The Methods and Math Behind Understanding Channel Performance With Michael Kaminsky, CEO of Recast

And if you get locked into this world of we can only spend money if we can track a click that's going to end up putting you on the wrong path and sort of in the Facebook marketplace death spiral of, look, we're just going, we're going to be trapped into a world where we're just bidding against our competitors for CPAs. And it's hard to grow a business that way.

I'm thrilled today to have our guest Michael Kaminski on our podcast here today. Michael comes as founder CEO of Recast. And Michael, thanks for joining us. Want to tell the listeners a little about yourself and about Recast. Yeah.

Howdy y'. All. Great to be here. Thanks for having me. Jason, I'm Michael. As Jason mentioned already, my background is in econometrics and statistics. Super passionate about statistical inference. Really? How can we use data plus software and code in order to learn things about the world? That's where my focus has been for all of my career. I was really lucky to get started in academia and then sort of work at a bunch of startups with a lot of really smart people. And for the last couple of years I've been working on Recast, where I'm combining that passion for econometrics plus software engineering to help marketers learn more about what is most effective for them. How can we use data to generate insights and then actually go out in the world and use those insights to drive additional profit for these businesses that we're working with? So at Recast, you know, we're in the family of marketing mixed models, which effectively means we use a top down statistical model to try to learn about incrementality for all of the different marketing channels that our customers are operating in. And yeah, I'm really excited. I've been known Jason for a while, really excited to be here. Huge fan of Simon Data. And let's get into it.

Awesome, Michael. So let's dive right into the big question. I think you said a word that I know is top of mind for many of our listeners today and that's profit. Yeah, the macro was changed. And for anyone who's caught some of our most previous episodes, we've actually had a whole line of conversations around this, starting with Aaron Federers at Transparent Partners where we talked about clean rooms. We had SRI and Albert on our last show where we talked about an incredibly interesting story about her journey with Bobby, you know, where she tried to understand, you know, which of her channels were effective and which weren't. You know, for those who didn't listen you know, to that episode, Bobby actually had a supply chain shortages where they actually had to shut off channels one at a time to understand what was effective and what wasn't, you know, and today we're going to be talking about and really unpacking, you know, how every one of you can better understand the channels in front of you and what your optimal marketing mix can be. You know, at the end of the day, you can log on to Google Analytics and Google loves to sell you search keywords. And search keywords are incredibly easy to understand, whether or not they're converting, you know. Yet at the end of the day, for most brands, a marketing strategy which, you know, just, you know, affects around last click attribution, you know, for search and maybe some social sparingly way, is not necessarily the best way, you know, to build a comprehensive brand strategy and demand strategy, you know. So today we're gonna be diving in, you know, to your tactics and strategies and how to think about these problems, you know, in a way that, you know, doesn't necessarily require that you shut off your marketing channels like Bobby did, one at a time, but instead can allow you to look at your business holistically, you know, and really just ask, why am I posting the Instagram for when I actually aren't driving any indirect sales? And how do I even think about that? Yeah. So maybe to kick things off, Michael, I know you gave a bit of a background for Recast, where it came from and your background, but maybe you can sort of just talk, maybe just give a bit of a founder story or a bit of context around sort of the mission, the vision behind Recast, and some of the core problems that you're trying to solve and help your customers understand.

