00:00:00 Phil: What's up guys. Welcome to the humans of MarTech podcast. His name is John Taylor. My name is Phil Gomez. Our mission is to future proof the humans behind the tech so you can have a successful and happy career in marketing. What's up everyone? Today we're taking a deep dive into customer data and the stack that enables marketers to activate it. We'll be introducing you to packaged customer data platforms in the more flexible options of composable customer data stacks, and getting different perspectives on which option might be best. I've used both at different companies and have had the pleasure of partnering with really smart data engineers that have helped me level up on the data side, and I've had the chance of using up and coming data tools. And I'm really excited to to dive in today and get my little spin on the popular debate. Okay, let the battle begin. Here's today's main takeaway. The debate between packaged and composable CDPs boils down to a trade off between the out of the box functionality and tailored flexibility. With industry opinions divided on what offers greater long term value, key factors to consider. Obviously, our company needs and data team size, but if you do decide to explore the composable route, consider tools that focus on seamless integration and adaptability rather than those who claim to replace existing CDPs. Okay, first things first, let's get some definitions out of the way. Let's start with the more common packaged CDPs. A CDP. A customer data platform is software that consolidates customer data from various sources and makes it accessible to other tools that you use in your company. The end goal is being able to personalize customer interactions at scale. I've been a big fan of RP Choudhary of Data Beats, and he articulates the components of a CDP better than anywhere I've seen in his posts. Composable CDP versus packaged CDP. So let's unpack each of those eight elements and let's get takes from different industry pros if they agree with our beats eight components. So number one, CDI customer data infrastructure. Fancy way of saying this is where you collect first party data directly from your customers, usually on your site or on your apps with a tracking code. Number two is ETL data ingestion. It stands for extract, transform load. This is simply about pulling data from different tools that you currently use, maybe on other teams, and integrating that into a central storage component. Um, obviously the data warehouse isn't always the central point of, uh, storage. Uh, like I said, CDPs will just like have their own database. And this is where the collected data resides, essentially a central repository, right? Number four is ID resolution. This is pretty much how you connect the dots between various interactions a customer has with your brand. And maybe they're using different platforms to access your, your website or, you know, different mobile apps. Maybe they're using different devices like their iPad, their laptop, their desktop, their phones. Audience segmentation is the fifth component. Um, marketers are more familiar with this one here. Like your marketing automation tools have an audience segmentation component to that already. This usually comes with a drag and drop UI for easily sorting your audience into different buckets based on events and behavior, demographic traits, or any other factors you have. Um, that's flowing into that audience segmentation tool. And then number six is reverse ETL. This is where stuff gets a bit more, uh, contentious, especially in the market with data and MarTech tools. But reverse ETL is exactly the opposite of the ETL portion that we talked you through. So instead of like getting data from existing tools and putting it into your central storage spot, this is the reverse. It's taking, uh, the data from your central storage pod, whether it's your warehouse or, uh, any other, uh, centralized database and pushing it out to other tools that you use. And then seven and eight are data quality and governance slash privacy and compliance. You can argue about like whether these are separate or all in one. In summary, eight components that make up an all in one CDP. So like most companies who say that they're a CDP, they should in some shape or form meet most of the criterias under this list. So collect first party data and important data from other tools into a central database, run ID resolution to figure out like different users coming in from different devices, focus on quality and compliance, and then finally having a segmentation engine and allowing you to push that data to other tools. That's how we boil down the simple definition of an all in one customer data platform. But I've asked recent guests if they agree with all eight of these components. In my mind, you know what you what you described as a package. CDP comes from a long line of software, right? That's Borisx Jabez, the co-founder and CEO at Sensus, a reverse ETL tool that allows marketers to activate customer data from their data warehouse. Actually, if you go back in time, there's also Dmp's and SEPs. Right. There's a lot of three letter acronyms here. Uh, and what I would say is they were, they are products, uh, specifically designed for marketing and generally designed historically, in my experience for consumer marketing, like B2C companies. And they do, I would say three things, right? They help you collect events from your website, from your applications. They are a source of truth for that data. Uh, for the marketing team specifically. And they allow you to segment and personalize target, uh, you know, based on this data into other, marketing tools, whether those are advertising platforms. If you're coming from the DMP side of the CDP world. Uh, and again, just, just using all the jargon, um, or it might be into a, you know, email or direct mail tool, etc.. Right. So I think that's probably the most complete definition, I guess you could say of a package CDP. Yeah, I wouldn't necessarily argue, uh, there are definitely some details to each of them. That's Tamara Grosberg, VP of customer Strategy at action IQ and Enterprise Customer Data Platform. Starting with data collection and ending with data activation is is definitely critical when it comes to audience segmentation, drag and drop UI. Absolutely critical in terms of providing this data democratization and self-service, I would say. From this we also offer, Um, I wouldn't call it a bi tool, but more of a insights dashboards that allow people not just build the segments, but understand the overlaps with other segments, understand. Maybe, you know, distributions on some key KPIs. So get more understanding of the behavior of this