Episode Transcript
[00:00:01] Speaker A: Hey everyone, welcome to this video. Or should I now be calling it a podcast or vodcast or a vlog or I don't know what I should be calling it. In any case, this is the inaugural something cast for Martech Therapy. More about that brand name to be announced soon, so please keep an eye on your LinkedIn feeds. Today I have a special guest, Michael Katz, the CEO of mParticle. Michael's here to discuss his recent CDP 2.0 blog series which provides an in depth look at the changes taking place in the field of customer data platforms. Before we start though, let me quickly tell you about Michael. Michael is a seasoned entrepreneur and technology leader who has built and scaled multiple successful companies in the data and marketing technology space. As the founder, or sorry should say, co founder and CEO of mParticle since 2013, he claims to have pioneered the customer data, the CDP category before it was even formally recognized, although other companies make the same claim. In any case, I'll be knocking on David Rapp's door to be Cernum at some point. Prior to mParticle, he founded Interclick, growing it organically to $140 million in revenue over five years before taking it public in 2009 and selling it to Yahoo in 2011 for $270 million. Now, for those of you not aware of Michael's communication style, let me enlighten you. Michael is known for his direct, no nonsense approach to business and industry discourse, notably so that you think that he's actually Dutch.
Well, this was demonstrated when he boldly declared DMPs are dead. Welcome to the CD P era exchanger event. Sorry for those voices, a statement that proved prescient as the industry evolved. He consistently challenges industry hype, particularly in the packaged aka traditional vs composable CDP debate, emphasizing the fundamental importance of data quality, integrity and privacy over marketing buzzwords. In its comprehensive article, Michael explores various aspects of modern CDPs, including the concept of zero waste performance, the role of AI in machine learning and marketing, and compute costs in data warehouse native solutions, and of course the different types of CDPs available in the market today. So let's go ahead, get Michael onto the call and get this show started.
Hey Michael, thanks for joining. I don't know if you know, but I just did a introduction where I announced that this is actually the inaugural episode podcast for MarTech Therapy. You are the first guest and hopefully not the last.
[00:02:41] Speaker B: What an honor.
[00:02:43] Speaker A: It definitely is. Definitely is.
We're doing a lot of rebranding or we. I should say I still one man army. But we're looking to expand with other professionals and act as therapists in the world of the crazy world of Martech. And as your article definitely does describe the CDP 2.0, I'm not sure is it a manifesto or a vision? I read a little bit like a brief history of what's happened so far and where we're headed.
What made you write this article?
[00:03:12] Speaker B: Well, this space has got very crowded, convoluted, you know, frustrating at times.
So I think for me, I wanted to set the record straight and give people some historical context around how we got to the point that we're in, because I think you would call that. I think another number of other people have observed there's been a bunch of change in the space recently.
So part of it was to deconstruct like, okay, why did the CDP thing become a thing in the first place?
And then what's transpired since then?
And so taking everybody through the journey from the rise of the cloud, creating a whole bunch of new consumer platforms and say marketing tech and ad tech tools to help drive growth on those consumer platforms and in the process having a lot of disjointed views of the customer, lots of customer incomplete customer profiles. And so CDP 1.0 was really about solving the integration challenges. Integrate your data sources, integrate your customer profiles, and then make it easy to integrate all that data downstream to your marketing tech stack and ad tech stack. Right. So the key benefit for the first number of years was just like it was just operational efficiency. You either had the problem or you didn't realize that you had the problem. But pretty much everybody had the problem.
Now what happened, I think starting in 2017, especially with the rise of GDPR privacy and the impact that it had on ad tech and marketing tech kind of took the industry in like a very different direction than where it had been in the past.
Previously, brands had invested a lot of money in third party data, third party cookies, third party tools, and they had really ignored the importance around building a first party data foundation. But like the ripple effect from not just new regulation, but also Apple and Google changing their policies around, you know, what identifiers you could access and what you, what you couldn't. And we could like, we could argue about whether that was just like, you know, a tactic for them to benefit their own ad businesses, but they did so under the guise of consumer privacy.
And what it forced was everybody to kind of rethink their acquisition and engagement strategies. Because what it did is it wiped out all the cheap acquisition right Remember, you know, this is like 2017, 2018, 2019.
People were just like addicted to the Facebook and Instagram cheap acquisition ads. Right. It was like crack to marketers.
The problem with that was like there was no real incentive to create better and better customer experiences. And so what it actually did, I think there is a net benefit to, to the consumer, although I don't think it's necessarily centered around privacy. I think it's that it changed the center of gravity.
And all brands who were heavily reliant on acquiring a bunch of users that went away. And so they had a focus on retaining a bunch of users. Right.
So in the process creating a really strong first party data foundation became mission critical. Like it was no longer a nice to have, it was a have to have. It was core.
[00:07:16] Speaker A: Yes, it must have.
