Become a Creator today!Start creating today - Share your story with the world!
Start for free
00:00:00
00:00:01
The Future of Marketing Measurement with Rockerbox | Ron Jacobson
59 Plays8 months ago
SUBSCRIBE TO LEARN FROM PAID MARKETING EXPERTS 🔔
The Efficient Spend Podcast helps start-ups turn media spend into revenue. Learn how the world's top marketers are managing their media mix to drive growth!
In this episode of the Efficient Spend Podcast, Ron Jacobson, co-founder and CEO of Rockerbox discusses the evolution of the company, their approach to optimizing media spend, and the importance of accurate measurement in marketing. He also shares his insights on the challenges and the methodologies of multi-touch attribution and the impact of privacy regulations on data sharing in the advertising industry.
About the Host: Paul is a paid marketing leader with 7+ years of experience optimizing marketing spend at venture-backed startups. He's driven over $100 million in revenue through paid media and is passionate about helping startups deploy marketing dollars to drive growth.
About the Guest: Ron Jacobson is the Co-Founder and CEO of Rockerbox, a premier marketing measurement and attribution platform. As a technologist and marketing measurement expert, he has transformed how companies approach marketing attribution and data-driven decision-making.
VISIT OUR WEBSITE:
https://www.efficientspend.com/
CONNECT WITH PAUL:
https://www.linkedin.com/in/paulkovalski/
CONNECT WITH RON:
https://www.linkedin.com/in/rjjacobson/
EPISODE LINKS:
https://www.rockerbox.com/blog
https://www.strava.com/
https://zapier.com/
https://www.amazon.com/Ogilvy-Advertising-David/dp/039472903X
https://www.reforge.com/
https://gdpr.eu/
https://www.apple.com/privacy/
https://ai.stanford.edu/
https://www.csail.mit.edu/
Recommended
Transcript
Privacy and its Impact on Elections
00:00:00
Speaker
ah Privacy is not only becoming an issue in and in our industry, it's kind of become something that is you know found itself in political elections, right? So we have more regulation coming and more eyes on this stuff. With these two things in mind, it would seem that partnering would be helpful with different ad networks, things like that. I think there are certain components or stages of that kind of sequence of events that AI can be really, really beneficial in, particularly in some of the so what part of of of that of that ah kind of five step process where we're already starting to see a little bit of a rockerbox with some of our
00:00:35
Speaker
behind the scenes work that we're doing. I mean, I think it comes back to like what we're just trying to accomplish as an organization. We want to build the world's best measurement platform. To me, that means you have all the right data on top of it. You have all the right analysis on top of it. I think the next step that we could do a better job at is ease of insight and ease of actionability on
AI Enhancements in Ad Networks
00:00:51
Speaker
top of that. I think those are two areas that we're not we're not doing as good a job as I'd like us to do. But if you think about it, imagine you know waking up ah on a given day Whatever question you want answered, you're able to easily get the data automatically assembled together. You're easily able to get a handful of suggested actions taken off that. And those actions are automatically actually orchestrated across all the platforms that you're using. And the feedback loop continues over and over again.
Introduction to Rockerbox with Ron Jacobson
00:01:20
Speaker
Welcome to the Efficient Spend podcast where we help marketers turn media spend into revenue. My guest today is Ron Jacobson. Ron, how are you? I'm doing well, Paul. How are you? I'm excellent. um Thank you so much for for being on the podcast today. Would love for you to give users a quick overview of your experience with optimizing media spend at Rockerbox. Yeah, so um I'll give a quick background on Rockerbox because it'll kind of help give some color. I'm co-founder and CEO of Rockerbox. We're a marketing attribution SaaS platform. um We work with hundreds of brands like Staples, Warby Parker, Los Hotels um to essentially help them measure all of their marketing.
00:01:57
Speaker
across paid channels, organic channels, ah digital, and even offline. So that could be from somebody running ads on TV, to sending mail to your house, to podcast ads, to that bottom-up funnel, clicking Google, that drives through a conversion. Yeah, we work with we're really fortunate to get to work with hundreds of the best marketers out there in day-to-day optimizations. How do they get the highest ah ROI out of every dollar being spent? um So this is ah very close to to our heart at Rockerbox. If we do our job well, I say this internally all the time, if we do our job well, we are literally giving information to our customers that enable them to grow and scale their businesses profitably. So it's it's a really big responsibility and we take it very, very seriously. But um yeah, it's... ah optimizing marketing dollars all day long. You have been doing this for a while. Could you give some context into what the growth of Rocketbox looks like over time, maybe in terms of both client mix and also product development and expansion?
