#59 - Data Science That Drives Business Outcomes

Change Agent Kumar Subramanyam of HP joins the podcast to break the role of data science in driving business outcomes, why activation doesn't matter if you can't measure it, and how to keep an eye toward future customer value. It's a conversation you won't want to miss!

Thanks for listening! Follow us on Twitter and Instagram or find us on Facebook.

Transcript

Announcer (00:01):

Welcome to Real Talk about Real Marketing, an Acxiom podcast where we discuss marketing made better, bringing you real challenges and emerging trends. Marketers face today,

Dustin Raney (00:14):

Man, am I excited about today's episode? It's not often, but when we do, I'm super excited to hear from people in the field actually, brands that are putting technology to use, that are thinking forward [00:00:30] thinking. I had the opportunity here over the past couple of years of needing our guest today, Kamar Rahm, global Head of Marketing Data Sciences at hp, and that is Hewlett Packard. So Kumar, thank you for joining us today on Real Talk. It is truly an honor. We had the opportunity to have a great conversation just a couple of weeks ago while out in Palo Alto, you are a true change agent, which is one of the reasons I immediately were [00:01:00] like, man, we have to get Kumar on the podcast. Really excited about our conversation today. If you don't mind, tell our guests about how you got to where you are today at hp.

Kumar Subramanyam (01:10):

Absolutely. Happy to. And first and foremost, thank you Dustin, and thank you Carl for having me on this podcast. Really love to share my perspective, albeit a little more philosophical than more tactical to be specific. So how did I get here? I've been [00:01:30] at HP in this role doing data sciences for the last 14 years. Some people would say, yeah, you've been doing exactly the same job for the last 14 years. Isn't that wrong? Or isn't that not a good thing? No, I've been moving up the ladder, but essentially I've been the person who hired me many years back. I have his job, so to speak at this point, but how did I get here? As the head of data sciences and data sciences [00:02:00] means different things to different people. But let me quickly give you a brief of what we do. The data sciences function in marketing, the global data sciences function actually has three slightly different elements to it.

(02:12):

One is the measurement data infrastructure, which includes managing all the data that we collect from media, from our campaign brand market share. I mean, talk about all the different types of data that we get, and the first part is assembling it, and we [00:02:30] have the whole infrastructure team that sits underneath me to manage the data, collect the data specifically with the focus around measurement. Data sciences is predominantly measurement focused. Yes, we do some audience work, audience targeting, et cetera, but 90% of our work tends to be more around measurement. Then the third piece is reporting ultimately, whether you call it dashboards, whether you call it scorecards, that's the third piece. Now, that's sort of the set [00:03:00] of services that's managed by my team. Again, it's a global team across all three regions, and then we have some functions that are centralized. Now my history coming into hp, again, kind of dating back, I graduated like let's say early nineties. I'm not going to date it for many different reasons, but let's just say early nineties, early to mid nineties. Let put it that way. I graduated out [00:03:30] from Villanova. Again, Wildcats, anybody's out there. I know you guys are Razorback fans, but hey,

Dustin Raney (03:37):

Go hogs.

Kumar Subramanyam (03:38):

Go. Hi now. Thanks Justin. So long story short, graduated out, worked in Wall Street for a couple of years, moved to the west coast. I worked many years at Sun Engineering, and then Sun was in the heydays of sun, and then fast forward, quit Sun. I think around 2005 I started my own company, two of MI [00:04:00] basically sold those two venture back, and then in 2000, and I'm guessing immediately after the recession, I was just punting around wondering what to do with my life. Then fortunately, I have a DHD, which I'm always open about. I don't conceal it. I'm not on any more medication to be specific. I think especially once I got into my fifties. But a DHD gives me a unique perspective, which [00:04:30] some people don't necessarily realize. When I used to get into a bus, I would count the number of doors.

(04:36):

We all count the number of doors, you want to know where to enter or exit, but then the secondary element is like how many windows, how many people in this bus or how many handrails or whatever. It's start counting. So it became a conversation about data. That's when I realized I had the startups that I did and even post that always data was pretty core and essential [00:05:00] to everything that we do. Now, fast forward, that's when I came into hp. My entire job was all about data management, data sciences, although we used to call ourselves analytics back then, and then reporting of course is a vector that sits on top of all of this. Hopefully that gives you a little bit of a rounded perspective.