[00:07:17] Speaker B: Yeah, it's core to your survival and digital success.
Which is why I think like that was kind of the tipping point where like everybody started to say, oh well, I have CDP capabilities, I'm a CDP too, I can do what they do.
And I think like when you deconstruct a lot of those offerings, not really doing all the things that CDPs do, what they did was they created a set of connectors, they made it so that data was more mobile or more portable throughout the ecosystem, which I think is kind of a feature of most modern software at this point. Even going beyond marketing tech and ad tech.
[00:08:04] Speaker A: This is what you refer to as a built in cdp.
[00:08:07] Speaker B: Yeah, typically when there's like a customer engagement platform that has quote unquote CDP like capabilities.
[00:08:18] Speaker A: Yeah, they have profile, unification, et cetera.
[00:08:22] Speaker B: Yeah, yeah, they do all the basics. Right. So when I talk about commoditization, it's that there's more players in the space and there's more solutions or alternatives that have the basic functionality.
So now the CDP's are at an inflection point and you have to really choose like do I want to play the commoditized low cost solution provider game, which from our vantage point is a, is a race to the bottom. Like we'll tell customers if all you need is like some data movement, just like you don't need, you don't need us. Like go with whatever the cheapest possible solution you can find and really don't spend more than like 10 grand a year if you need to do more sophisticated things. And this is kind of the other path that CDPs can take is to create premium offerings around either creating and bundling adjacencies or integrating intelligence. Like we've We've chosen our path.
[00:09:27] Speaker A: Yeah. But CDP 2.0 focusing on a specific vertical as you mentioned.
[00:09:32] Speaker B: Well, I think the vertical focus becomes critically important for a couple of reasons.
One I think like look, all the CDP players have built real technology and they real teams. Right.
But there's a bunch of nuance and the capabilities that each of the CDPs has built probably lends themselves more favorably or less favorably to certain verticals.
For us it's like real time identity and activation along with some of the kind of zero waste stuff that we built and the governance tools means that I think we're incredibly well positioned to help multi channel consumer brands deliver targeted relevant experiences across all screens and devices.
But if you're batch based and if you're completely reliant on a number of batch processes, the multichannel thing you kind.
[00:10:39] Speaker A: Of miss that becomes a challenge. No, definitely, huge challenge.
[00:10:43] Speaker B: Yeah. You have to move with the speed of the customer. Yeah.
[00:10:46] Speaker A: And definitely. And look, I've always been a person, of course in the beginning, you know, I've been, I've been partnering with a few CDP vendors over the last few years and I've, I've absolutely shown my in or favor to a few in the past which is, which has kind of been a mistake.
But now that I've, I've come to the realization that every customer has their unique needs, their you know, existing platforms, existing goals that they need to work towards.
And I agree with you that even though some might not consider the latency issues with these batch solutions a problem, there are always going to be companies who will put it at their number one in terms of priorities when selecting a cdp.
How have you experienced that so far?
[00:11:37] Speaker B: Yeah, to your point it's all customer context dependent. Right. Look, if you're a marketer and you're running, you know, five to 10 audiences, really small numbers, your customer experience isn't all that dynamic. There's long purchase consideration cycles.
Go with the batch based system, go with one of these compostable CDPs. Yeah.
[00:12:07] Speaker A: Slow decision processes.
[00:12:09] Speaker B: Yeah. Cause that's like it's going to be better suited for their needs and it's also going to be more cost effective. But as you start getting into more sophisticated use cases, whether you're running like multi brand portfolios, you're running across multiple regions or you have, you're operating multiple plat across multiple platforms and you want to engage across a variety of screens and devices like that no longer is suitable. You need something that is a lot less latent, a lot more real Time, a lot tighter governance controls, a lot less waste from a compute standpoint. And so it's really just like you got to align the key capabilities of the vendor with the key needs of.
[00:13:02] Speaker A: The customer and going quickly back to those built in CDPs. As you wrote in your first of all, what are we going to call it, a manifesto or just a blog series, kind of the Bible.
We'll get some people chasing us down with the black books, call it blasphemy there. But again, it was very thorough and especially towards the end and I have to say it was relatively unbiased. And of course, and this is just my perception, there was a little, I don't want to say animosity sounds like a harsh word, but I think in the end you were proving the case for mparticle for the product, which is fully understandable. Having it written on the mparticle website and coming back to the whole built in CDPs that you kind of labeled a lot of these engagement platforms who are enhancing their product with CDP capabilities, I believe you referenced that if they've commoditized just the basics of a cdp, you call them, they'll be able to solve so called small CDP problems. But products like mParticle, the more specialized or the more traditional, the ones who are moving beyond that CDP 1.0 can solve the big CDP problems.
Can you dive a little bit to what your thoughts are around what do you see as a big CDP problem? I've got a few, but I'd love to hear your thoughts on that.