Shift in Focus: Attribution over Ad Management
00:02:52
Speaker
Yeah. So we launched our attribution business five years ago. Um, before that we actually manage ad dollars. So for five years, we had a different business line completely where we were day to day in the transit, you know, in the trenches, um, running programmatic advertising campaigns for,
00:03:06
Speaker
thousands of brands. At that point, it was like Vanguard and you know Citibank and Chase, so some pretty some pretty big companies. We ended up getting into the attribution space um because, frankly, when we were running media, we felt like we were never getting the signal that we needed to optimize against. We felt like the brands that were looking to rockerbox to manage their ad dollars were judging us with tools that actually didn't tell them the full story. um So I felt like the brands were being disempowered. And then myself, as the media buyer, I would get reporting from the brands on a weekly basis, one row in and ah in ah in an Excel file, saying how all of my 30,000 ads were performing. And I'm like, how am I supposed to use that as any signal to optimize again? So that led us to launch what at the time was called Rockerbox attribution
Media Mix Modeling and Data Foundation
00:03:52
Speaker
platform. That was five years ago. um Our first product was multi-touch attribution product. We ran and grew that for around three, four years. ah Two years ago, we launched our first media mix modeling product. And this all comes from our thesis that what we see is at scale, most brands end up using multiple types of measurement. They use MMM and MTA and surveys and holdout tests and benchmarks. And we want to, we want to build a platform that enables all of that. The last component of it is having a really, really solid data foundation.
00:04:18
Speaker
um You can't do measurement unless you have all of the data, both user level and aggregate, cleaned, centralized and normalized. um And even with consistent taxonomy, knowing that prospecting and Facebook might mean the same as branding and Google, how do we normalize together those two things so we can actually provide analysis. So that's where Rockerbox is today, a data foundation with modular measurement on top of it. I want to dive deep into the the measurement today. Before I do that though, in our previous conversation you had compared um kind of measurement solutions to weighing yourself and using a scale. ah Can you kind of talk through that? Yeah, it's funny. A lot of customers come to us and they look for rocker boxes sort of be this silver bullet. ah They say, hey, you're measuring your market you know you're we're hiring your rocker box to measure our marketing. That means you're going to help us find what's working and grow our business really well. The problem is there sometimes isn't anything working. It could be that all the marketing isn't effective. It could mean that the company and their brand doesn't actually have a great product to market. It could mean the price point is too high. The analogy I was kind of using, it's it's like going to your scale and seeing that, you know,
00:05:25
Speaker
Hey, I tried this diet last week. A week later, I go to my scale, my weight hasn't changed, and yelling at the scale because the diet isn't performing. It's not the scale's job to figure out why the diet is working or not. It's to be that source of truth, to analyze it. It's figuring out what the right diet is, the right exercise. That's a whole other beast, but you want that scale there that's independent, that's neutral, that doesn't have a stake in the in the game to just be there, to really be the barometer so you know what's working and what's not working. That's why I think the good the job Well, I think that's what the job of measurement should be. I think at times marketers potentially, either rightfully or wrongly, I think probably wrongly, but um think that the job of it is to go beyond just the scaling and actually to be prescriptive of here's the diet, here's the exercise program. And if you do all these, you're going to have X result. If we're giving that position, it's hard for us also then to be a neutral person, arbitrating what's
Personal Experiences with Measurement Tracking
00:06:16
Speaker
working, what's not.
00:06:16
Speaker
Sure, sure. Measurement can be very objective. um i love I think it's Paul Graham that said what gets measured gets managed. And it's funny, even in my own life as I've grown to be a better marketer, I find myself measuring more things in my personal life and I'm tracking different things. And it helps me, it holds me accountable too. Yeah. It's so true. i One year I said I was going to run 1,000 miles in a year, and I made this amazing spreadsheet with Zapier between Strava and my Google Sheets, and I tracked it every single day, and I hit the goal with one day to spare it in the year. The next year I go, well, you know, now I'm a runner. I'm just a runner, so I don't need to track this anymore. I'm just going to naturally run 1,000 miles ah ah you know a year. Next year I ran like 110. So without tracking, you're right. it just
00:07:00
Speaker
you It's so true. And again, that's why like ah we take what we do so so seriously. If we're providing these measurable, trackable insights to our customers, we think that they have the ability to really grow and scale their business efficiently. Speaking of measurement and and tracking, I want to ask you a very pointed question, which might be a little clickbaity, but it's something that um a lot of folks are talking about now.