Dustin Raney (05:21):

No, that's great. And man, HP massive company to be the global head of marketing, data science is a super [00:05:30] important role, and you talked about measurement a lot when we were out in Palo Alto, our member and just how things are changing. We talked about ID deprecation, cookie loss and stuff like that, and you brought some kind of perspective around things you guys are doing in spite of some of those challenges around measurement. You want to talk a little bit about that? I think you were quoted saying that [00:06:00] marketing activation is great, but it doesn't mean anything if I can't measure. Can't measure, right?

Kumar Subramanyam (06:06):

Exactly. Exactly. So when I look at even the evolution of the data sciences function, I'm just going to say last five to six years, about five years back, maybe six years to be more specific, right? I mean, again, my dates might be a little wonky here. So we had a DMP through the DMP, we used to run a lot of our media activation, cookie-based activation. More specifically [00:06:30] though, biggest struggle that we ran into. We were doing a great job selecting all these audiences, but at the end of the day when we had to assess the return on marketing investment associated with those cohorts or audiences we're like what? There were a couple of companies that actually had models that allowed us to get that, and then we had our own internal MMM, but the end of the day, it was a struggle for us to think about it. So the question that we started [00:07:00] asking is audience activation that critical without our ability to be able to measure. So that's where I'm guessing about five years back or four years back, and again, my dates are always a little wonky, so my apologies. We made a massive shift towards a measurement first approach, which is if we can measure, we don't put our money there, we don't run tactics there,

(07:24):

And that's been core to how we have operated over the last five years minimally.

Kyle Hollaway (07:29):

So [00:07:30] unpack that a little bit more. Obviously measurement, totally get that aspect of if you can't measure it, then why do it to really understand where your money's going, where your emphasis and such. So in the last couple of years, you said it's about four or five years, you've really started to shift that focus. Talk to me about the ecosystem. Where have you seen challenges with measurement? Do you see certain pockets where [00:08:00] measurements really effective and working well and areas that maybe measurement's very hard to get to?

Kumar Subramanyam (08:10):

Yeah, absolutely, absolutely, and I'll touch upon each one of those, right? When we got into this whole space of measurement, the first biggest struggle that we had is how do you think about measurement from a framing perspective, measurement could mean a lot of different things. So this took us a while to figure out about north of six months, in my opinion, and [00:08:30] the most important thing that we figured out, step number one was to frame measurement into three specific buckets, impact planning and optimization, and we call it the IPO framework. I know it's like a corny term, but hey, it's easier to remember than calling it IOP framework or whatever. But impact is all about what have I done as of yesterday? It's backwards looking. It's like, okay, the money that we spent, what did we [00:09:00] achieve as a result of it, whether it be sponsorships, whether it be media, whether it be emails, whatever that example is.

(09:07):

The second one is now that I know what I've accomplished as of yesterday, what should I be doing in the next 30 69 gs? That's what we call as optimization, very much short term focused in terms of shifting dollars, whether it be from meta to Google or whatever those examples are. Then the third one is basically planning, which is focused around, okay, now that I [00:09:30] know what my historical performances, if I had 50 million to spend in the beginning of the next fiscal year, where would I put that money? And planning essentially has two slightly different subsections to it. One is strategic planning, and another one is scenario planning. Strategic planning is, okay, let's think about the print category for hp. Print category includes a lot of different things. I mean, we have products, we have inkjet printers, we have laser jet printers, we have supplies, we [00:10:00] have instant ink.

(10:01):

We have many different ways of looking at it. So the problem is how do you take the $50 million in this example and optimize your spend across these different subcategories, so to speak? Why, as an example, if let's say we push an inkjet, so to speak, instead of laser, what does that mean from a business impact perspective? We need to account for it from a profitability, from a revenue, et cetera. And so how do you balance [00:10:30] that? That's the larger context of scenario planning and being able to turn around and say, okay, hey, this is the minimum that we need to be able to spend in every quarter within a region or within a market, and this is the maximum that we should be spending. Because the other element of this is you spend stop spending by the time you ramp up back again, it takes time, whether it be from a media perspective or even from an optimization perspective.