[00:14:32] Speaker B: Sure.
I think I'll start by just saying the big problems of yesterday are today's small problems. And that's really what was meant. Right?
[00:14:43] Speaker A: Yeah.
[00:14:44] Speaker B: And so taking it back, the big problems of yesterday centered around all this fragmentation that was happening. Right. Access was incredibly difficult. Right. So difficult expensive too. So the problem that needed to be solved was operational efficiency. Ultimately, by and large, those problems have been addressed in spades. Right now we could argue about some of the more idiosyncratic differences in the quality of integrations and things like that, but I don't believe that the market really assigns too much value to that. So by and large we'll just say like those problems have been addressed.
The new opportunities create new challenges. You know, I think last year it would be tough to argue that like it didn't usher in like the era of, of AI and running highly compute intensive machine learning workloads in certain environments is going to be really, really expensive.
[00:15:56] Speaker A: Yeah.
[00:15:57] Speaker B: The problem directly is that within the cloud Data warehouse, the costs scale linearly and then exponentially as you run more queries, which.
[00:16:11] Speaker A: Reducing the latency.
[00:16:12] Speaker B: Yeah, yeah. With more audiences and then as you increase the rate of refresh.
[00:16:20] Speaker A: Yeah, definitely.
[00:16:21] Speaker B: It's a multiplier effect. So the bigger problems of today are how do you create cost optimized compute cost optimized solutions that don't erode all of the value that hopefully the marketing team is creating? Because our research has shown if you use one of these compostable CDPs, like for every dollar of progress that the marketing team makes, you will be giving 20 to $50 back to the cloud data warehouse provider.
[00:16:56] Speaker A: And that's based on the research that mparticles then? Yes, because that was definitely one of my questions.
How do companies using these composable solutions track down these potential hidden costs? I know I'm going to get a lot of flak on this because I think a lot of them will say there are no hidden costs. But from my experiences, from what I've seen so far and the arguments that I've heard, I can definitely see that it will go beyond the basic license costs. If you're looking at these solutions purely from the CP standpoint, you'll be paying for that part. But behind the scenes you still need to do a lot with your data. So how do you see what would be the best way for enterprises to kind of start thinking about these costs? As we move into a period where AI and ML is definitely going to be more front and center, we're going to be, you know, we're going to be more dependent on this moving forward.
[00:18:02] Speaker B: Yeah. I mean, one of the things that we've been doing is making sure that the CFO is involved in the conversation so that the, the data team, who rightfully so, has their really strong opinions on the right type of data architecture, but they've got, they've been fleeced by a few of these compostable vendors to focus on. Like if you've ever looked at a cloud hosting bill, you know that like 3% of the bill is storage and 97% of the bill is computer. This whole kind of zero copy thing is trading storage optimization. Right. Don't create too many data copies because all these bad things can happen Right. At the severe expense of compute.
That's not an equitable trade off. There's no CFO alive who thinks that that makes sense. Right. So first and foremost it's like get the CFO involved, start to do a total cost of ownership analysis. And total cost of ownership is like, once the data is connected, every time you want to add a new use case, how much engineering involvement is required? Because for mparticle, it's zero. But if you have to go into your data warehouse and create a new table and do a bunch of joins just to support the N Plus 1 use case, and there's infinite use cases, you probably need somewhere between three and five staff and like data engineers on. On staff. Right. So that's probably close to a million dollars right there, depending on kind of what part of the world you're. You're in.
[00:19:49] Speaker A: Yeah.
[00:19:50] Speaker B: And then you have to do a load balancing exercise. You can just run side by side. You baseline your existing cloud data warehouse bill and you run a bunch of audiences and you create a like, for, like comparison. And we did this and.
[00:20:06] Speaker A: Yeah, I was just about to ask, have you done this in practice where you've matched a composable solution towards a solution like mparticle to kind of see what the effects are?
[00:20:16] Speaker B: All right, yeah, we have, we have. And we're going to be publishing some pretty exciting research over the next few weeks.
Yeah, it's astonishing. And I think our findings in that 20x to 50x more expensive figure.
[00:20:34] Speaker A: I'll put the graph up for the viewers.
[00:20:36] Speaker B: Great. It doesn't even account for the entire set of tools. It doesn't account for the ETL tool, It doesn't account for the reverse ETL tool. It really just looks at the cost, the increased compute cost as a result of the architecture imposed by the compostable. Guys, let's call a spade a spade.
[00:21:12] Speaker A: As I mentioned when we were talking before the show started, I think you have Dutch roots because you're pretty straightforward. I need to be straightforward here. Are you calling it compostable on purpose or. I mean, I believe it's composable. All right, cool. We'll take that with a grain of salt then. But I just wanted to test the waters there. All right.
[00:21:34] Speaker B: I appreciate the Dutch compliment.
I'm transatlantic. I identify as American and Dutch, clearly. Yeah.