Evolving Multi-Touch Attribution
00:07:24
Speaker
Is multi-touch attribution dead? I think it's more alive than ever. And again, I'm probably biased, but I generally think it's correct when I'm saying. I think the notion and the way that MTA was first envisioned five to 10 years ago as a methodology of tracking every single marketing touchpoint that leads the user to convert by leveraging third party cookies. I think that concept, kind of the original notion of MTA is also the original sin of MTA. That is a false,
00:07:55
Speaker
ah It's a false methodology. It's a false conception for how you can even think of tracking every single touchpoint. I think if you look at it from first principles perspective, you know Is there a third party cookie when you watch TV? No, there isn't. Is there a third party cookie when you listen to the radio? Is there a third party cookie when you get mail sent to you? All these different areas where we're spending a lot of dollars and have nothing to do with third party cookies whatsoever. So that that initial foundation was a false premise. So I kind of reject even the question, even though everyone's asking and I completely think you should ask it. I think the way to think about it today is
00:08:28
Speaker
If you were building MTA from the ground up, how would you do it? um You would use first party data. That means you'd capture phone numbers, email addresses, physical mailing addresses, whatever information you can about your customers in an opt-in privacy compliant methodology. And as much as possible, you would leverage that data to connect the dots between all the marketing channels. I'll give you a great example. Rockerbox is able to measure the fact that somebody viewed an ad on Pinterest or Reddit or Snapchat and then came to your site and converted without a click. We're doing that because we struck deep partnerships with those partners that enable us to join their data with that first priority I was just talking about, again, in a privacy-compliant way. That's that's significantly better. That's significantly better than using a third-party cookie to try to track a view that may have happened on a computer with a conversion that happened on a phone. um Those are two different devices. Third-party cookies won't go across those two devices. Your email is your email on both devices.
00:09:17
Speaker
It's just an example, but I think approaching MTA from the perspective of first party data, ah lever leveraging modeling where necessary, there is a modeling component to some channels that just don't have um really good granularity to it. And also recognize, you know, really strong identity resolution, but all that to be said, MTA is not perfect. Other methodologies are not perfect. I think what you need to know is what is the question you're looking to answer? What is the methodology that is most geared towards answering that question and understanding the drawbacks of that methodology? So when you take the results, you recognize that it's directional. It's not gospel. It's not the word of God. So that's my take anyway. There is a level of belief with a lot of this stuff. Curious ah on that your your point about partnering with with different networks. so There's kind of two trends that are happening right
Challenges of Privacy Regulations
00:10:06
Speaker
now. Privacy is not only becoming an issue in and in our industry, it's kind of become something that is
00:10:13
Speaker
you know found itself in political elections, right like this this idea the relationship with tick-tock in china for example, so we have more regulation coming and more Eyes on this stuff and at the same time we know that more and more people are adopting cross-device using multiple things They have multiple phones, they have laptops, TV, etc. With these two things in mind, it would seem that partnering would be helpful with different ad networks, things like that. But at the same time, ah you know the large players want to control their data.
00:10:53
Speaker
How does that play out with like Facebook, Google? right Do you think that they're going to want to share more of this stuff um to give advertisers more access and kind with a partner like Rockerbox or how do you see that going? i mean Listen, you said Facebook and and Google. So you picked the two largest, most entrenched winners of the ad market to date. I think they have little to no incentive to change anything if they could. For them, Stasis is victory. um So no, I don't expect any day for Mark Zuckerberg to be called. Well, he's not going to call me anyway, but I wouldn't expect Meta to come and say, hey, let us share more first party data with you.
00:11:32
Speaker
um So what have we done there? like I think that's where we have to take an approach that some data sets have to be modeled against. um And it's actually interesting, even if you look at what Meta and Google are doing themselves in some of their in-house attribution, they have this whole concept of modeled conversions. Because even they are starting to face a gap. you know Apple with iOS 14.5, suddenly you're not able to get access to 100% of usual data. That means you're not able to ah connect 100% of impressions to conversions. So even for like a meta and a Google, if they went from being able to connect 100% onto 60%, they're modeling that 40 delta. So even the the biggest, most entrenched players have to do some form of modeling to fill in
Probabilistic Modeling and Platform Partnerships
00:12:09
Speaker
gaps. That's what a rockerbox is doing for channels like Facebook and YouTube. ah We actually have products in market around that. We call it synthetic events. It enables us to model the impact of views from an aggregate data set against the usual data set.
00:12:21
Speaker
It's actually interesting, though, too, because we got into this not because we were trying to solve the problem of how do we measure facebook ah views on Facebook and Google, but it was how do we measure linear TV. So we first got to this problem of on linear TV, unless you're procuring ACR data sets, which we currently aren't. The only data set available are called post-log reports. I don't know how familiar you are, but essentially, CSVs saying ah where and when and what channel an ad was served. I always say 1230 in Tuscaloosa, Alabama on planet Earth, an ad was served. So we would get those from our customers as aggregate data.
00:12:55
Speaker
We had to build a model to figure out, fine, based off that and based off what we see with our first party pixel, can we probabilistically determine which users that arrive on site do we believe ah were impacted by TV? That muscle that we had to build, taking aggregate data and connecting it to user level, is the same muscle that we ended up using for Google, for views on google for um google sorry views on YouTube and for views on Meta. So in a way, I think you're right. I don't think the largest players are going to share their data. Why would they? Having said that, I've been extremely hardened that we've been able to get, again, Snapchat, TikTok, Roku, Pinterest, only missing Reddit, to name to name a a handful of really big players that I would have never thought would be willing to share this information. They're partnering with us, partnering with us very heavily. Again, privacy compliant opt-in methodologies but um enable that kind of give them a give them the opportunity to compete in this market. And frankly, they're doing a really effective job at it.