Absolutely. So maybe I'll give the background on Recast or the slightly apocryphal founders journey here. So before starting Recast, I ran the data science and analytics team at a company called Harry's, which probably if they're, if you're a male in the audience here, you have surely been advertised to, at least if I had anything to say about it, they make men's grooming products. When I joined that company, it was an e commerce store, right. We sold online, we primarily advertised on Facebook. And Google had grown the business quite sizably just on those channels. It was great. Things were going well when I joined. I helped to build out the effectively the multi touch attribution platform at that company. Worked really closely with the marketing team, thinking really hard about, okay, how do we measure the effectiveness of our marketing spend. During my time at Harry's though, things got a lot more complex. The business expanded into other non trackable distribution channels, selling at Target, at Walmart, on Amazon, places where we can't put a pixel right, you can't put a pixel on the scanner at Target, you can't do it at Walmart, you can't do it on the Amazon checkout page. And so and the marketing team started advertising in more and more offline channels or sort of mid funnel, top of funnel channels as a marketer might say. So you know, investing into podcast and radio and TV and channels like that, where you know the multi touch attribution model that I had built, and I knew really well, we just recognized that it had flaws, it was going to over credit more bottom of funnel channels that are closer to clicks and purchases and also myopically only focused on the E commerce site. And so we sort of realized that the business was missing big parts of the full picture. And so I sort of wanted to think about well how do we actually understand that full picture right, where we can't track people across the Internet. And I started learning about these techniques called marketing mix modeling which CPG companies like Procter and Gamble, Gillette, Coke, Pepsi have been doing for 50 or 60 years. And I was like, oh this is really interesting, right? We're going to have a regression type model we're going to look at historically when we tend to spend more money on certain channels, do we tend to get more revenue? Find those patterns. I went out and I looked at building it myself and I was like this is really hard. I went out and I talked to a bunch of vendors and I wasn't really happy with what was on the market. They were using outdated methods from, you know, 30, 40 years ago. They weren't updating the model frequently enough to be responsive to the way that modern marketing works, which is that the platform effectiveness can change really quickly. And so I identified like look, there seems like there's a problem here, right? We need some way of measuring full funnel marketing effectiveness for online and offline channels. We need to be able to do that because of changes to privacy and because the future is omnichannel for retail, right? You're going to sell on your E commerce store, you're going to sell on Amazon, you're going to sell at Target. And so it felt like there was a real gap there. And I left Harry's, teamed up with my co founder Tom and we started recast to really try to address that gap, to try to Take modern statistical methods and apply them to this really, really hard problem of estimating incrementality of marketing investment for non trackable channels.

Yeah, no, I mean it's a great mission. And you know, ultimately the question that everyone's asking is, as I said originally, some channels are easier to track than others, but it doesn't mean that they're not effective. So maybe we can go into some.

Statistical methods and apply them to this really, really hard problem of estimating incrementality of marketing investment for non trackable channels.

Yeah, no, I mean it's a great mission. And ultimately the question that everyone's asking is, as I said originally, some channels are easier to track than others, but it doesn't mean that they're not effective. So maybe we can go into some specific examples. I think about out of home, I think about billboards, I think about tv. How does all that stuff fit in with my Google search budget? How should people think about this? And how does some of your analyses and your product help your customers unpack all that?

So it's a really good question. And I mean, look, I am a huge fan of Google search and Facebook advertising, or meta, let's call it right. But those two channels are incredibly powerful and there have been a lot of great businesses built on the back of those two channels. And I think if you are starting an e commerce store, those are the best places to start. The trouble comes when you start to graduate from being a small e commerce store to really trying to be a big brand. Right? And that's where you start to explore these other marketing channels. And that's where things get a lot more difficult. And this is a problem that we see a lot as we work with a lot of brands that started out, they built a decent sized business on the back of Facebook and Google and now they're looking to take that next step. They've already tapped out Facebook and Google. They just can't continue to scale up those channels anymore. They need to expand to other marketing channels.

So.

And they really struggle with that because the only thing they know is last touch attribution. Right. And once you start expanding into podcast and radio and TV and influencer, Last touch attribution doesn't really work anymore. And so this is a real challenge for marketers to try to cross that chasm to, okay, now that we need to get out beyond those two core channels that we built this business with, what are we going to do? How are we going to measure this? And I think sort of recognizing that things are different in that world, once you're in sort of mid funnel and top of funnel marketing, you're going to need other techniques. The things that got you here won't get you there. Right. Is a sort of a phrase that I that comes from a book, but that I like to share because it's really important in this context of, okay, we need to do something else. We need to branch out beyond what we've been doing before and actually figure out how are we going to measure channels that don't have clicks. Right. And there you start to explore more advanced methodology. It's really convenient to be able to go look at the Google Analytics panel and be like, look, the CPA for this keyword is X and the CPA for this Facebook campaign is Y. But once you get into channels where you can't do that, you need some way of bringing that story together and actually making smart decisions about how you're going to allocate your budget in this new world where things can't be measured quite so easily.

Yeah, no. And I'll sort of share a personal anecdote from a good friend of mine. And I think in the early days of startups and direct to consumer brands, yes, you rely on Facebook and Google, but you also have that buzz from your early adopters. And I think part of the challenge is as you widen up your addressable market and try to reach a new set of customers, you need to build that brand base and point of awareness. One of my good friends, he now runs a very large and successful, you know, B2B SMB SaaS business. But he launched in New York City and part of his launch strategy was to buy some subway. And, you know, the launch was successful. You know, this sale process included Google Ads, but also included a sales process where you actually talk to someone for 30 minutes, you know, and before you buy the product. And I remember, you know, when he went to launch in Boston, the next cities, he asked his board member, he's like, look, I don't know if these subway ads and these out of home ads are working. They're really expensive. You know, they're hundreds of thousands of dollars. I think maybe the next cities, I'm.