[00:21:43] Speaker A: And your attitude and your business doing is. It's definitely Dutch. It's no beating around the bush.
[00:21:48] Speaker B: Yeah. I mean, because, look, at the end of the day, when you waste a ton of compute cycles, who benefits? The cloud data warehouse provider, who loses? The customer flat out, like, full stuff.
[00:22:05] Speaker A: And I think the impact, I mean, another topic that I want to discuss in a bit here is agentic AI.
Everything is going to be covered in a layer of AI. So I can only imagine what that's going to do with Those hidden costs, or at least with those compute costs, it's not going to come for free. And I'm interested to learn how the market will develop. I mean will the composable CDP vendors, the data warehouse native vendors, how are they going to communicate this? So it's just a hypothetical question. Well they aren't now, but I wonder how is this going to change their approach? And it brings me to a quote from Dylan Fly. If I pronounce his name correctly, he's senior svp, so that must be Senior Vice President Sales over at Simon. He wrote something interesting which I think you can agree with. He writes, and I quote, the composable trend has significantly changed the CDP category over the last past several years. It's created much more cost and resource efficient marketing data operations. I'm not done here yet. So it has also shifted ownership of the CDP toward the IT or data Persona and pulled the CDP further from the stakeholders and business outcomes it was intended to impact. And I find that quote amazing because I guess everywhere you go the question that always brings out a six pack of beer on table is who owns the cdp? And I think the question we should be asking is how are we going to keep driving benefits out of IT to solve those business outcomes? How do you see that developing? And what do you think the impact of composable CDPs has been on that part?
[00:23:50] Speaker B: Well I think the legacy that composable CDPs will leave on the industry is the increased focus on the importance of being truly composable. Yeah, not the kind of co opted BS product marketing that they use.
[00:24:09] Speaker A: But composable from the. Sorry to interrupt. But composable from. Can we compare that to the Mac Alliance? Everything that falls under Mac Alliance. What's being determined as composable?
[00:24:21] Speaker B: Yeah, look the concept of composability within technology markets existed long before a couple of mouthy startups from San Francisco decided to bastardize the term.
Composability is about.
[00:24:36] Speaker A: Not my words. Not my words.
[00:24:38] Speaker B: Yeah, mine. Attribute it to me and me only.
The thing about composability is like modularity and configurability and flexible pricing. All of that stuff is really good.
What we say is that all CDPs should be composable and most of them actually are. Very few are selling this like big monolith and forcing you to buy stuff that you're never going to use. Right. So that's really good.
[00:25:10] Speaker A: Hell even Salesforce is doing composable with. You can use Salesforce marketing cloud all by itself if you don't want to use commerce cloud for instance, but their whole platform. I get it. Yeah.
[00:25:20] Speaker B: Salesforce will sell whatever they.
[00:25:24] Speaker A: I don't have enough alcohol in the house to start that discussion.
[00:25:28] Speaker B: Yeah.
So we got sidetracked. What was the original question?
[00:25:33] Speaker A: No, the question around the kind of the ownership of a cdp, how composable has shifted it more towards the engineering department. While others seem to focus on the impact for the marketeers while the business outcomes have been left out in the cold.
[00:25:51] Speaker B: Yeah. I mean, you know, we've long championed this idea of like data as a team sport.
[00:25:56] Speaker A: Yeah.
[00:25:56] Speaker B: Right. And the point is like you do need both sides to come together. Both sides being data producers and data consumers. Now rewind the clock. Ten years ago the data producers were the web and mobile developers. Now it's the of core data and IT team.
[00:26:17] Speaker A: Yeah. And even the, even the services you use as a company, your payment gateways, et cetera, there's so much data available to you.
[00:26:23] Speaker B: Yeah, exactly. And in a healthy organization, there's processes and controls, but there's collaboration between the producers and the consumers because there has to be. Right. In theory, kind of all on the same team working for the same company. The end result that they're all trying to drive is growth and increase in their, in their equity value. Yeah.
[00:26:52] Speaker A: Definitely the shipping.
[00:26:53] Speaker B: Yeah. You know, I think where the disconnect comes from is like marketers, like the idea of like the technical marketer just never really came to fruition. Like they're few and far between and the ones who are like, these guys are rock stars. Like I love working with some of our customers who have, who have technical marketers. There's a guy at Marks and Spencer, I think he's one of the brightest minds in the space.
And on the flip side, you have data teams who think that they understand the nuances of marketing and how data needs to be treated to drive marketing and engagement because they've spun up a bunch of like bi and reporting dashboards.
[00:27:46] Speaker A: Yeah.