00:13:46
Speaker
What extent do you think calibration can play in individual publisher data to help you get a more holistic sense of contribution? Obviously, and MMM plays a role, multi-touch plays a role. It becomes so hard, even for, ah take the brand that I worked for, CELF, for example, we have acquisition on both web and app, we're running a bunch of different channels with a bunch of different measurement methodologies. So how do you normalize? You can normalize with an MMM, you can or you can normalize with incrementality tests. And that gets you part of the way there. But then the question becomes, what do you make of the attributed data? The CPA for iOS is going to be 5x of that what it is for web, simply because of the lag and all the issues with but scan. So what ah level do you think of like calibration should
00:14:37
Speaker
Should play and or adjustment should play to like, you know attribute a data for example Yeah, um I think that's kind of the billion dollar question of our industry these days to be frank um I'll say we personally rockerbox. We don't work primarily with app only ah advertisers um So companies that are exclusively trying to get app downloads that really relied MMPs. You're right. There's a whole different measurement issues there that I'm um but At least from my perspective. It's probably probably not the best suited to dive into there um I would say, like listen, there there are a lot of approaches that people are starting to kind of posit, right? One is you know you have your day-to-day attribution. You have your incrementality test that you're running on ah on a given cadence. You have your MMM that you're running on a given cadence. Mind you, the cadence of both of those are getting faster and faster. ah Mind you, there are costs to running, um let's say, holdout tests and geotests. So I think it's important to understand the pros and cons of all these.
00:15:31
Speaker
I would even removing those though, like I think on the calibration front, we've been doing this for years, like ah just on the MTA side. And I'm happy to talk about our MMM product as well, because we're really proud of what we built there. But even on the MTA side, look at ah the direct percentage, right? What percentage of conversions are being attributed to direct? I personally, I mean, we have some numbers in the past, we saw an average 21% for our clients. That seems really high, like 21% of your customers just happened to be going and clicking on your website, having never heard of you before, like something had to, either you're really lucky or some somehow you're getting in their mind. We already do some calibration there, so for years we've been thinking about how can we use post-purchase surveys as a way to calibrate? How can we use promo codes as a way to calibrate? How can we use vanity URLs? a bit more you know A bit more obvious, but still, all ways to try to say, you know what
00:16:16
Speaker
Can we take that 21% and lower 10% by using other signals? What I'm trying to get at is, anyway we had really good success there too. On average, we got from 21% down to around 10.5% by virtue, like literally kind of 50% by using just those signals, plus identity resolutions, some of the partnerships, et cetera. But all that to be said is, is that like the, is that methodology, um like is there a scientific rule saying that's the right way to do it? Absolutely not. Like it's not A squared plus B squared equals C squared in terms of like, this is like a rule of physics, but it's how do we take you know Best intentions, best of data that we can get into our best job. oh All that to be said, calibration is something that everyone's going to have to do now, but how do you do it? um Are you taking into account an incrementality test 100%, 50%? Are you really going to take into account an incrementality test from Q1 and Q2? Are you going to take that Q1 test in Q3, that Q1 test in Q4?
00:17:04
Speaker
um I'm seeing brands do that. Is that right? Probably not. Do I get why they're doing it? I do. like It's hard to run tests all the time. So i just don't think there's a I hate to say, I don't think that there's an easy answer here. There is no silver bullet. I think there's a lot of work we all have to do in the industry to try to get um a bit more concrete with like best practices on this.
Google's Insights and Advertiser Methodologies
00:17:22
Speaker
Google actually just launched a white paper that was pretty interesting in terms of their perspective on it. Mind you, it's from the Google perspective, too. So I think we have to take that with a grain of salt. They have a bias. But but I thought it was actually a pretty good first take. And um I was glad to see them actually get a market with that. Yeah, it feels like there are there's a lot of triangulation you need to do. If you took a look at you know all advertisers globally, there's going to be a distribution of measurement expertise and sophistication. right I would think within your client mix, you work with some of the more sophisticated advertisers and maybe you work with some folks that are just getting started.
00:18:01
Speaker
um I wonder though, if you take a look at kind of the the most sophisticated clients, what does their measurement stack look like? What do you think is like the thing that... What's the North Star that advertisers should be striving for? And frankly, that's the direction we've been going for the past couple of years is working with more and more of these um larger enterprise sophisticated brands with really complicated um businesses to be frank. you know They have web-based conversions, they have digital conversions that happen offline. That's like an example I always use is you apply for like a loan, ah you know a mortgage on the website, and then you find out a day later you got the mortgage. That secondary conversion that's not happening on the website, so it's not client side. um We have retailers with their own and operated, we have them selling through wholesale, ah we have them with mobile app as well. So how does a company that sits across all those different um distribution channels think about marketing measurement?