Not going to bother.

The board member responded. And he was like, look, if the launch in New York City worked, do the same thing that you did in Boston. Now they're in 100 cities across the country and this is their plan. And they've certainly gotten a lot deeper around analysis. But the fact of the matter remains the halo effect of doing proper Upper funnel awareness, you know, is totally critical. Yet at the same time, the decisions that people use, you know, the data that's used to make those decisions rather is so, you know, associated past. You know, it's like, well, like, I guess it worked before. Like, you know, like, you know, and again, this great, you know, happened, you know, you know, five, six, seven years ago when you, when the capital markets were a lot different than today, it was like, obviously you're going to do it. And today if you have the conversation, it's going to be actually a lot different. Like, I spend $400,000 to advertise on Boston's tea. Or are we just not going to because we think we can get it done without it?

This is like, this is such a hard, this is such a hard question for everybody. And I think there's a bunch of interesting points here. So one is, if we think back to the history of marketing, the last 10 years are really the anomaly, right? If you put yourself in the shoes of a CMO of Pepsi in the 1980s, they had to do everything sort of by the seat of their pants, right? Because there just was no other data. There's no E commerce, there's no clicks to track. You have to take these big bets on out of home or new channels or print or tv and then do your best to stitch the learnings together, run experiments where you can. But this idea of like a direct connection between, you know, we showed an advertisement to Bob and then Bob made a purchase two days later, didn't, didn't exist at all. Like it was inconceivable. And so really it's the last 10 years that have been the anomaly. And I actually think that it's part of it has been really good. Obviously, like a lot of great businesses have been built, but it's also, I think, weakened the capacity of a lot of marketers to really take these big bets and big swings and think holistically and strategically about how does our marketing fit into this bigger picture. And your anecdote here, I think is a really good example of like, you know, it seemed to work once. Why would we like, let's, let's double down, let's click into it, let's see if we can do it again. If it doesn't work, we'll have a learning. If it does work, we can sort of see if we want to keep expanding it. Does it only work with launches? Does it work at other times? And those sorts of like bets and swings of, look, look, we're going to Go try tv. We're going to see if it works. We're going to go run an experiment. We're going to see if, you know, podcast is going to work for us, I think is a really healthy muscle for marketing teams to build. And if you get locked into this world of we can only spend money if we can track a click, that's going to end up putting you on the wrong path and sort of in the Facebook marketplace death spiral of look, we're just, you know, going, we're going to be trapped into a world where we're just bidding against our competitors for CPAs and that's. It's hard to grow a business that way, at least in today's world. Right. Ten years ago it was possible when the, when the cost per impression was a lot cheaper on Facebook and Google. But now it's so competitive that if you can't make decisions without a click, you're going to be in trouble. And it's just going to be really hard to grow the business from where you are today. Because all of the margin gets competed away on the hyper efficient Google and Facebook channels.

Yep. No. A hundred percent. Yeah. So Michael, I don't know if you have any specific examples of channels or learnings that you've come across over the past years. Maybe something from Harry's or something from Recast where it was sort of like, oh, wow moment, this is actually working. And here's a new way of thinking about my channel mix.

So it happens all the time with our customers. And I think every brand is different and all of their marketing strategies are different. And so I wouldn't necessarily say there's patterns that we see, but we've seen tons of examples of customers sort of going both directions in terms of recognizing that something is working or isn't working. We've worked with a customer that was spending a lot of money on streaming tv. Their streaming TV vendor was claiming credit for a lot of conversions. And then we ran their data through Recast and the Recast platform identified that look, when you vary your TV span, your number of conversions does not really seem to change very much. Right. It seems like there's actually not very much of a connection here. They took down their streaming TV spend basically to zero. They turned it off and number of conversions stayed flat over that entire course and it really was doing almost nothing for them.

They.

