Balancing Data: The Art of Offensive and Defensive Strategies

Chris: [00:00:00] All right. Welcome to another data driven podcast. My name is Chris Detzel, and today we have a special guest on the podcast. It's Willem . Koenders. Willem,

Willem: how are you? Hey, Chris. Thanks for having me. I'm doing great. How about yourself?

Chris: Doing well, man. We were just talking, I always think of the pre show, and you're telling me this is your first podcast to be on.

Chris: I was shocked with all the stuff you wrote and everything else, and so really appreciate you coming on today.

Willem: This is very much the inaugural event. It's a pleasure to be here. Let's

Chris: see how it goes. That's going to go great, I have no doubt. Tell us a little bit about yourself, who you are, what you do, and the kind of people that you work with on a daily basis.
Chris: Of course,

Willem: sure. My name is Willem Koenders. I'm originally Dutch, so speak with potentially I did a little touch of an accent here, but I spent my entire working life working with data. So right now at CS, I'm a leader of the advisory service around data management, [00:01:00] data governance.

Willem: I can lead their practice globally. And before that, I spent a decade with Deloitte. Maybe interesting to share as well that did two years of that in Europe. About six years of that in the U S if you add it all together, I led a practice and helped found it co founded in Latin America for two years.

Willem: And as right now I'm in Morocco actually. So a lot of these, experiences have been scattered all over the world, but yeah, for the most part in the last couple of years, I've worked with a really extraordinary. String of chief data officers working on kind of the biggest topics across financial services, banking, insurance, retail, CPG technology a little bit now, medical technology.

Willem: Yeah, it's been it's been a fantastic run. Yeah. That's great. And,

Chris: and, you've had such a vast background. The background, if you've dealt with a lot of customers or a lot of. People, CDOs, like you said, and so I'm really excited about today's topic around balancing offense and defensive strategies and what that really means.

Chris: You [00:02:00] mentioned that as we were talking back and forth, and can you talk a little bit about, what you meant, and how do you define offensive and defensive data strategies?

Willem: Yeah, for sure. Here's at least my interpretation. And so when I think about offensive and defensive data strategy, it's almost like maybe let's take both of them and understand what they can be.

Willem: So if you have an offensive data strategy, you are focused on enabling positive outcomes, right? So you want to leverage data to achieve some sort of positive outcome, like growth or revenues or profitability, drive innovation. And. Drive competitive advantage, and it tends to have these tend to live on the business side of an enterprise, and they tend to be heavily focused on AI and analytics use cases, right?

Willem: That's those are typically more offensively oriented strategy. If you have a defensive data strategy, it's More about preventing negative outcomes. [00:03:00] So again, it's not necessarily enabling positive one, but preventing negative one. So think about data security, privacy, compliance, risk management, all those kinds of risks that need mitigation.

Willem: They tend to sit on the legal or the accounting on a regulatory sides of an organization. And they really, and data strategies that are heavy kind of on the defensive side, they tend to emphasize, compliance, governance, and security capabilities. So that's how you can characterize those, how I correct, characterize these two types.

Willem: Is there

Chris: any like specific examples of a, an offensive strategy and a defensive strategy, like a real life thing that you can think of?

Willem: Yeah, I've worked for for a lot of years within the banking industry, for example, where they tended to be extremely. Strong on a defensive side, right?

Willem: They have very strong regulatory, regular regulations in terms of for example, their liquidity reporting. And if they, if you don't pass those tests, so if you don't pass a stress [00:04:00] test, if you don't pass those basal tests and you can't evidence what data you are using for these tests, where it's coming from, it's provenance and they can turn off your lights, literally they can shut down your organization.

Willem: So like that, they are extremely focused on the. We're at least on the defensive side. Right now that's a little bit in the past they start to open up more on the offensive side as well. Offensive, think about a company like Netflix, just comes to mind randomly for me if you think about what these guys do, like, how data and AI is at the heart and core of their business model, like, how they track what you watch, how it may impact the experience that you're having, how it can help them recommend the next the next recommendations there, their team, and then just do, they're doing this on the go.

Willem: Does not like they did it like five years and that's a modernist scene, like they learn every day. And so that's all about that customer experience, like leveraging and the insights that they have. So there's just two kinds of examples. No, that's good.

Chris: And why don't you think it matters to recognize these two strategies?