(10:53):

So that's the most important thing that we had to solve. And then the second part of planning is actual scenario [00:11:00] planning, which is now turning around and saying, okay, let's assume 10 million of that 50 million is going to go towards our inkjet marketing in this example. So what are the tactics that I need to run? Do I need to put all 10 million on Google, or do I need to spend five here, five there? Do I need to front end with YouTube followed by paid search? All those things become part of the scenario planning vector. Now, as we thought through these, and [00:11:30] this has been an evolution for us, the one thing, whether it be impact planning or optimization, the one thing we need to be absolutely clear about is, or rather we had to be absolutely clear about is what is that north star metric that we care about? And we talk about North Star. Is it market share? Is it revenue? Is it profitability? Today, we don't do much around profitability, but that's sort of where we want to go. But minimally on market share, minimally on revenue and on units, [00:12:00] those are the three vectors that we care about. So that's sort of how the framing looks. Now, where is the good and the bad and the ugly, right? HP is 80% commercial, about 20%, give or take consumer. The problem is we have the solutions for consumer across

Kumar Subramanyam (10:53):

Kumar Subramanyam (12:17):

The board across six of the eight markets, but B2B is a massive struggle for us right now. And if there is a problem that's like this big monster that's sitting in front of us, how do we measure [00:12:30] the impact of B2B across the board? And more importantly, it's not just about measuring. It's like, okay, I have to go in front of the sales team and have them sign off on the measure, otherwise I'm basically taking credit for every Christmas card that I sent to them. So anyway, that's my 2 cents on that.

Dustin Raney (12:51):

And I would think that digital transformation the past 10 years, especially in B2B, have [00:13:00] been way more difficult than B2C because business owners usually when they would walk into a store or something like that, there's contracts that would be written, but it's consumers. If I own a business and I'm buying for my business, I'm typically not telling you. So that kind of becomes a first party data issue. Right?

Kumar Subramanyam (13:23):

Exactly.

Dustin Raney (13:25):

So I mean, with that said, that's where identity [00:13:30] fits, right? Are you guys starting to think about how to connect those dots between business owners and consumers and even on devices and whatnot?

Kumar Subramanyam (13:41):

Absolutely. Absolutely. And again, I'll rewind a bit, right? Going back to the days of the cookie, and I think I made a joke about this several times. I have multiple ality disorder in addition to the A DHD, right? So now let's keep count of all of these issues that I have. So you got to assume what my spouse goes [00:14:00] through on a daily basis. But when I think about cookie data, at one point I did an assessment about my personal data and there were 38 versions of me. I get it. I have multiple personality disorder, but I don't have 38 versions. I can tell you there aren't that many hours in a day. So now, if I were to unpack that, that's historically been the struggle, not just from an activation or a measurement perspective. How do you use industry standard [00:14:30] ID graphs to be able to measure value for marketing?

(14:33):

And it hasn't worked, but when the ecosystem thanks to GDPR and dovetail by CCPA and whatnot, while as a marketer, I enjoy the laws or what, that's a whole other argument. But as a result of that, what came about is brilliant. Everything is now driven off people data. The interesting thing is there's only one Dustin, there's only one [00:15:00] Kyle, there's only one Kumar. Maybe we have three different email addresses. Maybe we have a professional email address, we have personal email address, and we have a junk email address. Each of us tend to have that. So the question is how do we connect those elements from a marketing standpoint? And that's been one of the things. We've actually done a very good job on the consumer side, and the struggle is now kind of evolving that anti RAF to kind of get to a B2C two B.

(15:27):

That second part is tough [00:15:30] because the B2B part of it is easier in my mind because you probably have my email address, my HP email address, and you can mark it to me. And if I'm a decision maker, it's pretty easy to reach out to me. But how do I drive this conversation in the larger media ecosystem, which is mostly consumer focused and drive a message home that drives some sort of an awareness into my head as a business decision maker at some point. I think that's the shift that I think [00:16:00] we're beginning to see in the industry.

Kyle Hollaway (16:03):

Yeah. So do you see that shift certainly as we recognize individuals and being able to, like I said, be more people based, but do you see it marrying up more with historical contextual signals to understand it's Kumar and it may be Kumar at your personal address, but you [00:16:30] are interacting on office depot.com or something, and we're able to associate, oh, that's the same Kumar, that's Kumar at hp, therefore this may be a business purchase. So where does contextual come in conjunction with the people based identity?