particular segment that would allow them to create more efficient marketing campaign. Uh, as well as we have two types of audiences, uh, one we call the rule driven and another one ML driven. Uh, we have a component, uh, machine learning component within our platform that allows our clients to build the audiences, um, based on predictive models. So, um, That's definitely a component that a lot of clients are interested in, especially depending on the type of the client. Maybe not huge enterprises with large data science teams, but more on the mid-market side. But sometimes even when it comes to large enterprises, a couple of those that we work with come to mind as we collect, uh, completely new data for them that they didn't have access to and they want to put something into market quickly. Um, they utilize our predictive tool to create these audiences. And of course, we work in collaboration with their data science teams to make sure that, you know, it passes the sniff test. I agree with, with those eight, um, uh, pieces of functionality that I think comprise a let's forget package, whatever. That's like an end to end solution, right? That's Michael Katz, CEO and co-founder at M particle, one of the leading packaged customer data platforms. Because we'll like, we'll kind of tear apart the, the nonsense argument in a bit, but like, yeah, those are, those are probably the, the eight main components of at the very least, like the first generation of customer data platforms. The value of these components, which need to be integrated is that the whole is greater than the sum of the parts, right? Otherwise, mix and match from lots of lots of different vendors. And so you start to skip certain steps. Not all CDPs, most actually, most CDPs don't have strong stories around data quality and data governance, right? So you're automatically operating off of like a, a weak or unstable foundation. Um, and the value starts to compound as you move from like the front part of, of, of the pipeline or system through the activation layer. So it's not just about how fast can you get data out into your, your application layer? It's about how, how fast can you do it relative to maintaining some level of, of quality. And, um, and consumer privacy protection. I don't particularly disagree or agree or. Yeah, I don't, I don't have an opinion on whether these are the components of the CDP. And that's Tejas Manohar, the co-founder and co-CEO at High Touch, another reverse ETL tool that's taken a bit more of a controversial stance. But what I think is most important is that, like, why do people persist in the first place? It's so that they can, you know, harness all the data they have about their customers and activate it across their marketing and to personalize the customer experience and, and drive great business outcomes. So that's the reason companies, um, pursue a CDP in the first place. Um, everything else is kind of ancillary to be completely honest, like everything else is a, is a means to an end or, you know, it's not worth thinking in components or features. Um, I usually think of it as three core things, which is that you need in a CD in a typical traditional package, CDP, whatever you want to call it. Uh, there's a way to collect data, there's a way to do some data transformation in between. So things like identity resolution and modeling. And then there's a way to activate the data, which usually, you know, has audience building or integrations, etc.. So a couple different takes there. Uh, Boris and Tejas actually agree that a package CDP focuses on data collection, serving as a primary data source for marketing and enabling segmentation slash activation of that data. Uh, tomorrow adds predictive modeling to the mix at action IQ and underscores the importance of data quality and privacy and definitely not skipping over those steps. Uh, Michael Katz points out that not all CDPs excel at every one of those eight components, and especially emphasizes the need for data quality and governance. Uh, just like Tamara did as well. So what's the hype around composable CDPs thought of as the new kid on the block? Composable CDPs promise a lot of different things compared to the packaged option. Composable CDPs take a modular approach to data management built from separate, easily interchangeable parts and easily as kind of an Asterix here. This design and how you kind of build this offers way more control over your data processes. And you can also customize this to fit any of your particular business objectives or requirements. Um, the composable route provides a contrast to a packaged solution based on balancing specialized benefits against workflow complexity. So here's an example of a composable tool setup. So instead of going with an M particle or an action IQ, you could have snowplow as your customer data infrastructure collecting first party data. You could have Airbyte as your ETL tool, getting data from your other email tools and your customer support tools, pushing that into your data warehouse. That could be BigQuery, could be snowflake. And then you have a reverse ETL tool that could be census, for example. And then you could use DBT or a mix of other stuff for data quality. So basically five tools plus like maybe one or two other specialized ones in a stack to compete with an all in one package solution. Um, but not everyone actually sees the composable route as an entirely new thing. Yeah. So, um, when I think about composable, um, CDP, uh, there's actually something that pre-exists, you know, a composable CDP that a lot of people already attuned and believe, as you know, here, here's an approach composable. And that's actually what commerce. That's David Chen. He's the managing director at Deloitte Digital and he leads their CDP practice. Like I said, my background, um, in the past, um, included a lot of different technologies, but commerce and web content management was, was one of them back in like twenty thirteen, twenty fifteen, it became very common to have what they called headless commerce. What's headless commerce? It was, uh, splitting up, um, web content management from just pure e-commerce tools where, um, the web content management tool would act as the front end and the commerce that had all the heavy logic and the, you know, the checkout pages and the product detail pages and sort of handling all the OMS and fulfillment options that was kind of separated out. And there were different like ways in which you could composedly build that. That was just like a very first start where there was sort of like three patterns you can follow. And over time, it got even more, more advanced. Um, some content management systems completely break apart from a