[00:27:47] Speaker B: Data for analytics, even if it's the same data set, needs to be treated quite differently for marketing purposes. And I don't think that there's a strong appreciation on either side for the kind of nuance and complexity that that respective side has to deal with in order to do their jobs effectively. And so what you've, what you've seen happen in the space, it's not necessarily data led buying decisions versus marketer led buying decisions. It's really about influence and collaboration. In certain organizations. The data teams are like, they're highly influential to A point of interfering and there's very little collaboration. And in other organizations there's a lot of collaboration and they have influence but because there's also collaboration it just, it works and they make pragmatic, rational decisions.
[00:28:49] Speaker A: Yeah, no, absolutely you're right.
I posted it on it I think was last week I rounded off a contract with an American car listing company.
I was trying to avoid brand names and et cetera in this whole, in this whole recording. So I won't name them but what I found interesting about them is from day one they had it right and you know that this, the project was going to be their success. I got brought on board to help at least advise them as a subject matter expert not only on the topic of CDPs but also for a CDP brand. But they had everyone in the room, they had something called a CDP boat and I've been trying to look it up what this it's a form of, I don't want to say agile but the boat consists of leads from different departments. So we had someone working with the backend, someone on the front end. We had a data scientist, data engineered. We even had someone who would become the product owner and the CDP champion. And I said, jokingly said in my post she didn't know it at the time because once this whole momentum got going of this team working together to implement this, that's actually where the, that's when the value started being generated. I, you know, hypothetically of course but it's, it's, it's, it's from that first day when people just realized we need to collaborate on this. And it always hurts me when, not emotionally but you know, when people say it or marketing or whatever department owns the cdp, it's that, that's the, that's the wrong way to start. We need all, everyone needs to be accountable for what happens with the CDP because as a company you all stand to benefit. And yeah that. But that brings me back to another point about the built in. Sorry I'm kind of jumping around topics here but no problem around I personally I do see a benefit of these built in CDPs in that they open up the door for small and medium sized businesses.
Been speaking to various customers over the last few days about an article I wrote or a video that I recorded about a Dutch company who, that who went bankrupt and personally I think that their choice of Martech did play a small role end that entire process just be costing a lot of money in terms of resources to get it integrated into the entire business again. It's all conjecture. But what I see now with these built in CDPs and in terms of pricing, I mean it's incredible. It's really lowering the bar for these small medium sized companies, family businesses to really get started and I'm seeing it happen.
I mean even if it's still CDP 1.0, if we compare it to the 2.0 Bible, it's still going to benefit.
How do you see that? What would be the way forward for these companies? I guess it all comes down to keep retaining more money at the end of the year below the line to seriously start considering more, more mature CDP products or how do you see that happening? What are some of the constrictions that they could potentially run into over time?
[00:32:14] Speaker B: Yeah, so I, I do tend to agree with the notion that for mid market, maybe smaller brands that having.
Well and in within those organizations have limited resources themselves. Maybe really small marketing department departments and smaller marketing budgets. Yeah, go with the built in solution.
You know email has been around for a very long period of time so there's generations now of knowledge workers that have used email tools. Yeah.
[00:32:50] Speaker A: Membership was my first one I believe.
[00:32:52] Speaker B: Yeah, yeah, yeah.
[00:32:55] Speaker A: So what was that, 15, 15, 16 years ago? Yeah, it's older than two of my kids at least.
[00:33:05] Speaker B: So if people are going to become more proficient with using customer data platforms and getting familiar with the concepts of it and how it creates value above and beyond just connecting to a set of tools, I think the built in solutions are like a kind of great entry point. Right. It gets them familiar.
Hopefully it helps contribute to growth and at some point what they'll need, assuming that they are successful is a more robust solution. Right. A more robust solution that can connect directly with their engagement solutions that you know, for us it's not only about ensuring that we maintain a certain level of functionality so there's no functionality lost when you're using mparticle and connecting to say braze. But actually like because of our intelligence layer we actually make the data better.
We augment customer profiles with richer context.
[00:34:20] Speaker A: Yeah.
[00:34:21] Speaker B: And predictions and so now you can do things in your customer engagement platform that you couldn't really do otherwise. Right. But that's like, you know, there's building blocks to get there that that is not the solution for everybody. And most small companies, probably not.
[00:34:38] Speaker A: No. And how has the.
I remember last year mparticle announced the kind of, I believe it's called differently but kind of the pay as you go model where you fully revamped the kind of the pricing structure how has that been working out? Has it been attracting more of the SMBs on board or has it resulted in existing customers working more smartly with the, with the kind of the credits that they have within mparticle itself?
[00:35:10] Speaker B: It's all, I mean, you know, when we announced it last year, admittedly it was like half baked. I think like we knew we were onto something because we had been introducing it to existing customers for a little bit. But the architecture of it was convoluted and it was kind of confusing. And I think with that one we probably shipped the org chart.
Over the course of the past couple quarters we've worked greatly reduce the complexity and just simplify it. And you know, I think what work, what we've been able to do is still capture the essence of why we moved that direction in the first place.
[00:35:55] Speaker A: Yeah.