00:18:51
Speaker
What we see happening is, at scale, um fundamentally, they they are running multiple measure methodologies. um They run MTA, they run MMM, they run geo holdout tests. There are varying ways in which they go about each of those. um Some companies will take certain components of that in-house. On the MMM side, particularly recently, we've seen a lot more using the open source tools that have emerged to build Beatemix models in-house. um I think that's been fine for one or two runs of a model. I've seen it be hard for a brand to build that into a consistent, recurring muscle organizationally. On the geo holdout testing side, I've seen a lot of brands also either leverage partner, um you know, every platform offers them for free. So a lot of brands will leverage the the free test that you know a meta or Google or Snap or Pinterest will will offer. A lot of other ones will actually build analytical capability in-house to design and
00:19:46
Speaker
um design, run, and actually determine the results of their tests. Mind you, for both of those, MMM holdout testing, they're also great companies that they can go to. So I think everyone's kind of making this decision. Do I ah do i build? Do I ah partner? or Partner, in the sense, would be leverage, I guess, with the platforms to provide themselves? Or do I or do i buy? um But again, at scale, I see them all leveraging at a minimum MTA, MMM holdout tests. uh, surveys, post-purchase surveys are, um, table stakes. Um, and they have a data warehouse that last part is actually beyond key. Um, it's not enough to, uh, have all the state and disparate locations. They are, uh, you know, uh, spooling up by their, uh, big query redshift, you name it. Um, uh, obviously snowflake, um, instance where they want to store a host of different marketing data sets. That's platform reported numbers.
00:20:35
Speaker
ah that's centralizing all their marketing spend, that's path to conversion, that's non path to non-conversion, and that's session data. At a minimum, those are kind of like the core five datasets that most brands want in-house, and they need an analytic team ah to support it. um That's kind of the best-in-class stack that we see these days. Plus well, there's more but I mean you could talk about reverse ETLs and stuff to push that push those results out They have some cappy infrastructure as well. There's there's there's a lot that goes into the Cisco Yeah for sure. And I think the point that you make is taken at scale um The fact that when you are a young startup just getting off the ground there are certain things that you don't need to be doing and you don't have to worry about this and you know, you're trying to get customers, right, right, right and That goes back to the whole thing about the the scale, right? like Until you actually have any form of diet or exercise that can help you lose weight, um there's no point in trying to... you know I'm going to butcher this analogy here, but like there's no point in going to measurement until you have something working. You don't need ah any third-party anything. Frankly, don't even use GA. Just look at the platform numbers. Until you see something trending in the right direction, there's no need to spend money on someone like Rockerbox or any of my competitors, I suppose.
00:21:41
Speaker
Right. um It's kind of akin to when you're starting ah a sport. like I used to play paintball as a kid and um you would see kids just getting started and they had the $1,500 angel gun. and I'm thinking to myself, like you need to start with something more simple, like learn how to shoot, you know learn these different moves ah before you get complex. um But part of that is the challenge of us as humans and marketers. We want to get more sophisticated, kind of, sometimes. Yeah, we want the nice shiny toy. And listen, some brands are killing it, so they can afford it. And they can get the nice shiny toys. I was talking this week with a guy I know that pays for three individual vendors for MMM, MTA, and GeoHolaTest. And I was like, wow, I'm amazed that you got the budget for three separate vendors for that. Their business is killing it, though. So I guess good to be them. But honestly,
00:22:33
Speaker
but ah being a bit serious here, like you want, to the Shiny toys are nice, but you need to make sure that you're going to use them. If you're not going to use them, if you can't work it into your businesses, um operating cadence, if you can't get it so that you're reporting out against those Shiny toys, they are exactly that, just toys. They're not really providing value. I want to talk a little bit ah ah about the the kind of well lean teams and what you what we can ah do based on company size and things like that. Before I do that, I'm working on kind of establishing a new term in the marketing industry that is based on um product market fit.
Aligning Marketing Spend with Demand
00:23:11
Speaker
So product market fit, ah basically aligning product to consumer demand and you've reached it once you are
00:23:21
Speaker
you have a product that is capturing demand, that ah that is satisfying demand essentially. and What I extrapolate this out in the paid marketing sense is paid marketing fit, which is aligning marketing spend with marketing demand. The idea that the first dollar that you spend should be on your highest demand customers. And then as you scale, you actually scale into lower demand audiences. you spend dollars first where there's a lot of demand, there's higher conversion rate, you're you get better performance, you get revenue, you have lower CAC over time, you can expand your CAC threshold, you can expand into these you know higher CAC areas like TV and and things like that. Based on the you know your client mixing your experience with optimizing marketing spend and seeing a lot of folks doing it, what do you think of this this concept? um And yeah, what what are what is your feedback on that, I guess?