Kumar Subramanyam (16:50):

Absolutely. Great question. If there are two ways of targeting going forward, one is people based, the other one is contextual, [00:17:00] right? Especially in the B2B world, contextual tends to, in my opinion, personal opinion, drive more demonstrative value to initiate a B2B conversation. Where I think we still are seeing evolving solutions, especially in terms of connecting the ID graphs, is how do I now turn around and attribute value? So now in the contextual case, let's say somehow [00:17:30] we have access to that email address and the phone number and whatnot, and we're able to do some sort of validation saying, okay, I reached out to Dustin or I reached out to Kyle, but then Dustin and Kyle have a consumer persona that in turn is also seeing ads of HP at the same time.

(17:49):

How are these two jointly influencing? I think those are some of the more critical problems that we see up ahead in front of us. And I'll give you an example. Ads Data Hub, [00:18:00] brilliant product does a great job. As long as I have kyla gmail.com or dustin@gmail.com doesn't do a great job when I throw your Acxiom email address in there. I think those are the struggles. I believe anti GRA providers need to step in and more and more, even if anti GRA providers aren't the ones solving for it or clean rooms the way to think about it. So that's another measurement. Clean rooms, not activation clean rooms, but measurement clean rooms is that way to think about it. So these are some of the ideas [00:18:30] that we're playing around with. I don't think we have a comprehensive answer, but objectively we would want to go there.

Dustin Raney (18:37):

Yeah. So you mentioned the C word clean rooms, and we did get a chance to talk a lot about that while in Palo Alto, and I think we both agree that there's a lot of promise, I think with the data clouds, the snowflakes, the Databricks of the world, of having places where [00:19:00] people can maybe share data safely without having to move their data. Is this an area where HP is really kind of investing?

Kumar Subramanyam (19:10):

Absolutely.

Dustin Raney (19:10):

And where do you see that going for you guys? I mean, are you seeing maybe even shift in budget to at some point activating through the context of a clean room?

Kumar Subramanyam (19:19):

Oh, fantastic. Austin, Dustin, anytime when I have a conversation around clean rooms with someone with all due respect with a Gen Z or even a millennial [00:19:30] or like, Hey, this stuff is new. No, it's not new. It's been around for ages, right? As Gen Xers, many of us know, and again, Acxiom has cleaned rooms, has had clean rooms forever, and I remember one of the first campaigns after I joined HP back in 2010 or whenever, was a joint effort that our store ran with Annex.

Kumar Subramanyam (19:55):

And

Kumar Subramanyam (19:55):

The way we ran that was through Acxiom. [00:20:00] Neither of us wanted to give the data to anybody else. We trusted you guys and voila, the campaign was one of the more successful campaigns, albeit it was more email driven. So well, the concept of clean rooms has essentially remained the same, but now with every company like I'm going to take New York Times or LA Times or Conde Nast, each one of them having their own CRM, how do you capitalize on that CRM without impacting the privacy side of things? [00:20:30] If Conde Nast has, what, 60 million? I don't know. I don't remember the number. If they have 65 million subscribers or whatever that number is, I can I get an opportunity to be able to talk to those people. Historically, we would've had to have some sort of a data share agreement, or NAST would've had to have a data sell option within their privacy policy.

(20:54):

Very troublesome. I mean, these are all struggles that companies go through. Now, let's bring a middleman in between, [00:21:00] whether it's Acxiom or somebody else. How can we mingle, commingle our data without exposing each other's private PIA compliant data to the other company, and how can we use it for the purposes of activation and measurement? These are the two problems that are the most pertinent to us in order to solve for this. We unfortunately, the industry today is siloed to a bit, and you see many different vendors coming [00:21:30] up with solutions, but not necessarily a cohesive cross-industry strategy.

(21:35):

I'll give you an example. We have a relationship with Habu, which is now a part of LiveRamp. We have a relationship with info Sum, which is part of LiveRamp. Now with all our data now sitting within Databricks. Databricks argument is, Hey, we can expose that data as a clean room. If you have your data in Snowflake, you could do exactly the same thing. So the options that we see is amazing, but I [00:22:00] think there needs to be some sort of a standardization in terms of methodology, policy, et cetera over time, and that's what is going to incrementally drive value with clean rooms. The way I think about this is, you and I had this conversation about DMPs and more specifically how easy it was to bring in a vendor like Acxiom or X Accelerate or LaMi by just checking a box and agreeing to A-C-P-M-V, [00:22:30] I would love to see CDP c DMPs or whatever those platforms are, essentially have the same capability.