template. Everything is these, uh, different components, even on a page in which, um, your operations teams would build, uh, even pages off of web pages that is. And so the composable architecture approach is something I've already seen in a commerce space. Now I'm looking at it and comparing it to the CDP space. Well, what's the difference? Number one, the composability approach of commerce went through the same sort of like, let's call it consolidation and the mashing and the banging that we're seeing today in the CDP space, which is like none of these things really want to work together. They're all a bunch of features and capabilities, but how do they actually plug in together in a unified way? And so, um, what's happened? Right? If you think about it's been ten years since then, more standards and partnerships have evolved where people know, okay, here is how all these composable things should fit together. That's what's missing right now in the CDP space. All that hardening of thoughts, standards, frameworks of how not how their tools work, but how they should work together. Um, that's really what's missing. But I see it happening, right? I there's no reason why it wouldn't follow that route. You just have to give it some more time. And has that changed? Has my beliefs changed at all? No it hasn't. We're still not there yet, in my opinion. Let's hear from Tejas Manohar again, the founder and CEO at High Touch. Two things happening. A lot of one, a lot of purchasing of CDPs. Everyone wanted the vision of a CDP in their company. They all wanted to be able to have their marketers freely self-serving, understanding their customers, exploring their data, building audiences, and then two, deploying those personalizations to the customer journeys across every channel and every tool in the marketing stack, whether it's a lifecycle or advertising or push notifications or in your app or whatever it is. But at the same time, a deep dissatisfaction with the state of the solutions today. Um, I think you don't have to, you know, trust my words on this. Gartner conveniently released a report for us, which actually, uh, said not for us. They released a report to the public that said sixty percent of customers who purchase a CDP are getting value from it. The marketing team isn't getting value from it. Um, so this idea that yes, you want a CDP, you really do. But there's a new way of thinking about this, that it's a lot more likely to be successful and allow your marketers to use data across the organization. And that's a way where, uh, marketing can access all the data that large companies are already investing in, in their data warehouse, uh, and start activating that across the customer journeys. And finally, let's get Boris's take the founder and CEO at census that we heard from already. Yeah. So first of all, I think there's a lot of reasons why people are looking at a different approach. Let's call it that. You're right. That census is built differently. And again, we came at it from first principles. I didn't come at it from when we started this whole company that, you know, you so eloquently put in my like intro bio there. Uh, we didn't come at it from a, hey, what are these tools? You know, what does the package keep doing? And we should do something different. It was absolutely not in our mind. What we wanted was to give marketers more, better, more trustworthy data and ideally to try to make the world a, what I would call a better place. We, we wanted to reduce the proliferation of these things. Right? Uh, and you're right, in a lot of companies, there has been a huge investment happening in parallel over the last few years for on what you would call data warehouses or other kinds of data platforms, uh, whether they're from Google or Snowflake or Amazon or Databricks, you know, it doesn't matter. There's so many options, but they are designed to store infinite amounts of data. They are infinitely flexible, I would say, at this point. Uh, and they are used to answer all sorts of questions. And so why would we replicate that to solve the problem, right? So, so yeah, census came at it, uh, from the start by saying like, well, we already have the information that I think we need. So let's start from there. Right? Uh, but I do think composability is, is, is a philosophy at some level, right? I always tell my own team it comes, I think of that word coming from the world of software. Uh, so like that's a bit of a technical aside, but really it means that you're building parts that are flexible and can work together seamlessly. It's not about breaking things down into pieces, it's about pieces that work together and in such a way that if you need to build something yourself or be custom in one part, it's that's native to the system, right? And, and so I think That I don't know about you. You've actually maybe built more CDP like implementations than I have. But I found personally that they tend to fall short somewhere in some dimension, somewhere. Uh, and if you're lucky and your business perfectly fits the, you know, the parameters, then I'm super jealous. Uh, but as soon as you want to be customized in some form, like you're going to need to do something outside of it and make sure that that interacts, you know, that it composes with it. And so I think that's why this is a trend is that people kind of realize they need more flexibility. I do think like, why, why is there more? Why is there a need for more flexibility? And I think there, there's, there's a couple of things that are happening beyond just the fact that you have a data team that marketers should probably be aware of if they're not deeply, deeply aware of already. One is the journey. The customer journey is getting way more complicated. Uh, it's not just you have a website Site. And you know, they you put a pixel on it and you track the. You know, shopping cart and you, uh. You're done. Right? Like Facebook even takes care of it for, for you for most of it, that was kind of a happy world that we lived in. If you're a marketer, like, let's say five years ago, it's just so easy. Just basically put a pixel and you're more or less done, right? Attribution is solved. Uh, engagement with the customer can be basically done because it's this very single path. Now, if you're in B2B, it's unbelievably complex, right? Like that's where we started before kind of getting deeper into the, into the consumer world, but sales cycles are multi touchpoint, right? Like app users are all sorts of variations. You have workspaces, you have different, you have admins, you have regular users, you have guests, you have so many ways to model the user and the user relationship. And it takes a very long time, right? It's not