00:24:18
Speaker
I think you're, I think you're nailing actually a really interesting point. um I would even, I think even maybe before that, something that adds to this whole thought is um business marketing product fit. maybe And what I mean by that is making sure that the KPIs that you're even providing to the marketing team. are actually KPIs that can make your business succeed. you know If you're giving marketing targets that will drive your business into bankruptcy, you could have marketing product fit, but your business will go under. so and I've seen a lot of that in the past couple of years ah where money was was cheap. To your point specifically, i think I think you're really right. i mean Fundamentally, I saw a really interesting graph. I'll try to find it the other day about how you can approach
00:24:58
Speaker
where to start spending your dollars first. And the thesis there was you want to spend it with the highest demand highest demand, lowest cost, lowest discount customers first, which is for the most part, your existing customer base. you know If you can send an email to your existing customers and they all buy the product with no discount, you've That's the best. It's sort of like, what's next after that? um Is it um you know that same cohort with a slight discount? Or is it going into market with ah the ah proper CAC you're looking for? But basically, this whole theory of there's probably a sequence of you know the bottom of the whole thing would be um sending an email to your existing customers with ah with a giant discount because you're just trying to get rid of the product. So it's sort of like, what is the slope, downward slope, of where you should be focusing your time on? um Yeah, it's ah it's a super interesting concept.
00:25:44
Speaker
Yeah. And um obviously there are diminishing returns within a given demand stage yeah based on the company size, based on a lot of things. And what I see ah happening over time is that as you scale into different demand stages, for example, you do TV, you grow the relative size of the lower demand stages. So by doing more mid funnel stuff, you actually grow high demand audiences. When you get into TV, you're doing more TV stuff. the size of the mid-intent audience grows as there's more people aware of the product that then get into that funnel and are captured by that different marketing activity. Yeah, I think it's um particularly looking back the last couple of years, um I think that a lot of companies messed up by potentially, well,
00:26:33
Speaker
Today, it's easy to say you messed up by under-investing in some of that more mid to upper funnel um ah marketing in you know mid-22 to end of 23 when markets start to tighten up. I get why. You know you want to double down on your strongest customers, the cheapest acquisition costs. I get it. I think a lot of companies are feeling the pain for that today. And so kind of what you're saying, you know that ah linear TV starts to build up that middle funnel, maybe even a bit more upper funnel. If you haven't been doing that for last year, you're you know you're fishing in the same well that you've been fishing in for the past year. um and you're not getting anything new into it. so it's it's a really hard it's really hard for I sympathize here for both companies and marketers to balance that, to strike the right balance between spending your dollars today that are going to be the most efficient, but doing that at the cost of your future self. um Going back to the whole diet thing, you know it's you know you eat that cupcake late at night. If you feel good now, you eat the pain of it a day or a week later.
00:27:28
Speaker
um So yeah, I think companies almost need to think about that marketing product fit, almost maybe even in terms of like different objectives, in terms of there's an objective to keep replenishing top of funnel, an objective to drive mid funnel to bottom funnel, an objective to drive bottom of funnel to convert, because they all will have different KPIs and different channels of work or don't work for them. And frankly, different measurement methodologies to speak to of the the efficacy of them as well. Yes, I very much see this as a measurement problem in yours in your example of eating the cupcake. Actually, what happens is if you ate the cupcake today and you ate yourself tomorrow morning, you might not see as much of a difference if you weigh yourself a day later.
00:28:08
Speaker
So sometimes you have to know what that lag effect is in market you know of ah specifically of of stuff like like brand marketing. um And just to contextualize this even further, you know if we look at our industry today, in the past couple of years, there's been a lot of layoffs happening. um you know I think more start more startups are focused on profitability. ah There's less companies raising big rounds. um I personally see like even little things like, my LinkedIn recruiter inbox is not as crazy as it was a couple of years ago. So I wonder, like with with this environment in in mind, what are your recommendations for teams that are operating much more lean? What should they not be doing? What should they be doing from your perspective?