Willem: Mostly [00:05:00] like anything, but like strategies about decisions, like always, it always comes back to making decisions as well. In this case, too, you can't do it all right. So sometimes there's a trade off. And sometimes a trade off is really about, I have a million dollars to spend, or I can onboard three people in my new CDO team am I going to have them, am I going to look for someone who can really reinforce the compliance with a certain regulation, or I'm going to read, onboard someone who's Really great and generative AI and helps you figure out how it can bring this to the different business lines, right?

Willem: So some of it is really just a trade off but I think that is that there's a little bit more also to just having a balanced data management Strategy in general so that you want to be to want to recognize that both of these sets of Objectives and benefits exist and you want to make sure that you have a right balance like you don't want to necessarily 100 percent go one way or without kind of losing sight of the other one, like you want to enable, not completely, shut down the innovative [00:06:00] innovation ability that you have, but you also don't want to like unnecessarily open yourself up to risk.

Willem: So those are some of those considerations of why it really helps to figure out which of these matters the most you know, to this, to, to your enterprise. Do you think companies have,

Chris: Both? Can they go both offensive and defensive? Or do you basically see companies go one or the other?

Willem: Not in the extreme so you can't, you cannot be completely offensive or completely defensive, it just doesn't work because if, at the end of the day, there are, if you push it too far, you push it far enough, there will be trade offs because offensive is all about getting the data to people as quickly as you can, whenever they want it, in the format they require it, defensive is way more about locking it down it just oversimplify it a little bit.

Willem: But like I as you may have seen like I that is if you push it very far right but I do believe very strongly so that there are absolutely things that you can do on a foundational enablement perspective to actually do both like some [00:07:00] things like having a. Base basic foundational capability to manage your metadata or having a catalog, for example, like doing, having some, minimum level of data quality.

Willem: That is necessary for both, like you need to have that anyway. If you want to enable like a customer segmentation or these new recommendations, like you need to know what data you have, where it sits, the quality, et cetera. In the exact same type of metadata, you would require to be able to identify How do I classify it?

Willem: How do I protect it? So there are bits where you can absolutely make decisions to enable both. If you push it a little bit further, then some trade offs will appear. Yeah, it makes sense. How

Chris: does how does data governance and compliance fit into these strategies?

Willem: How does data governance and compliance fit into the strategies?

Willem: Governance first. So governance to me is a foundational one. So data governance is almost a, it is actually interesting. Very often, like when I come to a new client, I also first want to [00:08:00] establish what do you interpret? What do you mean with data governance? How is it different from data management?

Willem: But in any case, data governance for me is really all about How do we ensure that data management, which is broader, how you actually do it? What are the roles, responsibilities, the policies? Like, how do you go about making sure that the data is appropriately managed? That's a foundational capability.

Willem: You need that for both. You cannot have a very effective offensive data strategy if you haven't figured out some sort of, foundational things around things like data ownership or some key data assets and things like that in the exact same way if you look at it from a defensive perspective.

Willem: You need to have Those those kind of foundational capabilities compliance is pretty, pretty completely on the defensive end, right? That is a defensive objective and a lot of capabilities underneath that are would sit on the defensive side.

Chris: Not to get off too much, but data governance is to go back to the data governance pieces, it's a pretty hard thing to do [00:09:00] when you go in and look at a organization.

Chris: Do you ever think, oh my goodness, , this is gonna be rough to get, a governance piece set up. How do you go in there and just say, you know what, you have to set this up and this is how you do it, and that kind

Willem: of stuff. Yeah, there's, depending on where you go, it could be a relatively young company that is more struggling with, they want to be able to scale and not get caught in kind of a breach because they grow too fast versus these, huge behemoth organizations that have stacks all over the world.

Willem: But, a couple of challenges that I pretty much always see, there's still a pretty negative connotation to data governance. That's almost the first thing that always, almost always appears. Like we tried this. It worked, this person was here, they left, like there's always that negative connotation to kind of battle, it's like a cost to the organization, if that makes sense.

Willem: And I've seen some places where they just try to rename it Data Enablement, but do the same thing. Which, I'm not necessarily a big fan of that, that kind of happens. [00:10:00] But like the, I guess the good news is you just got to be able to tell the story in general. So not, when we come in a new place, but that's the first thing very often you start talking to, the CEO, whoever it is, the head of data governance is like you want to change that story like data governance is not.

Willem: It's not a cost to the organization in that sense. It can never drive value in and off itself. It's always tied to stuff that drives impact one way or another. If you don't know how it's driving value, then go back to the drawing board, right? You may want to reorient yourself. That's like the one thing that I would say is it's I guess you have some tricks eventually into stories to tell, but that's that is definitely one to We run into very often.