just from they visited our add to their free user to a paid user. It just goes on and on and on and consumer. Same thing, right? Like now you have Millions of users. You span multiple countries, you have tons of channels. You now you're going to see an ad in different places. You got to make sure they work together. Uh, attribution is just much more complicated. Uh, and you have to adapt to that. And it all gets harder, if you will, uh, because of privacy. And, uh, you know, this is one of those where it must be interesting to be in the field of marketing because you as a consumer probably appreciate some of these things. And then you as a practitioner are really frustrated by these things. But, you know, the easy world of putting an external pixel on your website and having kind of your marketing problem solved is gone. And, and so you need better activation on your first party data. You need better collection of first party data and usage of it, which is kind of where does that sit sits somewhere in your product database slash data warehouse, right? Uh, and you need a way to govern that because, you know, whether it's the EU or California, they they expect consumers expect you to be able to, to kind of show them what you know about them and be able to delete that data. So, so I think all these, these are a bunch of trends that lead to people, I think, thinking about solving the problem of customer data platforms, which is what we outlined at the beginning in a way that is a composition of tools, uh, centered, I think, around the data warehouse. So pretty interesting perspectives here. We heard from David and Boris. Tejas notes that the shift to composable was fueled by this customer dissatisfaction with traditional packaged CDPs. Uh, over the last few years, and, um, aiming for a system that integrates well with existing data warehouses. But Boris emphasizes that the drive for composable architecture comes from a need for reliable data and adaptability for this, like over and like increasingly complex customer journey and privacy needs that companies deal with. David adds that while composability isn't necessarily new and exists in fields like commerce, CDP's are still maturing in this aspect, and the consensus is that composability offers a flexible, tailored approach to data management. But the concept is still evolving in the CDP space. Okay, so we've covered the components of the definition of a package CDP, and why there might be a need for some companies to explore the more flexible route of a composable stack. Let's hear from various different industry pros about where they side when it comes to the package versus composable CDP battle. But what I do see customers is they're either from the segment space and realizing that the licensing cost is really starting to trip them up over time, and it doesn't really fit something that they want to keep long term. That's why bales Chief Customer Officer at Blueprint X, they're a global growth as a service consultancy who provide MarTech sales, tech and work management solutions to then like to name drop one of our great customers, Sota dot io here in next door in Belgium. They're a data integrity solution that sits right on top of like snowflake or your, or your, your database. And it just tells you how clean or how ugly your data is. And so the way that I kind of see this is maybe not a traditional CDP platform, any duplicative database, but just a warehouse that has a plethora of all these tools that sit on top of it, that just tells you what exactly you have inside of your warehouse. And then if it comes down to these outbound API calls, like you mentioned on doing email delivery or campaign execution, I definitely think again, for the warehouse space, sorry for the enterprise space. That is where the future is. Yeah, I think, I think it really depends on on each company and the needs of that company. I think, I think we're probably not, I would say twenty five tools. I would say we're like probably three, like big three. I would say that's pinnacle. He's the CEO of Optimove. Their platform that combines three big, traditionally separate tools into one. They're a CDP, a journey orchestration tool and an AI engine all in one platform. Look, the question is, can you unlock more value by having things cohesive in one place? Well, we what we claim and what we believe is that data and channels like messaging channels need to live in the same place. And so when you say it's impossible, I was like CDP. Ultimately, the data is there to drive some kind of business value. The fact that it's sitting somewhere and you can push it to another place, like ultimately the question is, what are you going to do with it? So are you going to is it for analytics? Are you going to drive better decisions because of it? Or are you going to, you know, delight your customers and use this organization that I was talking about before only because the two exist together. When you break them down and they start to get siloed. So yes, maybe handling data separately is a silo will be done very well by one tool. But then, you know, some of that data will be forwarded to another tool that does messaging. But that tool in itself is still inherently rule based. So what it does is you feed that tool with an entry point of like a smarter segment, but that tool will still decide. So decide and make decisioning rule based in a rule based nature. So what do we feel like is you need to feed AI and decisioning into the bloodstream of channels. So the bloodstream in the place where you send the message and you engage with the customer that that bloodstream need to be, you know, hyped with, with data, with, with decisioning. And that comes when it's together. It doesn't come to API like it doesn't come through because the APIs are good about sharing data. And let's think an audience from here to there, right? Let's say, and then that audience is going to start a journey in some kind of a system, right? The audience is going to go into an email system. It's going to start a journey. They're fine. It doesn't change the nature of that email system. That email system is still running with kind of like on on rule based food. Yes. I think there are a lot of debates around it, right? Like composable and this thing, right. And I agree and disagree with a lot of those points, right? Because eventually it is all everybody's trying to kind of market their own products, right? That's a run through his CEO and co-founder at Castle dot io. Their warehouse native customer engagement platform that sits directly on top of cloud data warehouses. But in my interview with Arun, he explained that before he went down this path of customer engagement, he