[00:35:56] Speaker B: Which is there's a problem with pricing on like on a user profile basis or like monthly tracked user. Because the problem is like your more valuable users have a lot more data than your less valuable, less engaged users, but you're paying the same for.
[00:36:20] Speaker A: Exactly.
[00:36:20] Speaker B: For both.
And that doesn't really make sense or.
Yeah, or you have some of these models that just create like an upcharge on your cloud data warehouse bill, like 20, say whatever you're paying them, you know, pay us, you know, 20% premium. Yeah, marketers don't understand that. Like they have no way, like it's, it's a total black box. Right.
[00:36:50] Speaker A: Yeah.
[00:36:51] Speaker B: And so we felt like first and foremost the best measure of value was moving towards an events basis. Because that way if there are users with lots of events, meaning they are more valuable because we're charging on an event basis, like that user is going to be more expensive than the person who maybe comes to the website and drops off right away. So it's all kind of. It was about creating alignment between costs and value, first and foremost. And then the second part of it was to instantiate this credits model. Because the other problem with CDP pricing is you kind of have to opt into the core thing and then there's all these modules and guess what, 100% of the time customers are going to be wrong. Whether it's slightly to very wrong about their desired configuration, they make a set of assumptions and then when those assumptions turn out to be like slightly to very off, they get penalized. They have to buy more of one module or they get hit with caps and they can't use enough of part of the system. Yeah, we heard that for years and years.
[00:38:06] Speaker A: Still hearing it.
[00:38:08] Speaker B: Yeah. Yeah, yeah. So like there's this asymmetry that benefits the vendor at the expense of the customer. For us, we recognize, you know, you're probably not going to get it right 100% upfront and so you shouldn't be penalized if you need to swap out credits for other parts of the platform. You don't have to pre commit to using those things, but you're not penalized if you do. And so you create kind of a win win.
[00:38:35] Speaker A: Yeah, I've seen the ramp up especially with new implementations of CDPs. It's not something that happens overnight and anyone who says it does, don't take it seriously. There's a lot involved in getting, especially on an enterprise level and getting that CDP up and running and there's a lot of things that users have to pay for from day one but they won't be able to use for at least four or five months in some real worst case scenarios. And I always find it intriguing that they're still forced to pay this. Like there's no ramp up pricing or, and quite often there's not, there's no ramp up pricing or even you know, a way to kind of facilitate the spike in acquisition of users to the website or to the mobile app that just really picks out your mtus.
I don't know. Personally I hope that moving forward we see some more development in terms of pricing there. Again, I can't look into your kitchen and behind the scenes to everything.
Yeah, that, that needs, that gets involved there. But that I think for like you said it needs to benefit the customers in the end.
[00:39:51] Speaker B: Yeah. You know there's, there's three problems. I think you alluded to two of them. Right. Like being able to ramp up and overspend during that, during that ramp up.
Then there's like the ebbs and flows of the normal course of business. If you're, if you're a business that has seasonality.
[00:40:12] Speaker A: Oh yeah.
[00:40:13] Speaker B: You shouldn't get penalized for the spikes, the expected spikes in the business. Like everything could be on an unexpected.
[00:40:21] Speaker A: Growth for that matter, even if you're not seasonality, I mean seasonality you can kind of average out over the, you know, a 12 month period. But if you experience unexpected growth, which is good, but you get hit hard on the other end.
[00:40:34] Speaker B: Yeah. And then, and then when there's like multi year arrangements, you should be able to roll over unused credits because again every customer goes in with like the best intentions and the best set of assumptions that they can make and you know, what's the Saying it's like all models are flawed, but some are helpful or useful.
It's kind of the same thing. It's like you're never going to be 100% right, so why get penalized? Like the whole thing is just really, really strange to me.
[00:41:11] Speaker A: It is, it is. I mean, and like I said, I hope the 2025, 26 brings some refreshing new thoughts and ideas around pricing that would benefit the customers. Because as we're seeing, as Scott Brinker pointed out with Franz Riemersma, the whole consolidation of Martech solutions is happening. I believe the Simon Data article pointed out that the, you know, the multiple CDPs has decreased from 2.8 on average to 2.1. But that again is taking a kind of broad perspective on what a CDP stands for, stands for.
And I think that's only going to reduce. So in the end there needs to be the benefits for the customers who are going to commit to you as a vendor, as an architecture that you also commit to them on price flexibility, moving, moving forward and growing with them. And it's a lot of marketing and sales talk that I usually hear, but I haven't really seen it come to fruition yet.
[00:42:08] Speaker B: Yeah, well, I think, you know, even going back to the CDP 1.0, you know, ascension, it was, you had all this, all this fragmentation, you had all this unbundling of various point solutions. It became really easy to start a new ad tech or marketing tech company and spin up a nice little tool and hopefully find your attack vector. So for 10 years, 10 plus years, we were in this phase of unbundling.