(22:35):

You know what? This clean room info sum as an example, has a great relationship with the CTV vendor in the uk. I'm going to run my campaign through that vector, so to speak. How can I do that today? There is no such standard, and that's sort of where I feel some of the CDP or CDMP vendors need to start thinking about the cohesiveness [00:23:00] across today. It's a little siloed. I mean, you have IT service providers, like the live ramps separated out. You have clean rooms separated out, and you have companies essentially managing their own data. In other cases, that's the ultimate value.

Kyle Hollaway (23:17):

Yeah, I think that's an interesting call out, and I do think we're starting to see nibbles at that, right? I mean, certainly you mentioned Databricks and Snowflake, you've [00:23:30] got Delta Share, you've got Iceberg, you've got some standardization coming to at least the data taxonomy and the mechanism for sharing the data. Now there's still, I think the governance side to needs to be more largely standardized on that. It is a simpler option to be able to say where the data sits becomes kind of [00:24:00] nascent to the conversation. It's like it sits wherever. There's some standard mechanisms for reading that data, standard mechanisms for doing the match or the share and then how to take the results,

Kumar Subramanyam (24:13):

Correct.

Kyle Hollaway (24:14):

So yeah, I think it's a great observation.

Kumar Subramanyam (24:16):

Yeah, I mean, what if there was an option where I could go into this UI and again, visualizing this, like the way the DMP used to be, I see a set of clean room providers, and [00:24:30] so clean room provider once says, okay, we have data from Best Buy Clean Room Provider two says we have data from Amex. Third one says, we have data from let's say a CTV provider in uk depending on the campaign. What if I checked that box and it immediately did an overlap analysis with our CRM and say, you know what? Overlap is minimal. You are going to get the biggest bank for your buck by just checking this box. CPM [00:25:00] fee or whatever is the equivalent. Some sort of a currency is going to be so much you're going to spend $15 CPM or whatever that example is, and then voila. And these are the distribution or the activation platforms, what I want to push it to? Yeah, you could push it to Google Meta wherever or LinkedIn or whatever you desire. That is ultimate Nirvana in my mind.

(25:26):

You have that with the DP, but the problem with the DMP was nobody [00:25:30] was willing to literally give their cookie data away. And because policies were still evolving, GDPR was evolving. CCPA was evolving. It became a struggle for companies to essentially give us data. I remember dating back like seven years, we've tried to broker, we actually bought Adobe DMP knowing that Best Buy was on it. Microsoft was on it, and Intel was on it. We said, we saw this great opportunity to work with all of these three companies. Fast forward, [00:26:00] there was one campaign where Intel shared data and that was it.

Dustin Raney (26:05):

So Kumar, do you think part of that was also, I mean when everything was, the cookies were solving everything, fraud ad fraud of how did you know that your ad that you're serving is actually going to human and not maybe some made up site or there weren't, and is that part of something that you measure as well?

Kumar Subramanyam (26:30):

[00:26:30] Yeah, at that point, I think it was a huge concern for a lot of the companies that I spoke to. In fact, I don't know how many international trips I made five years back with my boss back then talking to every retailer in Germany, in France, in the uk, and everybody was willing to work with us. The problem at the end of the day turned out to be, it was the data engineering team. And the data science team was always philosophically open to doing it. [00:27:00] But given the veracity of who you are going after, not knowing who you are going after was one of the biggest tumbling blocks with privacy. And none of the companies, including our own privacy, was ready to jump on it. Right now, I see that conversation changing because a lot of clean rooms, unlike with the DMP in the past, now you have real data, real people data. You don't have 38 versions of me. And more [00:27:30] importantly, you are working on opted in data where you have a signature that ties back to say, Kumar gmail.com was in fact opted in and can be used in the context of a campaign, which I think is the biggest win now. And of course, I did talk about the evolution that's needed even within the clean room industry in order to bring this forward. But

Dustin Raney (27:58):

Yeah, Kumar, [00:28:00] I guess let's get a little bit provocative. Now. We talked about the ME toos, but being a change agent, what does that look like in the context of a company like hp? I mean, I think we had the conversation of maybe learning from some of the walled gardens how brands are starting to focus more on first party data, and really it's about offering up experiences that keep people engaged with your brand longer, right? [00:28:30] It's like if you think about the Facebook algorithm, what keeps you on that newsfeed so long? Well, it's learning, constantly learning from the time that you spend, do you feel like, I certainly see it, but HP in your brand that you guys can start incorporating some of those and are you already doing things like that to get people to, I don't know, spend more time with hp?