AI and Efficient Team Management
00:28:57
Speaker
um They should be focusing as much as possible, if not 100% on the actual number one thing that matters versus focusing 50% on the top two or the top three things that matter. um I think I've seen this time and time again. ah The thing that causes a company to die is more so um not being ruthlessly focused on the things that are most important. So you go super, super fast there. um that So that'd be my number one. Number two, then, it's, and by the way, it's not just asking for more resources, because what everyone always does is say, let me get more resources to move faster. I would say no, it's ah cut the least important projects um and cut, get rid of those. And it sucks. But I'm telling you, like the worst thing, again, is not focusing all your energy on the most valuable thing. The second part there is figure out then, okay,
00:29:54
Speaker
I'm at least focused. ah How do I do that better? This is obviously like very ah ah sexy in the moment, but I think AI is an amazing tool these days. um i I preach this all day long in my company. If you're not using AI every day, you're likely failing. Every single day, everybody should find some way to use some AI tool to make their their job better. um And it's a new habit to build. like I find myself sometimes just doing something repetitive 10 times in a row. I'm like, why am I doing it this way? um So I think that's probably the two key things that piece of advice I would give. Expect no more resources to come. focus all of your energy on what is actually the number one most important thing, cutting off things that are even number two, um and find tools that make you just more effective. Automate your life away if you can. Yeah, I think AI can be one of those that can be a little bit of a shiny object too. If you spend too much time on it, you might be taken away from other things. ah Double clicking on that though, um if ah you were to tell someone, okay, spend 50 to 100 hours getting really good at AI,
00:30:51
Speaker
What would that curriculum look like from from your perspective? For marketers specifically, do you have any opinion on that? I'm not the expert on this one. I okay i think there are actually, I can probably put a new touch to the folks, but I think we're a lot better at this. Please. um For me personally, I just think getting to the notion of using, this is my personal one, and um ah using the ah some sort of a chat GPT, chat GPT, or Gemini, or Claude, you name what it is. um Think of it like an intern you have. An intern you have 24-7 that will never get upset, um that is really pretty smart and just needs feedback. And as long as you act that way and think to go to your intern all the time, it's like the best employee you could ever have. um But it's just building that muscle to think every single time when you like have a question, you want to bounce idea of somebody, go there first. um All that said, like I saw a tool this weekend that, I forget the name of it, um makes unbelievable
00:31:44
Speaker
audio recordings based off somebody's voice, you give it a little bit of voice, you can basically make a whole podcast off of you, you give it a, like it's unbelievable what you could do these days. I'll get the name for you too, but um yeah, im I'm not the, I actually would love to consume ah the right agenda in terms of a rather syllabus of how to become better at AI. Do you think that AI plays any role in measurement and kind of product expansion at Rockerbox? I think that measurement is actually a series of problems. um I think when you're actually thinking about marketing measurement, it's because you ah
00:32:21
Speaker
You have some goal and you arrive at some question and then you try to get the data to answer the question. Then you try to actually answer the question and then you try to figure out what do I do based off that answer. And unless you do all these things, you're doing nothing fundamentally. I think there are certain components or stages of that. ah kind of sequence of events that AI can be really, really beneficial in, particularly in some of the so what part of ah of of that of that ah kind of five step process, where we're already starting to see a little bit of rockerbox with some of our
00:32:53
Speaker
behind the scenes work that we're doing. So I think there's that side of it. um I think AI is also going to be really interesting for things. This is maybe less ah a bit less on the measurement side, but in the audience creation side, coming out of measurement. So help me make audiences that are most like people who come to me via linear TV. I think that's an interesting example because it requires, as you to know, who came via linear TV, which requires some measurement. But then you want to start creating audiences that look like that, which can be really done effectively with AI. so um Yeah. I mean, we'll see. We'll see. Yeah. Cool. Um, going on that, I know we have about 10 minutes left. Uh, I have some questions for, for productions of, of the future.
Seamless Data Integration and Privacy
00:33:30
Speaker
Uh, so, uh, we, we look at, um, we look at the years 2034, right? So we're, we're 10 years away. Um, in 2034, what does rocker boxes product mix look like? I mean, I think it comes back to like what we're just trying to accomplish as an organization. We want to build the world's best measurement platform. um To me, that means you have all the right data on top of it. You have all the right analysis on top of it. I think the next step that we could do a better job at is um ease of insight and ease of actionability on top of that. I think those are two areas that we're not um we're doing as good a job as I'd like us to do. But if you think about it, imagine walking you know waking up ah on a given day,
00:34:08
Speaker
Whatever question you want answered, you're able to easily get ah the data automatically assembled together. um You're easily able to get a handful of suggested actions taken off that. And those actions are automatically actually orchestrated across all the platforms that you're using. And the feedback loop continues over and over again. um I think it's a pretty cool place to be from a measurement perspective. um A lot of that actually stems into feeding results back into the buying platforms. So making sure that if you are holding ah Pinterest to account, I'm just picking Pinterest, um for third party measurement, then Pinterest needs to know that. So they can actually do their optimization systems against that as well. Other areas I think could be interesting too, coming from looking a decade out, um I think there are going to be more interesting ways to do the intersection of user level data that's even increasingly more privacy compliant, um such that examples are,
00:35:01
Speaker
ah You can have it so that two parties can share data in a way that neither party actually knows what data was shared by either party. Analysis can be provided on top of it in a way that is no use of the data. And nobody actually, there's a full end-to-end audit trail of who had access to the data, where it started, where it ended, how it got deleted. There's some really interesting research being done there. So we'll see where that goes. I think they're interested. There are a lot of questions in the business models behind those who actually wants to pay for that and who will be paying for that. But I think there'll be, and this goes a bit full circle to where we started this conversation. I think the trend will be for more sharing of data. I think it just has to be done. Well, more sharing of first-party data in privacy compliant, increasingly sophisticated ways that enable all parties to accrue financial benefit while preserving privacy of the individual. And I think that
00:35:48
Speaker
we will get better and better at that as and ah as an as an industry. What resources would you recommend to pay attention to these these trends and research and things like that? I mean, you can check out the Rockerbox blog, but I think there I would focus a little bit on some of the just the measurement companies out there. look at what they're producing. A lot of them produce some pretty interesting white papers. There are actually a lot of universities. I can follow up with some names of ah research that I did, but a lot of universities, professors are coming up with a really interesting research here. One example was like, I read a paper a couple of weeks back about how a small number of incrementality tests, geo holdout tests can actually be leveraged across many, many advertisers.