I think coinciding with the rise of CDP 2.0, I think we've entered a period of rebundling. So you said consolidation. I think about it a little bit as like convergence.
Right.
[00:42:57] Speaker A: Well then the built in CDPs are doing the right thing. Maybe.
[00:43:01] Speaker B: I think that they are to a large degree. Now it doesn't necessarily mean that they can create value on a dollar for dollar pound.
[00:43:09] Speaker A: No, no, definitely. We're just waiting until you do emails as well, right?
[00:43:12] Speaker B: Nah, that's probably a space where we're never going. I never say never, but never say never.
[00:43:18] Speaker A: No, indeed.
[00:43:19] Speaker B: You know, we've, we've made that a kind of a deliberate part of our ethos. Yeah, but you know, we did acquire a machine learning company and a customer journey analytics tool a couple of years ago.
So we do see the system of action or data movement converging with the system of intelligence. I think that has to be because it's not just about dumb pipes anymore. You need intelligent data movement. Right now you could say, okay, at what point does the rebundling end? I don't think that there should be one software that addresses everything for all categories across whether it's ad tech, marketing tech or just software in general. I think that would be way too much. The adjacencies have to be accretive in terms of creating incremental customer value. So we're going to kind of see those lines continue to blend. And I think what we're going to start to see more and more are suites. Right. I think Lytics announced what they call it like the X cdp.
[00:44:40] Speaker A: Oh yeah, the Experience cdp. That's right. So that I think it comes down to kind of a content layer where they can actually push the or have the user profile. Some extra elements in the data layer that the CMS, I believe a WordPress and Drupal can latch onto.
[00:45:00] Speaker B: Yeah. And admittedly that's a very different approach than we're taking, but I commend them for placing the bet and saying we have to move beyond.
[00:45:13] Speaker A: And it's a true pain point from what I've seen is when companies do deploy cdp, setting up those pipelines, the sources and the endpoints, that's fine. Finding the data works great.
But then they talk about, hey, all right, but how do we get personalization on the website done? How do we get content personalized for the individual user? And that always from my experiences and I'd love to hear from other people who found a way to do this, but on an easy way. But it always seems the biggest challenge we face is how do we leverage the user profiles for content. And of course the landing page, the hero image, you know that I don't want to say I can do that, as the Dutch say with two fingers in my nose.
But that's not the challenging part anymore. How do we go about serving content smartly?
[00:46:06] Speaker B: That's right, yeah, yeah. No, I think, you know, I'm not making any kind of prognostications.
[00:46:12] Speaker A: No, no, no, no, no.
[00:46:14] Speaker B: But I do.
[00:46:17] Speaker A: I think it is going to be a trend to get these integration with cmss. But I don't want to be the first one to call it, I think Linux all credits to them that they've seen this and I'm definitely going to be watching how that rolls out because there's also another CDP or well actually they're calling them a built in now who has an integration with Adobe Experience platform. So it's slowly coming. But it's going to be interesting to see how well that's going to evolve. But on that topic, you also mentioned something about a cloud data warehouse overlay architecture that's going to address data governance and compute optimization. Looking at the time we have left, is that an interesting topic to kind of wind down about? It's not something we grab whiskey here and kick back because it's a heavy topic.
[00:47:07] Speaker B: I think it's also 9:47 in the morning in New York.
[00:47:11] Speaker A: Yeah, no whiskey yet.
I know.
[00:47:14] Speaker B: You guys are a few hours ahead of us.
[00:47:16] Speaker A: Exactly. Yeah. Six. Yeah. But it's. That caught my attention because you mentioned that a lot of parties are actually more or less overlays on top of the cloud data warehouse, which is true. I mean, it's just a fact.
But that mparticle is planning something similar. Is that something you can kind of do a sneak preview on or kind of tease us a little bit more about what you have in store?
[00:47:41] Speaker B: Uh, yeah, I would just say like, wait, wait and see. You know, in the coming quarters where we're heading there, I'm pretty excited. But I think that there's just, there's a much better way of doing things. Right? Like the warehouse native thing became it like composable, became conflated with like a number of things, right? It was like warehouse native and zero copy and time to first query, which they tricked people into thinking that was like the same as time to value. Right. A couple more things. But regardless, like, if we just hone in on like the zero copy thing.
Zero copy is only a marketing scheme to reallocate compute costs, unoptimized compute costs, back into the cloud data warehouse. Right? And the reason that they're unoptimized is because every time you need to do an audience refresh, you need to scan all the data again. Yeah, you shouldn't have to scan all the data again. And this is where zero copy, in its kind of most literal interpretation makes absolutely no sense. It's just, it's unproductive because you can create.
You can create a cache, a temporary cache of the data and instantiate an architecture that allows costs to scale sublinearly rather than exponentially.