actually debated building reverse ETL. So he's well versed in this battle. Um, for me, the different I mean, the way I see composable and packaged CDP is, is how I see maybe an open source and a closed source system, right? I mean, these are different things, which I probably have a lot of, not a lot of people have kind of talked about. Um, so a composed like or a packaged CDP right on top of the data warehouse. It gives me the flexibility to innovate on top of the data warehouse, right? You can actually, if you feel that it is not like a complete right. Then I can add, I can probably add like more tables. I can more add more transformations, right? I can plug external tools to the system, right? To give you an example, right? Um, there are tools like maybe there are tools like, uh, a thing, right? What they do is actually they do identity resolution on top of the data warehouse, right? And they do kind of fuzzy resolution, not like deterministic. So they'll say that, okay, this row and this row might be same. And because of that, I'll just kind of join them together. Right? And these are innovations which are actually happening on top of the data warehouse now. I don't want to be in a closed system right where I'm not like, you know, I cannot actually have these innovations on top of my data warehouse. So that is actually, I mean, obviously, there are a lot of other arguments around it, which I think a lot of people have said, but this is just one analogy, which I don't think a lot of people have actually seen this thing. And this is how I see composable versus packaged CDP. Yeah. So, um, I would agree with the statement that CDPs have to evolve. Um, but I like what business doesn't have to evolve over, right? Um, you strip out all of the kind of endless debate around package versus composable because it is all just like its its product marketing. That's that's all it is. And finally, that's Michael Katz again. The CEO of M particle, a packaged CDP. But what I think is undeniable is like the rise of the cloud data warehouse as a really important system inside most organizations, for them to be able to have a, um, hopefully have a single source of truth, right? Um, now, now the problem is from a, just as like a matter of practicality, data quality varies widely, um, across, um, even within an organization, but definitely across organizations. So it's not as simple as like, hey, stand up a data warehouse and all your problems are solved. There's a lot of work that has to go into getting it right. That starts with having a data strategy, right? So I would, I continue to vehemently argue that reverse ETL. When when when tapping into a weak foundation, it's just like it's garbage in, garbage out. And it's a faster path. Like faster garbage is still garbage. And that's like, that's all they're peddling. And so my, my point around like the sleight of hand, um, tricks is like they're, they're trying to create distractions from what like the real issues are, and the distractions, as I call out, are like the importance of, um, zero data copy, which is actually not possible if you want to create a best, best in breed stack. Um, then there's, uh, this false narrative around the fact that CDP's are more vulnerable to security threats, which also, um, completely, completely unfounded. Um, and then, you know, the, the last argument, um, I think keep me honest here, but I think it's centered around was it, was it privacy or. No, sorry, it was faster time to value. Yeah. It was deployment times. And, and the point I would make is, and the point I did make is that time to initial value is not the same as time to sustained value. Something that's easy to get started on is usually harder to, to maintain. Like nothing in this life. Like forget CDP's for a second. Nothing in this life is is free, right? So you either pay for it on the front end or you usually pay for it on the back end. And the back end payments usually contain compounding interest, right? It's like buying a house. So like we got it like we got to separate all the, all the noise. I will commend the, um, these kids running reverse ETL companies like they have super aggressive product marketing and they've been noisy enough to. To get a bunch of attention, but like, I'll say it here, it's it's designed to trick. The market. It's not designed to create value for customers. And. And that's the problem I've had with it for the past couple of years. I think the argument there is stemming from the desire to package reverse ETL tools in a way that sells to marketers, but I think that causes a lot of confusions of what is a reverse ETL really doing. That's Prateek Desai. He's the CEO and founder of One to one, their personalization agency that works with enterprise clients. And they've actually recently released a product called Ragana, a composable search and sort personalization engine that's built on top of your e-comm platform. And he's got a lot of hot thoughts about the CDP battle. And so, you know, in terms of a CDP, one of the big aspects or features of the CDP would be the ability for identity resolution. We know that reverse etls don't do that. That doesn't negate the value of reverse ETL. It just means it's not a like for like or apples to apples comparison, right? And so now if I start to sell reverse ETL as a replacement for CDP, I need to make sure I'm selling into an organization where the marketer is positioned to work with their data warehouse owners to get that data structured in a way that works for them. Right? We talked a little bit earlier about what happens when your data isn't structured in a way that works for your personalization or marketing program. And so kind of roll that up into what does that mean? Right? I think let's not forget that different databases of your users. That was a situation that was created because marketing did not have a seat at the data table. And so CDPs were really trying to create a solution for marketers that did not have that seat. Now, I think we can treat this entire innovation as an impetus for change and get the marketer marketing organization a seat at that table. But we know how that kind of unfolds, especially at the enterprise level where buying, you know, a reverse ETL tool isn't the impetus to change an organizational issue, right? We talked about operational excellence. Buying a new tool generally doesn't solve operational issues. So as we start to kind of pare that back, I think to answer your question, going back to reverse, you know, the big the big conversation around reverse ETL versus CDPs, I do think they're both solving problems. I do think they're solving different problems. And the shift from what the problem is that CDPs solve into the