)

you

A little known fact about Mike True that might find shocking and surprising.

Wow, that's a really good question.

Jon Blair (38:39)

You heard mine earlier with the touring metal band, so...

Michael True (38:40)

Might not be surprised.

I enjoy jumping out of airplanes and I still get petrified every time I'm about to do it, but I like jumping out of planes. And shout out to Hans from Broomate. We went in San Diego at the Seminan Conference. This dude walks up with his own gear and he was like, this is my 320th time jumping out of a plane. Craziness.

Jon Blair (38:51)

man.

I still haven't done that

man. Kudos to you for being able to do that. That is, I don't know if I ever will. Maybe I will, maybe I don't know, who knows. that's, okay, there you go. There's a way to connect for sure man. Before we do, no pun intended, land the plane here though, where can people find more information about you and about Prescient AI?

Michael True (39:13)

Maybe in Austin. Maybe we throw a DTC skydiving event the third week in Austin. I love it.

Yeah, I've been very active on LinkedIn this year. So hit me up on LinkedIn. It's Michael, Michael True of Prescient AI. I'm on Twitter now. And then, yeah, if anybody wants to just learn about MMM in general, hit me up on LinkedIn or can reach out to me directly. It's mike@ prescientai.com and we got a chat and make sure you've got the right people to support you as you think about it.

Jon Blair (39:56)

Yeah, I highly recommend reaching out to Michael if any of this piqued your interest. What they're doing is really, really cool stuff. as you're scaling your DTC brand and you're expanding into additional channels and you're leaning more on top of funnel, there is a place for the tool that they're building in your marketing mix to really understand measurement. And again, be able to scale profitably with confidence. So definitely hit up Michael if you're interested in learning more about this.

Also don't forget, if you want more helpful tips on scaling your profit-focused DTC brand, consider following me, Jon Blair, on LinkedIn. And if you're interested in learning more about how Free To Grow's DTC accountants and fractional CFOs can help your brand increase profit and cash flow as you scale, check us out at FreeToGrowCFO.com. And until next time, scale on. Thanks, Michael.

Michael True (40:43)

Thanks, Jon. Appreciate you.

Sherilee Maxcy
Next

Podcast: The Secrets of Elite DTC Brand CEOs