Kumar Subramanyam (29:00):

[00:29:00] I'm quite sure, right? I mean, there is, right? My role, unfortunately, we don't look at the creative, sorry, my team doesn't look as much at the creative side of it from an engagement et cetera perspective. But if I ever to answer this more generically for us, the way I would take a step back, and I would basically say for us, it's a journey for us. The context of how we can essentially decipher this is [00:29:30] it's a journey. I don't know whether we are there yet, but I'll give you one example which might help. When I go back to the days of, and I did touch upon this earlier maybe before the beginning of this conversation too, is we had the DMP, we were running our campaigns through the DMP building, our audiences, et cetera. And when G-D-P-R-C-C-P-A and all these things came in, the first question that we asked [00:30:00] as we went in-house and we said like, okay, hey, how many people do we have within our CRM 350 million? That's massive for us. And we didn't capitalize that. Why? Because we were taking our CRM and converting it to cookies through the live ramps and whatnot of the world. We were essentially, what do you call it, priming an ecosystem [00:30:30] that was faulty in the first place by doing that. And now overnight, Kumar Gmail had 38 different versions of him.

(30:39):

That's the historical problem. So what we had to do was first and foremost think through this problem. Then the second thing that we had to do is we had to work with all the big companies. Google was one of the biggest ones. Meta was another one, LinkedIn was another one. And then of course, trade Desk was another big component. We had to work very closely with each of these companies to understand their [00:31:00] evolving direction. Google, in fact, the ADS data hub wasn't even that big a deal like four years back, if I remember it. So that is now evolved into a massive product. So how does that Id stitching in the context of measurement help us? So that's the other thing. And the third thing was the platforms in between itself. Like CDPs, I've had conversations about CDPs with every major vendor, and where I struggled with was like, okay, it's a database for lack of a better way of putting it where I [00:31:30] can run more SQL queries of the wao.

(31:33):

Whereas what we wanted is a replacement for the DMP. How do I solve that? A replacement for the DMP would mean, hey, it's not just about creating audiences and customer 360 segments and whatnot, right? In courts, I need to be able to activate them, meaning I need to be able to push it to DV 360 or search 360 or YouTube or Meta or wherever. That's number one activation. The second one is I need to be able to enrich [00:32:00] the data and enrichment here meant, Hey, I have access to buying behavior of customers, but I don't know whether men like our products or women like our products or families with children like our products. What is that example? When we went from in the cookie world, we believed that we had these options available to us by checking a bunch of boxes, but when it came to real data, that became a problem.

(32:26):

So this is sort of where a partnership with a company like yourselves became [00:32:30] very critical. We had to rethink this process of bringing your enrichment into the ecosystem, and then as the guy who leads measurement, it would unfair for me not to talk about it. Can I measure this? I think that is the third vector that we looked at. So long story short, what we realized is we didn't need a CDP. We actually needed A-C-D-M-P, and that's what we rebranded it as where enrichment activation, onboarding, measurement, these are all core components of the platform, not just audience [00:33:00] segmentation.

Kyle Hollaway (33:02):

Well, I love that. And creating that more unified view and capability is really what is going to drive you forward. Because again, you're dealing with people based data, but it's distributed across a lot of walled gardens and such. And your ability to effectively integrate with each of [00:33:30] those and be able to get the insights back, the measurement component to that so that you understand the efficacy, the cost, the ROI associated with each of those including, and the challenging part is including overlap analysis. So am I double spending or am I actually getting accretive value by adding another connection? It's elegant, and it certainly, I think would prove to be very [00:34:00] effective, but it's also challenging, right? Because a entity of one, HP a big entity, but you're doing this and going about that process. So how does that get replicated out into the industry, I think is your point about standardization and building more of a meta layer that is effective that allows [00:34:30] people to take advantage of without having to recreate.

Kumar Subramanyam (34:33):

Exactly. Yeah. And also the benefit of being a big company helps in the scenario, right?

Kyle Hollaway (34:39):

Absolutely.