reverse ETL world. I think we need to understand what that means from an operational lens and a data excellence lens. Solve that root problem that CDPs evolved because of if we can solve that root problem, then I think reverse ETL provide a tremendous amount of value. Or if we're starting to sell into generally where we see SMEs, where marketing already has a seat at that data table, reverse Etls make a huge sense, right? They make they make a huge amount of value within that type of organization. So I guess the question is, is there a need to keep up based on the idea that do we believe all potential buyers, all potential clients of these technologies will have the same problems that can be solved by composable architectures? So it's really tough, right? Because I do believe we're moving more and more into this idea of a modern data stack where we do have a single database, we have a single view of the customer. But I do struggle also to see how quickly that can happen with this, with these enterprise level customers where, you know, the advent of CDPs and personalization programs was not the impetus to bring marketing to the data table. And I don't believe that reverse etls will be the thing that, you know, tips the scale in bringing them to the data table. We need to solve that problem first before reverse ETL cells and solves to every single potential client. So, you know, some summing all that up, I think all of our clients have different issues, and all of those issues need to be looked at subjectively. And then we need to bring out the right tools. So sometimes that right tool is a CDP, right? Sometimes that right tool is a reverse ETL. And then if we can get deeper and we can start to actually solve their operational and data excellence issues, then I think we can move into that modern data stack. And I do think that's the direction. So some really good back and forth in this battle here, waging from a couple of different perspectives. We've got Wyatt and Prateek on the agency side, Penny and M.K. who run packaged solutions. And we've got Arun who's a team composability if you speak. But let's recap them a little bit. So Wyatt notes that customers are grappling with the choice between packaged and composable, but he does see the future in the data warehouses that serve as the foundation for multiple tools. Penny, however, argues that the choice between composable and packaged CDPs depends on a company's specific needs, but that for most scenarios, data and messaging channels should exist in the same ecosystem, not merely connected via APIs. On the other hand, Arun likens the debate to the open source versus closed source discussion, emphasizing that composable CDPs allow for more flexibility and innovation. M.K. criticizes the sleight of hand tactics used by some reverse ETL companies and dispels the myths, such as CDPs being more vulnerable to security threats, and points out that quick deployment doesn't always equal sustained value. And finally, Prateek thinks there is confusion being created in the market by marketing of reverse ETL as a replacement for CDPs, which can lead to a ton of misunderstandings, especially since they serve a different function and are not exactly comparable. So let's finish off by diving into this portion of the battle here a little bit. And what I think is a bit more contentious in the market. So let's talk about the confusion in the market. Can reverse ETL actually replace a packaged CDP? All of the confusion around this stems from one reverse ETL vendor in particular, high touch. They've written plenty of controversial articles over the years claiming that the CDP is dead, and that friends don't let friends buy a CDP and that they can actually replace a CDP. I sat down with Tejas, the founder and CEO of High Touch, to get to the bottom of why they think they can replace the package. CDP here's what I would say. You know, we power Marketingdata activation for some of the largest companies in the world, and they call us their CDP, right? They use us as their CDP internally because instead of using high touch, you know, they are not, they don't have to buy CDP now or they churned off a customer data platform, which is the case for some of our companies like blizzard, Activision, one of the largest gaming companies in the world. Previously on a customer data platform. Same with Warner, same with Red ventures, etc. won't name names here, but they were all on customer data platforms before. But our product doesn't look exactly like CDP. You're right. Um, and some of the things, I mean, we're only really missing one thing, which is that data collection part. Um, you know, we, we always find a way to offer that to our customer, whether it's through partners like snowplow and the future, we're already working on ways to make that easier and easier for our clients so that it's not a blocker and so that we can, focus, they can focus on what's most important, which is being successful at activating their data. And that's really our differentiator at high touch, making sure our marketing teams are successfully able to activate their data. And, uh, and personalize. Um, the other thing I will mention is that, um, in our first blog post, CDP's are dead. Uh, we, we said something at the beginning, which was that, you know, every CDP is basically going to, in a pivot, uh, to our approach at Hitesh over the next few years, uh, or go extinct. And, um, probably not exactly what I wrote in the blog post, but something to that effect. And I do think you can start to see that happening in the space. Um, so what I'll say is that, you know, we're, we're kind of leading the way for what the future of CDP looks like, which is in architecture where companies own their own data, have infinite flexibility, um, are able to activate across all different channels in the company, um, and are able to have marketers in the product. Um, and, uh, we feel confident that that's what the future of this space looks like and that you know, what this is building today will just be called, you know, will be what all the CDPs are sort of pivoting to in the future. So some large customers do refer to high touch as their internal CDP, but is that enough to be able to claim that the CDP is dead and that they're able to fully replace it? I asked Michael Katz for his take on his argument. Yeah, and I think that that represents a very narrow, um, subsegment of the market, right? Like if, if we keep the conversation rooted in business value and what's required in order to optimize business value, like the, the fragmented do it yourself approach doesn't necessarily get you there. Like it's, it's cool for like the, the hobbyists, but I mean, most, most businesses