And that's like, you know, we're talking high six figures a month type of budget on this channel and it was just like that was net savings for them. That went straight to the bottom line because it really wasn't doing anything for them, even though their vendor was trying to claim a lot of credit. We've also seen the opposite, right? We've seen other companies where, you know, they, they've, they're trying to step into tv, they're worried that it doesn't work because they don't have clicks, they don't trust the vendor, but it really is highly incremental for them. And this is, and that's like, that then becomes their next channel that they can actually scale into meaningfully after they had tapped out Facebook in Google Search and weren't able to scale into that anymore. So we've been able to see all of the sort of the full range of outcomes here from look, this channel you're under crediting, but it's actually a lot more incremental for you than you think. Pinterest is a good example, actually. That's a pattern that we've seen now a couple of times where Pinterest, I think, builds a lot of awareness, at least for some brands that are especially advertising to like the Pinterest customer or the Pinterest user, where for a bunch of different reasons, I don't think that that channel was getting full credit and they were able to expand their advertising into that channel fairly substantially, grow the business that way. And so a lot of like the work that we're doing with these, with these companies is identifying those like small, small, sometimes very large areas for improvement and then helping them grow the business by either cutting spend from their less efficient channels where they don't have insight into or where they're not actually getting very much incremental lift from. And then identifying opportunities to grow into channels where there actually is room for that incremental lift 100%.

And sort of another area at least speaking on behalf of Simon Data, is we're always aggressive around retargeting and even if it doesn't lead to people submitting a demo request, I always hear from people that, oh, we see your ads and you guys are top of mind, we see you everywhere. So obviously maybe we should have a conversation, Michael, around doing some analysis on our data. I think really what we're talking about here is the bridge between that intuition and the direct customer feedback into the raw numbers, the data and the statistics that can validate what I believe might be true to things that we can actually statistically verify. Hey, this is working.

Yeah. And I think, you know, again, I think good marketers take a test and learn approach Right. They are willing to take big bets and a tool like Recast can help you identify. Look, there's potential in this channel. Let's go take a bet on it. Right? Let's go. Let's plan next month, April, we're going to triple spin into the channel. And it's a risk, right? There's a risk that it doesn't go well, that, you know, the statistical model didn't, didn't hit it perfectly. And we're going to learn something on the other side. But the value in the test is we're going to learn something on the other side. Right. We think it's a good, we think it's a, it's a good bet to take and we're going to go. If it works out, it's going to be worth, you know, millions of dollars of extra revenue for the business. If it doesn't work out, well, we'll have, you know, spent a few extra dollars on marketing. But that also shouldn't be valued at zero. Right. We'll have gotten something for that. And so all of those different points I think need to, need to come together and marketers, CMOs especially need to build that muscle of thinking in bets, being willing to take risks and being able to generate the institutional buy in for that. You need to be able to go to your CEO and the finance team and the board and say, look, we're not sure what's going to happen. There's a big opportunity here. We think it's worth investing in number of thousands of dollars in order to go test it. And if it works, it'll be a huge payoff and if it doesn't, we'll be out a little bit. But the only way that we're going to grow the business is by placing smart bets over and over and over again. Like in the days where you can just dump a ton of money into Facebook and then, you know, build a hundred million dollar revenue, business on top of that are over. It's just like that's. Those days are gone, they're never coming back. And I'm sorry, but the future for marketers has to be on figuring out what the next channel is and then taking risks. How, what is our influencer strategy going to be even if we can't measure it as precisely as Facebook? And you need to have an answer to that question. And it's a hard one. It's uncomfortable. Especially if you came up in the last five or ten years where you could do everything based on Last touch attribution. But, but Again, I just think that the future is in these less trackable channels. And the marketers that are going to have success and the brands that are going to have success in the future are the ones that can get comfortable with how can we take bets, do the best that we can on measurement, even if we don't necessarily have all of the clicks and all of the impressions and tracking people across the Internet.

And look, what you're making is actually an incredibly subtle point. And I think one of the biggest challenges with the macro is, is that there's so much pressure around you. Make make sure that you're show you have clear return on ad spend, make sure that ROI is 100% clear. And look, you should only be spending on channels. You know, your direct response response channels have to be working. But at the same time, look, if you double down your direct response channels, you will have no brand halo, you will have no brand awareness, you have no customer education around what you do. You know, that's, you know, either differentiated or valuable. Really the expectations in this new world is one of coming up the other side with an incredibly efficient business that can affect customer acquisition costs that are fully sustainable. So while if you took a strategy today where you were incredibly myopic around optimizing for last quick conversion and backed off the gas and everything else, you may work for a couple of quarters, all the work you've done over the last year in building your brand will be for nothing. So really the question is, how do you balance the two and how do you think critically around what you're actually doing? And how do you fast forward one, two years out in identifying that right channel mix that permits efficient CAC in 2024 and 2025?