Willem: Yeah, that was good. And,

Chris: How do you measure the success of a integrated data strategy? Is there any tips, tricks there?

Willem: That's another tough one, right? Very, if you look out there, there's this famous. [00:11:00] Statistic that is quoted that says CDOs don't last or last on average less than three years.

Willem: I think it's 2. 4 or something that's right. Very important part of the explanation there is that they struggle to articulate how they're Creating value. And so that all of that has to do with what you just said, measuring a data strategy or an integrated data strategy.

Willem: And so a couple of things that I believe I've seen that work really well is first of all, the realization like a data strategies in a sense, not separate from a business or organizational strategy. It's not right. It's not you're doing something else. You're not a different kind of company. Like it's really the interpretation of what can you do?

Willem: To drive value somewhere else. That is sometimes this, there, there are some data leaders out there who believe they are the actual business leaders in some ways, and for the most part, you're not like you're helping, marketing guys or you're helping clients guys like that's what you're effectively trying to enable.

Willem: Like their business process need to work. And so with that in mind, [00:12:00] there's just this sure you're aware this kind of re vamp or reintroduction of this concept of. Data assets and products which I think is very helpful in this area because before you had CDOs or data teams, or they were, they had a annual budget of like several million dollars.

Willem: And then at the end of the year, they would be asked like, what did you do? What'd you, and they said I've got 2000 data quality rules, completeness is 97 percent and like things like that. I cataloged 10, 000 fields, but I, okay, what does that mean? Nobody knows what that is.

Willem: Yeah, exactly. Is it which are to a data governance professional? Like those can be very impressive, things to to have done, like what you can do with an asset and a product, where you basically focus on. Actual kind of logical groupings of data that have inherent value can be customer data can be transactional data can be a mix and by definition they have several use cases.

Willem: Attached to them [00:13:00] with quantifiable business impact statements. And now as a CEO or a data leader, all of a sudden you, you don't say, I have this many data quality rules or this many fields catalog. Now you say my team is keeping 11 data. I'm making up the numbers, but I'm keeping up. 11 data assets up and running who are supporting 37 use cases, which are supporting half a billion dollars in revenue and about 25 million in risk mitigation.

Willem: And so that's a better story for sure. Yeah. And of course the thing to matter, like it doesn't really matter is half a million or 700 million or 300, because when you get to that point, people are like, Oh, okay. We know that ROI is so high. That's the, but in any case, that is what I've seen work like anything else.

Chris: I agree. I think at the end of the day, like it's all about the business outcomes and what the data is doing, like the, if I'm telling people, some of the things that I'm doing, I give you all these metrics and it's not that metrics aren't important. [00:14:00] Obviously they are, when you talk to a leader, a business leader.

Chris: They don't care. It doesn't mean anything to them. So you gotta, it's just an, anything that you do with data is what is the meaning and what's that story that you're trying to tell, you gotta tell, you gotta understand some of that. So what are some of the pitfalls or Colin pitfall falls one measure trying to measure the success.

Chris: You mentioned a couple of them, but wondered if there's a few

Willem: more. Yeah, common pitfalls in terms of measuring success or maybe slightly broader in this kind of concept of offensive versus defensive. Like I, the first one thing that comes to mind is like that you completely overemphasize one side, right?

Willem: Where, which happens, it tends to happen. Like it's someone who tends to be very strong in either of these two, who then grows into position to Own the strategy. I knew things that's the key to to unlock unlock the value. And so you completely focus on kind of the AI or analytics and the offensive side or completely on.

Willem: Making sure that you just declined. I was at earlier [00:15:00] this week on data sharing where they basically said, because they had the chief legal chief compliance, et cetera, officers all the room. And they said, we're locking it down. None of this gets shared ever again until all these rules are in place.

Willem: And then until not realizing that that means your organization grinds to a halt. It literally doesn't work. You can't. Do that. And so in any case, the overemphasis on either side is some sort of a pitfall kind of oversimplifying it for sure. I already mentioned the minimum foundational capabilities.

Willem: I go the same time. I'm almost like, how do you say that? I'm like, I'm invalidating what I just said myself, but in the sense that there are foundational things that always make sense. Don't like. Don't try to get the business case super detail done for like the first initial foundational things around like metadata or catalog You won't get it like you won't get it in year one ever It's always a little bit further down and you got to believe in it You got to tell that story and go with it like that's once you have that and once you go As I mentioned from one or two assets in year one to like now you're supporting [00:16:00] ten or fifty That ROI will be there, but there are some foundational things that always literally always make sense if you right now are a company out there that in any meaningful way is data driven and you have not paid attention to a minimum amount of metadata management.