definitely within the enterprise. Like they can't afford to just be hobbyists. They need to deliver something. And I think the day of reckoning is, is coming. You look at a lot of the, the sloppy habits that were developed, certainly during the pandemic. But I think we're a byproduct of of zerp, right. Where you just had data engineers running amok and not getting proper business requirements and just doing things that they felt like were right on behalf of the business, given a very narrow and limited perspective of the world and little to no understanding of the true nuance and complexity of running successful digital marketing campaigns. And we've definitely seen that go away. Um, we've started to see reintegration of the end users into the, into the buying decision. So marketers, I would say over the past, I don't know, nine, twelve months have started to take the power back from from the data engineers. Um, we're, we're a tool that is built for marketers, right? Like we make it easy once data is, is in the system to be able to, to, to contextualize it and then to activate it in a low code, no code manner, right? So you don't have to create custom SQL scripts and do a bunch of transformations. You don't have to know your exact schema, right? Like everything is, everything is available through like a point and click UI. And that's just not the case with, you know, certain CDPs for for sure. But, but definitely a lot of the, a lot of the tools that have been built for like the, the modern data stack or the data warehouse ecosystem. So like usability is, is still very much a factor. And like, again, if we're focused on value creation, there's lots of different ways to get there, but the fastest path is usually the fastest path between any two points is usually a straight line. So there's value in having an integrated platform where there's seamless workflows and you don't have to work between if, if, if what we're saying is like there's eight different components to an end to end CDP. Yeah, sure. You could use eight different systems, but like what happens if and when something goes wrong? How do you troubleshoot across eight different systems? You have to get eight different account managers involved. Like the whole thing becomes a bit of a mess. So there's a ton of value in, in, in the integration of components that can be configured, um, in a way where you don't have to buy all of them and then, and then customize to, to meet whatever business requirements you have. Not all reverse ETL platforms have taken a loud, controversial approach to marketing, though. I asked Boris Best, the CEO and co-founder of census, if his platform aims to replace the CDP. Yeah. Uh, not only do we not replace it, I think there's tons of customers that have been using census for years in combination with CDP. So, so I think the, the, the that's just not the reality either, right? Whether it's our claim or not. Um, so I think first composability to me is a, like I said, it's a, it's a philosophy, it's a way of thinking. And it means that you build tools that integrate seamlessly with others and allow the user, right. Whether that's you as a marketer or, uh, your colleague at, at genius who wants to, you know, from the data team who wants to, to customize things further, whether it's the other parts of the ETL and warehousing stack, right? For them to be able to work with this without making it obscenely complicated, right? And that's so. So not only do we not kill a traditional CDP or. Claim that they're dead. Uh, I think we are trying to give more users more trustworthy data in more places, right? That's that census. And if, you know, we're trying to not add another silo to your business, right? So we're actually saying you have data infrastructure. Let's leverage it. Let's, you know, let's not make another copy. The census is not like, hey, here we are the new thing, right? Um, and so the yeah, I think it's a little bit weird to, to, you know, I think it's just common in, in marketing, in brand marketing to, to talk about things in terms of like, we're killing this or we're, you know, destroying that. And, you know, just look at YouTube any given day, it's like everything's being destroyed for some reason. Um, but I do think composability is, is net better for the world, right? And so the same way software like you ask any savvy engineer, any any grizzled engineer. They will talk about using very. Like the way they design their software is that it doesn't break when other software's combined with it. Like that's what makes it really good. Mhm. And programming languages even get discussed in this way of like, how do they solve composability? Uh, or for composability? And so I think just a warehouse native way is a really good way to approach that. I think everyone benefits. Like I said, I think, I think you are, you're a better standard bearer for it than I am. Uh, and if you have a tool that does identity resolution for you, whether it's called a CDP or it's called anything, an ID tool, right, I don't care. Then let's make it seamless for you to be able to use that in tandem with what we have, right? And I think you have, you'll find that a lot of companies, there's a data team, there's a data scientist who's actually doing something that resembles identity resolution, by the way. And that just works seamlessly with our platform. And if it happens in an external tool used by marketing, great. What I would say is, let's just make sure it benefits everyone, not just the marketing team, right? You know, rewind to the first thing I told you. It's like it's in the name of the company, right? Census is a very bold name for a company and very tough from an SEO perspective. But yeah, it's really tough for me to ask you your perspective, but, but it's really the goal. It's I want every company should not have twelve census, right? You should have one for yourself. And so if identity resolution is happening in your CDP, let's not lock that to just the marketing team, right? The sales team, the finance team, the privacy and compliance team, all of them need that too. So that's kind of where I take it as like, this is where composability wins over non composability, but it has nothing to do with like census and versatile kill CDPs. David Chen, the managing director at Deloitte Digital, who leads their CDP practice, is really close to the composable versus packaged debate. Despite wanting to move on from the debate a few times, I asked him about the perspectives of census and high touch and which side he kind of sides with a little bit more. So, so Boris and I actually, I moderated a panel, um, uh, sponsored by census on this whole composable versus CDP a couple of couple