Kumar Subramanyam (34:41):

Well, not every vendor is ready to adopt what you breach, but this is sort of where some of the smaller companies stood out and they said, you know what? We're willing to work with you to essentially help build that. In fact, the two CDMP providers, or we call it CDMP, although they may not label [00:35:00] it that way, the two vendors that we worked with, ActionIQ and Am Purdy great companies, the big companies, while they were great at building a product, we realized, right, okay, the things that we want needs to happen now or the next six months. Now we're seeing that whatever we talked about three years back, that's built into ActionIQ and Amperity as being now becoming standard with the bigger companies as well, which is like I was at Salesforce Tower last week. [00:35:30] Yeah, last week I was with Adobe a couple of months back, and they're directionally going there, which is fantastic. But the key for us was we had to be ahead of the game. We had to be three years ahead of the game, and that meant partnering with maybe smaller providers, building a strategy around evolution. And in fact, it's actually benefited companies like ActionIQ in my mind, where now their product is, I believe they call it a customer hub, as opposed to calling it CDPI may be wrong, but that [00:36:00] requires influence, driving influence a vision that is collectively agreed upon between the company and hp. And more importantly, for someone to put energy into building that as part of their product, they need to hear something similar across other people like-minded people. And I think that was important.

Dustin Raney (36:21):

Yeah, I think one of maybe the key terms there is composability, right? It's like Theon IQs of the world. They kind of went into market saying, Hey, [00:36:30] we're not going to give you or make or forced you to buy some monolithic, massive platform.

Kumar Subramanyam (36:34):

Correct?

Dustin Raney (36:34):

It's like you already have technology, correct. How can you take some of the components that we have and put it into play speed the market like today? So that's brilliant, and it's definitely an area that Kyle and I have kind of heard across the board as something super important across every industry right now,

Kumar Subramanyam (36:54):

Thousand percent. And that's the biggest evolution in my mind in this industry, is not around CDPs [00:37:00] or measurement or any of these. The biggest evolution has been a piece of tech that should have been around 25, 30 years. Back when I was in grad school, one of the very first courses that I took was distributed systems,

(37:18):

How data needs to live, where it needs to live, and you don't sit and copy stuff. Well, it took, what, 30 years I'm assuming, for stuff like that to happen. [00:37:30] And you now have the source snowflakes and the Databricks and whatnot, the zero copy rule. And to your point, Dustin, this whole argument of why move data around when all that you need is a platform like ActionIQ or Emper, whoever it is, sitting on top of that and whether through Delta share or Compulsivity, whatever we call it, being able to use those methodologies to be able to build your segments and whatnot. I think that's the biggest shift. It helps us. And this [00:38:00] shift is not just with your CDPs or CD mps, but also with measurement platforms. We see this, see this with as being one of the key drivers of second party clean rooms. This is amazing. But it required time, in my opinion, in this industry for it to be where it's today.

Dustin Raney (38:20):

Yeah, that's a great point. I know we're running low on time now, but there is a question I did want to ask you. Go for it. Just from a, [00:38:30] customers are changing, people are aging, and you have generations coming in that are consuming content and you're having to reach them through other channels. How are you guys overcoming some of those challenges with the shift in interest, the shift in behaviors that are coming in with Gen Z, gen A?

Kumar Subramanyam (38:52):

Oh, it's a brilliant question. I'm glad that you did ask this, right? And the reason why I say this is while I [00:39:00] credit our CRM, which has got three 50 million people in it, our realization a few years back, I don't remember the exact dates, and again, I'm really bad with the dates, so my apologies. I'm going to say maybe 2019, we did an assessment of our CRM. The average age of a US customer was 54 years old.

(39:29):

Now, that's [00:39:30] fine. It's great to at least know the average age of your customer, but then guess what? In 15 years, with all due respect, I am marketing to someone who may care less about our products. So it was absolutely critical for us to rethink our strategy. Fortunately, marketing decided to rethink our positioning from a messaging perspective as well as in terms of where media dollars were being invested. [00:40:00] The shift that we did was to go put about 70 to 80% of our money on Gen Zs and millennials. Fast forward today, or rather end of 23, we did a reassessment in the two prior years that we push this newer thinking, our cohort, our average age came down, but more importantly, we increased our millennial and Gen Z pool by [00:40:30] plus 10%. Why is that important to us? Lifetime value, lifetime value for a Gen Z millennial is about 2,500 bucks for us. Average minimum for someone who's a Gen Xer or about 50 is about $1,500. So right there is a massive, massive value. And I'm talking about averages.