[00:49:15] Speaker A: And you hit the nail on the head there because I think a few weeks ago I read an article about even BigQuery doing this, that if you write out a SQL query to query your raw data, it'll come back with data set. But I think they're even focusing more attention on that. Even if you have a little typo or you just make one little change to that query, you're not going to get impact. And I know something like this has always been around for the die hard BigQuery fans out there, but I read somewhere and I'll dig up and publish a link somewhere but that they're going to optimize that even more. And I think it's for the same reasons that these costs and the amount of data that people are now have available in these data warehouses is staggering.
[00:49:59] Speaker B: Yeah. Well look BigQuery, like Google's in a position to do that because they have the largest ad business in the world. Right. So they can.
[00:50:09] Speaker A: Until they sell Chrome. Right. It's going to be interesting. I've read somewhere that it might impact them 66% market share. It's wow.
[00:50:17] Speaker B: Yeah, but I mean I think YouTube is really driving the lion's share. I mean YouTube and Google search still to this day, I mean it's like they're, they're the equivalent of utilities on the Internet. Right.
So they have the financial means and wherewithal to be able to go and subsidize those types of decisions that are truly in the best interest of the customer. Whereas you have people like Snowflake. Snowflake is just, their business model is to arbitrage compute. Right. Like they, they buy a bunch of compute wholesale and you know, across the different cloud providers they mark it up and they, and they resell it.
[00:51:03] Speaker A: Yeah.
[00:51:04] Speaker B: So the more compute that they can run, the more that they make. It's not in their vested interest. Like. Right. It's like, and this isn't even like conspiracy theory nonsense, it's just like follow the money one hop. It is not in their vested interest to create optimized like have an optimized query engine. It's just not.
[00:51:26] Speaker A: And their recent announcement of the partnership with Anthropic is going to drive, it's going to drive through the roof.
[00:51:33] Speaker B: Totally. Totally. Yeah. I mean it's, it's, it's, it's impressive but it's impressive in the way that like this, the Stackler family pushed a whole bunch of opioids on a bunch of Americans and made billions of dollars. It's evil.
[00:51:51] Speaker A: Yeah, definitely. Hey, to end it all off with first, thank you very much for your time. It's eight minutes to the top of the hour. As the radio DJs used to say when I was listening to US radio living in Dallas many, many years ago.
Just kind of, you don't need to commit. What is your biggest prediction? Maybe not for mparticle or CDP specific. But for Martech at least. What is your biggest prediction for 2025? You're not allowed to use AI or agentic AI.
[00:52:27] Speaker B: Gladly.
I think it's that the CFO continues to be kind of front and center in between the marketing team and the data team because that's really how you create the right type of balance. Marketer has their views. Not saying all marketer views are right. Data team has their views. Not all their views are right. The CFO has to be inserted into the conversation. I think that that started to happen a couple years ago when you had like the economic downturn and you had all the scrutiny placed on software spend.
But I think that this is.
It created a more kind of healthy dynamic, but it wasn't necessarily like a formal thing. Yeah, this probably isn't like the sexiest prediction, but it's one that I think makes pragmatic sense.
[00:53:19] Speaker A: Well, if I think about personally, I mean if I look at myself as a consultant, I need to position myself between marketing and engineering, but my bills are usually paid by one or the other and I sort of expected to lean to one side or the other. And it makes it a little bit difficult at times to remain unbiased in a situation like that. And with your prediction around the CFO playing, taking a part in that.
Personally, I would like to see solutions like CDP be managed from a kind of a new neutral team within the entire organization. Since it's, you know, we're talking multi, what do I keep using? Multidisciplinary, cross departmental. Lovely buzzwords, but they shouldn't be tied in terms of responsibility and accountability to just either marketing or engineering or a business unit. It needs to be a broad solution. So I liked your thought about that CFO part and I see him as a neutral party in the entire situation looking at business outcomes. So that. Yeah, that's a great. Yeah, I'm going to be hoping for it. I'm going to do a search for CFOs to connect with now on LinkedIn and say, hey, listen to what this guy Michael Katz just has proposed as a thought for 2025.
[00:54:42] Speaker B: Yeah, look, I also do think in 2025 what we are going to see is like a bunch of marketing tech companies and more specifically CDPs who raised a lot of money in or around the 2021 ZIRP boom.
They're coming towards the end of their Runway and I think that that will further perpetuate a lot of this. Call it convergence or consolidation.
[00:55:12] Speaker A: Yeah, yeah, I see that happening too. I mean, at some point the differentiation is going to be difficult to argue in sales pitches and. Yeah, no, definitely. Well, that's 55 minutes. I, again, really appreciate the time. It's always good talking about CDPs and going a little bit more in depth than usual and.
[00:55:34] Speaker B: Absolutely.
[00:55:35] Speaker A: Thanks a lot for your time.
[00:55:37] Speaker B: This is great. Thank you.