of weeks ago or months ago rather. And I got a chance to connect with Boris and learn more about sort of his background and that what you said basically aligns with my understanding is, hey, Boris didn't set out to basically conquer CDP. He was just trying to build really good tech that helps, um, data teams, uh, do do better work and be more successful at their jobs. Um, in company. So I totally understand Boris's point of view now. I don't actually know if he actually believes what he's saying. Um, because, um, it kind of then makes me question what he thinks a definition of a CDP is. So for example, I see reverse ETL tools as one component of many for a CDP. So he can't, he shouldn't, in my opinion, claim ownership of a CDP by saying, well, we're the CDP because we do this one piece of, uh, in concert with let's say eight other things. And so I, I don't think he actually believes in that because is, is high touch responsible or any reverse ETL vendor responsible for bringing in data in real time and batch or streaming. Are they responsible for the transformations and all the ETL inbound? Are they responsible for the identity resolution or are they responsible for actually shaping the the data in the data warehouse storage locations, which reverse details plug into? No, I mean, I don't think even Tesla would would agree to that. Um, what he would say is he can plug into the native sort of enterprise data warehouse, um, use that tool to make it easier for front end users to query the data, build audiences and have all these connectors to downstream MarTech, ad tech, CRM, what have you systems. Uh, so, so I, you know, I think it is a bit of a product marketing, uh, buzz just to get people to talk about it. So some pretty interesting perspectives here, right? Like David doesn't think reverse ETL tools can replace CDP's entirely. His view is that reverse ETL tools are just one component of a complete CDP system. While these tools make it easier for users to query data and build audiences, they don't handle all the other tasks that a CDP does. Therefore, claiming that a reverse ETL tool can serve as a CDP might be misleading and is misleading in some cases. They can complement CDP's, but they don't cover all the bases. Since we last spoke to Tejas, High Touch actually added ID resolution and event collection features to their product. And while they are built on your data warehouse versus copying your data, many in the market today are arguing that this is making high touch start to look a lot like a packaged CDP that they initially claimed to be dead. Alexandra Lomachenko, a data and MarTech consultant, is a big fan of the composable architecture for some companies, depending on their stage and their team. But she shared some of her thoughts on the evolution of the CDP. I would say this is a little bit cliche, but there is definitely nothing. You know, one size fits all. I'm really wondering how they will react to their customers growing and their customers needs growing as well. At one point, their customers, they will start requiring more functionalities and more sophisticated functionalities. And these companies they will need and the vendors will need to make a decision whether they are accepting and they start developing your functionalities around what they already have to keep the customer and small companies, they would be it will be more sensitive for them to lose one of their clients, or they will pass on, and they will keep true to their values that you know, they are keeping their functionality slides. They are keeping their, their, are bills lower as well. And so I'm actually very curious to see, you know, where all these smaller vendors start appearing right now, where they will end up. Will they end up being customer data platforms and warehouses and BI analytics solutions and customer engagement platforms that they are fighting right now. Or, or they will keep true to their values and will keep serving a wider range range of customers. And actually, you know, if they take the second pass, I actually see that this is very healthy for the industry because maybe, you know, you won't be able to serve everyone, but at the same time, you are truly democratizing it. So different companies would have of different stage. They will have an equal access to data and to quality data. So maybe it will give them a bump and it will increase their chances of survival, which is incredible. Uh, but at the same time, again, um, it will mean that vendors will need to make a decision at one point whether they want to stop development or they continue turning into, uh, big platforms that they are challenging right now. So there you have it, folks. High touch believes they can replace a package CDP because a few of their customers refer to them as their internal CDP. But Michael Michael Katz thinks that represents a very narrow segment of the market. He argues that a fragmented do it yourself approach doesn't optimize business value and isn't practical, especially for enterprise. On the other hand, Boris and Senses have taken a much less controversial approach to product marketing and opted for more honest messaging, if you will. They don't claim to replace IDPs. In fact, they're happy to work alongside them. Uh, the core idea behind census is creating tools that integrate well with existing systems without adding complexity. Census seeks to distribute trustworthy data across departments, leveraging the existing data infrastructure rather than creating another silo. David seems to side with census on this part of the debate. He doesn't think reverse ETL tools can replace CDPs entirely, and his view is that reverse ETL tools are just one component in a complete CDP system. Finally, Alexandra argues that vendors offering composable, lightweight solutions are making data more accessible, and they're a good thing for the market. But she points out that these vendors will face a strategic choice in the future either expand their feature sets to keep customers engaged as their customers grow, risking the credibility of their original bold statements like the CDP is dead, or focus on perfecting their core offerings without overhyping their capabilities. So you heard it here first, folks. I hope you got some, uh, fun and insights out of the debate here. Uh, the debate between packaged and composable CDPs boils down to a trade off between out of the box functionality and tailored flexibility, with industry opinions divided on what offers greater long term value. But if you do decide to explore the composable route, consider tools that focus on seamless integration and adaptability rather than those who claim to replace existing CDPs. I'll catch you guys next time.