Willem: You're, you're lost. That's going to come back to bite you. And then actually in line with that, the last thing is a cliche, but super true, right? Which is don't start too big. Don't do it everywhere. Don't do it all domains or all data. Everywhere all at once. And just you'll pick one is pick one area, one asset or one subdomain, I would say, but but not just that, do it like smart, like something about that there's value, right?

Willem: Figure out if there's value before you even get into all the nitty gritty, untangling data lineage or whatever it ends up being that you, you end up doing. And then that business person, like your very first one that you would actually work with, don't pick the most skeptical one.

Willem: I pick someone that's right Yeah, no, it's really true man. [00:17:00] If I agree The first one you want to have somewhat a friendly face who's on kind of a business or similar side who does He needs help here and who can help you tell that story so that you can tackle the somewhat harder ones a little bit further like that to me is maybe some commonsensical, but like things like if you don't do them, they become become pitfalls.

Willem: It's weird. You

Chris: say it's common sense and it sounds like common sense, but I remember I used to work at Forrester Research and one of our, one of the analysts at the time, his name's Jeff Scott. And the one thing that he said, he was doing enterprise architecture kind of stuff, and you can imagine, whenever they're doing big projects, he said, look.

Chris: Go to the people that are for you and that will help evangelize. What happens over time is you'll get three or four kind of business cases, use cases, and that person that was skeptical now has to listen to what you're doing because you've created so much success in these other three or four areas that it's just going to be helpful.

Chris: Sounds like common sense, but it's not always common sense. You just. [00:18:00] Think you need to get this one thing going and they're not agreeing with what you're

Willem: saying, you know There's something buried in what you just said even like what the anecdote you shared which is about like celebration of success and socialization like that there's like this data governance negative connotation you really have to Be deliberate about calling out successes, send that email, have the council, call it out, put those people out there and say, this worked.

Willem: I had just also last couple of weeks, a global technology company very big, very very advanced in a lot of different ways. And there was two kinds of areas that were actually very similar and in a lot of different ways, similar tech stack, similar seem to be like maturity in a lot of ways.

Willem: And one was. Very well able to get those products up and running at least seemingly and the other one was stuck somehow. They just couldn't get it off the ground running. And just upon looking at it a little bit closer, like really the key difference was just this whole social. [00:19:00] wrapper around everything.

Willem: Like, why are we doing this? There was a weekly note that somebody just took 30 minutes to know here's the stuff we did great stuff, like here's the people who made it happen. And it seemed a little bit silly and cheesy, but it's critical really is.

Chris: I love that. I had a, I had somebody tell me, Chris, what you should do.

Chris: And I'm not saying I do this every day, but it's along the lines that you said is every week or two, write down all the things that you've accomplished that you did. And that, that, that happened in the business from what kind of you, you did and send that out to certain people, it could be your boss or it could be a business.

Chris: person or whoever you're trying to and just be consistent about it because what they do is a lot of times they'll just read over it and be oh that's interesting and then you know gives them opportunity and it helps you to know what you've accomplished throughout the year you know when you go and do your own review i think that's great i love that idea So is there like a link between offensive and [00:20:00] defensive strategies like data and data unification 360 views and MDM?

Chris: Is there a linkage between all of that?

Willem: Absolutely. Yeah, 100%. So there. I would relate it back probably to the, what I'd mentioned around data assets or products or what you would call them, like the, these 360 views or 720 views or these MDM they are almost the best examples one can think of, if you have this.

Willem: Data product or assets. And you do it. It unlocks. It unlocks the objectives on both sides and probably the best examples out there. That I could come up with and just to maybe impact that a little bit more. Yeah. From a defensive side, if you look at what you just mentioned, like a 360 view, say for customer, right?

Willem: And it may actually be enabled by an MDM, but let's say that there's just an asset or two in here around customer data, like no matter what kind of industry, but imagine it's something about like names and phone numbers and [00:21:00] ages and jobs and who knows contact preferences. Maybe there's some customer data that's becoming a bit more sensitive.

Willem: Out there today, but so for you to be able to truly depend on that, like for you to say Netflix or say like these truly where you let the model make the actual decisions, like it's not even just an analytics report, but you have actual decisions that are happening based on the insights that are built out of these views.

Willem: What would you need for that? To be able to be done, which is trust. You need to trust the data that it's correct, that it is timely, that you can understand it, that there's no black box here and how it got there. And so that trust is enabled by the defensive. Kind of nuances of these products or assets around insisting on metadata, insisting on the classification, the definition, like the lineage, where did it came from, maybe business rules around it.