I think that's exactly right. And I mean again, Facebook and Google are incredibly efficient channels. Unfortunately, that works for your competitors just as well as it does for you. And so if you get to this world, I mean, if your product is not super differentiated, and again, there are a few products that are like that, but most products face competition from other similar products, especially in the consumer space. And what ends up happening is that two or five or whatever brands are all on Facebook and Google and it just drives the CPAs up. There's just limited room for everybody to play. Google and Facebook end up eating everyone's margin. And so if you're a brand and you want to figure out how to get to that next level of growth, you just can't assume that those channels are going to continue to scale for you because your competitors are going to come in and they're going to try to eat your margin the same way that you did to, you know, whatever brand you disrupted when you came onto the market as well. And so, again, you can get some efficient conversions, but it's just if you flash forward two or three years, those markets are so competitive. They're just so competitive. It's hard to win consistently there when you have to compete against all of the other competitors that are doing the same CAC and LTV math that you are. And so what is your plan going to be when all of those cost per acquisition numbers get driven up by all of the competitors bidding after your same audience? And that's the question that CMOs need to be thinking about today, is how are you going to do that a year from now when more competitors enter, when these platforms are even more saturated than they are today and tracking is worse? And those are hard questions. Being a CMO is hard, but that's, that's the job, I think, right now is figuring out what is the plan, what is the strategy for that future world.

Yep. No, that's 100% right. So, Michael, we're approaching the bottom of the episode. You had time for the bonus question? Yeah. If you were to have this conversation again with anyone in the world, who would it be and why?

This is a good. This is a. I mean, this is a really interesting question. I'm not sure. I think I would love to have this conversation with like the CEO of the CMO of Nike. I don't know who that person is, but someone like that, who is. They have such a massive brand. They advertise on all of the channels, they sell through all of the different channels. How do they think about measuring marketing? Right. What do they think the future of marketing is going to look like? I think that's such an interesting and unique perspective, at least for me to, like, learn from and understand. You know, how do they think about measurement? How do they think about defending Nike's brand in this sort of continually changing world and where there is lots of competition from both upstart disruptor brands as well as from, you know, the platforms themselves just changing. And how do they. How do they think about managing all of that complexity? That I think would be a fascinating conversation and would. Would be a thing that I would love to hear at least again from these massive brands that have been doing marketing for a very long time and have been thinking about it very hard for a very long time. Where do they think the future is? How are they thinking about future proofing themselves from all of the disruptors that are trying to come and eat their lunch as well.

Yep. And, you know, I'm sure if you were to have this conversation with Nike, you know, the number one brand defining moment that they've affected, you know, in the last 30 years, now, 40 years now, must have been signing Michael Jordan. You know, you have an entire product lineup, half of all the shoes they sell, how Jordan's name, you know, and it's completely redefined, you know, and defined what the brand is today.

And I'm sure the point would be, like, there was no measuring. Like, they didn't have a measurement plan in place.

They took no last quick attribution.

There was no last click attribution.

Yeah, no, that's. Yeah, that would be fascinating. Cool. Well, we'll have to, you know, hunt down, you know, someone from Nike for a future episode. Michael, thank you for coming on the podcast today and sharing your wisdom with us. Really, when I reflect on the conversation, it's ultimately about one of efficiency and about where to spend. And as I said a few minutes ago, you know, it's about, you know, thinking about the funnel from top to bottom. Today you have a whole bunch of folks who are at the top of the middle of your funnel who have a degree of awareness. You have a degree of enthusiasm about what you do, and you have a degree and a capability to convert some of them. You know, but ultimately it's about, you know, thinking across the entire spectrum, you know, from that first touch point with the brand to the first purchase. And we even talk about, you know, second purchase and, you know, the broader customer journey beyond that. Yeah. But ultimately, putting it all together is what drives effective marketing and effective product adoption. Yeah. And if you're not able to measure that and understand how the different constituent pieces work together, then you just won't have a holistic strategy and you won't have holistic customer experience. And you certainly won't, you know, have a very difficult time meeting some of your efficiency goals to affect the customer, moving through the final and seeing the value that they'd otherwise hope to see. And so, Michael, I'm sure at least some of our listeners might be interested in learning more about Recast and what you do. How can they find you and how can they contact you?

Yeah, absolutely. So they can always check out our website, getrecast.com we just published an ebook, 50 pages, on how to actually use marketing mix modeling at a modern brand, how to operationalize it. So I highly suggest that they go check that out. And they can always sort of connect with me on LinkedIn or send a demo request if you're interested on our website.

Awesome. And thank you to everyone listening to this episode of the Data Unlocked podcast. If you'd like to learn more about Simon Data, please visit us on the web@simondata.com or email us at helloimondata.