(40:50):

And this was yet another area where this was one of the conversations, Justin, that you and I had many, many, many years back, is like on this subject [00:41:00] where we said, okay, hey, where is Acxiom adding value? Let's rethink about this. That's sort of where the value of the data portrait analysis kind of came in massive value that allowed us to go back and look, oh, yeah, we were 54 was our average. Fast forward, now we're available certain than that, but we increased the Gen Z and millennial cohorts by plus 10% in the last two years, billion. Why? Because we now have a portrait of who our customers are. We need to keep doing this [00:41:30] every year. If our end objective is to continue sustaining the lifetime value measures that we care about, and more importantly, market to the Gen Z millennials.

Kyle Hollaway (41:41):

I love that. And just as you said, it all comes back to data where you started the conversation and just the almost maniacal focus to say, what's the important element here that we're looking for to drive the outcome that [00:42:00] we ultimately want? And I think that is a huge challenge for our listeners that I think they can take away from this, is to really assess their strategies and not just say like, oh, data-driven and being, it's easy for that data tilt to wag the dog if you're not really being disciplined about finding what that piece of data is that you're wanting to track that has a material impact. And that was a great example there where [00:42:30] you guys have honed in on the generational aspect of your consumers, what their lifetime value is, and then being able to monitor your mix of that over time and then layer in your strategies to help move that. Awesome. Okay. That's awesome. Great job.

Kumar Subramanyam (42:46):

Thank you.

Kyle Hollaway (42:48):

Well, Dustin, we're about done. You want to throw the last question out?

Dustin Raney (42:52):

Yes. So Kumar, we always like to have a standard wrap up question this year, standard wrap up question is [00:43:00] a little less dystopic than previous years, but it's fun nonetheless. If you fed all the data about Kumar into ai, what are the three words it would produce to describe you,

Kumar Subramanyam (43:14):

To describe me? That's a tough one. Are you talking about my bubble profile?

Kumar Subramanyam (43:20):

Yeah, just kidding.

Kumar Subramanyam (43:23):

The three words are in my mind, we believe we're trendsetters or we are leaders [00:43:30] in the industry of thinking about the value of data, not just repeating the things that we have historically done. That's one thing that comes to my mind. We've always, and I'm not just talking about myself, my peer Freddy, who heads media. We've been great partners and we have constantly figured out how to push the envelope. So that goes without saying. The second thing about me is curiosity, figuring out what else is there. I mean, [00:44:00] we didn't even know about DPA or data portrait analysis until we started asking all the stupid questions that we had to, evidently, they were not stupid beyond a point, and we saw value. It's being able to ask the questions, what is important for the organization? And then being able to dovetail that back to where your data is.

(44:19):

And the third one is, in my opinion, which I think has helped us, and this is sort of one of the big wins that has personally been a game [00:44:30] changer, is I go back to our spend, our technology spent a few years back, majority of the money that was being spent was on ETL and building products that required us to move data around that. Fortunately there for me, and again, this is sort of, I dunno whether I'd say composability or whatever the example is, but really the core value is the federation is one of the [00:45:00] key areas where I'm a big, massive proponent. And again, if there's one thing that I would say people need to think about, federation has to be a baseline now because you have technology that enables you to do this. Five years back, you couldn't do this.

Dustin Raney (45:14):

Love it. And the awesome thing is that I would say those are just having met you, Kumar, our qualities about you that become qualities about hp. Thank you. Brands are a culmination of its people, and [00:45:30] they are very, very not lucky. But I don't know what this, it's great to have somebody like you in a position like that, driving that innovation, driving things that matter. Thank you very much again for joining us today on the podcast. This has been just an amazing conversation. I know our listeners are going to get a lot of value out of this, a lot of real world experience talked about here. [00:46:00] I know Kyle and I did, so we definitely want to have you back. It's always fun.

Kumar Subramanyam (46:04):

Happy to. Thank

Dustin Raney (46:05):

You.

Kumar Subramanyam (46:06):

Happy to. Thank you, gentlemen. As long as we're not talking about the Razorbacks, right? I'm happy to come back.

Dustin Raney (46:12):

We might make you call the hogs next time.

Kumar Subramanyam (46:14):

Yeah. I'll talk about Nova Pride.

Announcer (46:18):

Thanks for listening to Real talk about real marketing and Acxiom podcast. Find all our podcasts at Acxiom.com/real talk or your favorite podcast platform. Until next time.