Willem: And so [00:22:00] this is in, in, in a little piece I'd written somewhere, I called this literal, this is quite literally. Defensive off enabling offense, right? There's things that you do to make sure that you truly manage it So that afterwards the people from it with an offensive mindset, they don't even need to think they trust it and so they if I so if there's a link between offensive defensive strategies and things like 360 views and mdm like Absolutely, because they will, if you do it right, can help you enable both of these things at the same at the same time.

Willem: Love it.

Chris: What are some predictions, and this is our last question most likely but what are some predictions for the future around offensive and defensive data strategies?

Willem: Predictions or hope. What do you want? But maybe that. Hey,

Chris: what do you think? Hope? I like the hope. But if there are specific predictions, that's helpful.

Willem: So let me just at least hopefully these will coincide or they converge, which is like the data [00:23:00] governance of the past that has a bad connotation is the one, where you had armies of Analysts or consultants often that came in helped dig up like the system that somebody else put in there and Tried to figure out like where the data is coming from and retroactively measure data quality It's a data governance that has a bad reputation.

Willem: It also has a very relatively low ROI and investment like I predict and hope that The data governance of the future is one of data governance by design, right? So that is where if you have a new business process, a new platform, a new, whatever it is, like that people are going to become smarter to integrate these things when they are created.

Willem: So that It's really only a very small tax, once the engineers are there and they know exactly why they're creating systems, why they're taking them from one source and not another, if you help,[00:24:00] help fix the transformational life cycle, then I think that the strategy will go a lot more to doing these types of things by design, that, that's one thing, because those big Mammoth projects to solve kind of the spaghetti mess of the past.

Willem: I think it's something that nobody wants to own anymore. Yeah. And the second bit is very much open to me is all if you think about this whole spectrum of offensive data strategy, defensive strategies, and specifically the capabilities that kind of fall in there and maybe enable part of both of these is the.

Willem: Thank you. Smarter, more intelligent and AI driven approaches to do this, right? So how can you do some of these things with? The technology that is out there and more automated, better, higher quality. It's how can you do that better? And there's, you can do probably a whole other episode on literally unpacking those, but that's a massive.

Willem: Massive one, right? And so it's a little bit early to tell what's really [00:25:00] going to happen, but there's a ton out there. And some of these things are maybe just process discipline. Like it's, it is kind of automation. Like you just make sure you build something new. It just automatically is readable by your, your spiders, your data catalog.

Willem: But there's also things like how you can maybe use some of the LLM or Gen AI. I don't know. Capabilities to interpret, the data that sits in in certain systems and specifically also the unstructured data, right? The data governance was virtually completely focused on structured data because that's the only thing we could do, right?

Willem: Maybe not anymore, right? That's maybe That's a huge opportunity is

Chris: the unstructured data. Yeah. One of my one of our customers talks about that all the time. He thinks that AI is going to be a huge opportunity to help with some of that.

Willem: So that's exciting. Yeah. One, if I can just squeeze one, one quick, super specific example in here is like one of those.

Willem: Clients we work with right now, like they're a technology company. So they have tens of thousands of SKUs that they build and they have R and D, related to it. But the actual [00:26:00] truth, the actual kind of source of that data are filed documents because they have to file them in different places for compliance and patents and whatnot.

Willem: So the actual source of it is actual texts. They're actually in PDF that won't change for a while. And so all of that is eyeballed now, right? So if you go back to. prOduct IDs to descriptions to like dimensions, colors, everything about it, like they have to go back to these documents, like that is something that is not even not.

Willem: Potentially complicated to train. Like, how would you be able to govern that data a lot better with some of these new technologies that are out there? It's a good

Chris: point. I worked at a company that electrician company, and they had all these skews, big 10, 15 billion company, they had all these skews, and that's exactly what.

Chris: They were doing to, push that out on these product skews out on the the e commerce website, they sold to B2B. And so everybody was [00:27:00] eyeballing, copying and pasting descriptions and things like that, and it just, I was like, they literally had a team of four or five people doing this and that was an all day job.

Chris: And they couldn't get done. They still can do, I think that's where AI and stuff will. Also play a huge role in some of that stuff. So Willem, this has been great. Really appreciate you coming on to the data driven podcast. So thanks everybody for listening in to another data driven podcast. I'm Chris Detzel, please rate and review us and Willem, thanks so much for coming on to today's show.

Chris: Thank you, Chris. Take care.