00:36:27
Speaker
So like as an advertiser, you think, I said earlier, like hey, is your Q1 incrementality test actually going to be valid in Q2 or Q3, Q4? What this paper posits is that um not only will it potentially be valid, but you potentially can get benefit out of having the incrementality test of a different brand in Q1 or Q2 or Q3 and leveraging it for a larger group of people. These are, I think, some areas where machine learning and AI is actually going to play for some pretty interesting role in. But um yeah, there are some interesting universities. Google also. Google puts out, I mean, they the companies who can invest in research are doing some some interesting things.
00:36:59
Speaker
That's awesome. Yeah. If you could, uh, find that report, I like it in the, in the show notes. Yeah. What is the channel that marketers are using today? That in 2034 will be obsolete, obsolete linear TV. And final question in terms of like predictions for 2034.
The Future of Search Marketing with AI
00:37:14
Speaker
Actually, you know, I'll, I could go back and give a different one too. That's potentially more here. This is a, I'm most likely gonna be wrong about this. Um, search as we know it today, just seeing how chat bots are starting to work and become part of workflows. It'll be very interesting to see what the impact of. of it is on search as we know it right now. So we'll we'll we'll see. I can see a world where the the place that you go to find information right away is not in a search engine, but something else.
00:37:37
Speaker
Yeah, I agree with you. um My girlfriend always makes fun of me because she'll ask me about something, and I'm like, oh, like, ask chat GPT. And she's like, no, I'm going to ask Google. And I'm like, no, no, no, like, ask chat GPT. It'll get you a better answer. But again, Google's launching their stuff. So that's a question of, like, how intermingled do all these become? Is is just Google with AI Google in the future and that equals search? I guess ah last last prediction here then, because TV is a big one in terms of spend. And these are these are great answers, by the way. I love this. top so Today, probably the top three channels would be Google, Facebook, maybe TV. What do you think the top three channels would be in 2034 in terms of media spend? I think some sort of search but search type channel, i.e. a channel that you're using to gather answers to questions. I don't know if that's search or if that's AI, um whatever the social network of the moment is. um ah Probably might have had to guess, although I've been skeptical that
00:38:28
Speaker
In the past and I've probably been wrong every a single time I was worried about their um aging Demographics and whatever is gonna be the video content of the day be it would be at some sort of short-form video content ah Be it a reels be to YouTube but be at AR VR I don't know but something that you're watching you're watching things for extended periods of extended periods of time And who knows how much of a role VR will play by that point? um Remains to be seen um Cool ah Final couple questions here for just to to close. I would love to know, um I'm sure that you you know you read a lot of research and things like that. For for marketers specifically that are thinking about growth in their career and also you know business, do you have any recommendations for books that you'll often share?
00:39:15
Speaker
um I think there's some like classic books that are good, like even Ogilvy's on advertising, I think is a good book to read. um That's kind of more old school and creative, but I think there's there's some good lessons there. I'd say more tactically today, I think reforge has some pretty good courses. i'm Personally, like ah the way I learn is I'm just a doer. like If I want to figure something out, like I just start doing it and I suck at it for a long time. So I'm often not the best source for these sorts of questions. Other folks I know read a bunch and then they learn that way. I'm a bit of the opposite. So yeah, like I like to just try and fill throughout. And I know that you haven't, you know, you're not doing as much in the weeds media buying, but um given that this is the efficient spend podcast, maybe within your client mix, ah what's the most efficient spend that you've ever seen and what's the most inefficient spend?
00:39:57
Speaker
ah Probably both Google. I hate to say it. I remember the first time I actually saw a client kill on Google. And it wasn't just like that they killed it, but their ability to scale it was unreal. I was just like, there's just, at the time, it was actually around um horse betting company. and It was around, what's that big derby called again? Kentucky Derby. Yeah, Kentucky Derby time. And their ability to scale spend in the lead up to that was just unbelievable. And they They've business it amazingly. I think the counter is probably companies who just wasted millions and millions of dollars on branded SEM when they probably didn't have to do any of that. Cool. ah Ron, thank you so much for for being on the show. Thanks, Paul. Really enjoyed it.