Mastering the Data Organization Matrix (DOM): The Insights Supply Chain Framework - Part 3

[00:00:00] Michael Meyers: Organizations today face a fundamental shift. Being data-driven has moved from a competitive differentiator to basic survival requirement, yet most organizations struggle to convert raw data into actionable insights with fragmented collaboration and ambiguous ownership, creating persistent roadblocks. That's why Dr. Duru Ahanotu, the leader of Tag1's data strategy team, has created the Insight Supply Chain Framework. It brings structure to how you organize your data professionals and teams to leverage your data and enable insights to flow throughout your organization. This is the third episode in our series on the framework. In our first episode, we provided a general overview, and then the second we explored the critical decision between centralized and decentralized data teams. Today we're diving into the Data Organization Matrix or DOM, and the strategic question as to whether or not data is or should be a core competency of your organization and how that's gonna impact the centralization versus decentralization decision. In future episodes, we're gonna talk about how AI has been impacting Duru's thinking, and do a deep dive on the Data Maturity Curve and how your organization fits into it today, and how you progress along it over time. Thanks for joining us. Let's get started.

[00:01:23] Michael Meyers: Hello and welcome to Tag1 Team Talks, the Tag1 Consulting podcast. I'm Michael Meyers, the managing director at Tag1, and today I'm joined by Dr. Duru Ahanotu, the creator of the Insight Supply Chain Framework, and the leader of Tag1's Data Strategy Team. Duru, welcome back to the show.

[00:01:37] Dr. Duru Ahanotu: Thank you, Michael. It's good to be here.

[00:01:40] Michael Meyers: With three degrees from Stanford, including a PhD in management, science and engineering, and an amazing career that ranges from building expert systems to diagnose manufacturing robots to leading data science teams at Yahoo and Dictionary.com during their heydays, to helping our clients build and scale data ecosystems, Duru brings not only tremendous academic credentials, but real world experience to help address and help you build these data ecosystems. And today we're gonna highlight some of those real world experiences and provide you with some examples as to how he's applied this Framework across his career. Before we jump into the details on the supply chain, I wanna give you a little bit of insight into Tag1.

[00:02:23] Michael Meyers: Tag1 is the number two all time contributor to Drupal, which is the world's second most popular content management system. For nearly 20 years now, we've been the architects of the open web, leading the collaboration of the software and best practices that power millions of websites and hundreds of thousands of organizations worldwide. We're a full service strategic partner, applying that same architectural expertise across technologies and throughout your organization. From discovery and design to building and scaling complex applications, we lead AI strategy and implementation, design and manage infrastructure and architect large scale web applications across a wide range of platforms. We're trusted by industry leaders, including Google, Sumitomo, NTT Data and the European Patent Office to solve mission critical challenges and build lasting solutions. Check out Tag1.com to learn more about how we can help you.

[00:03:28] Michael Meyers: So, Duru, just to catch people up on episode one and two, where we talked about the Framework as a whole, just like a high level overview and then got more into the centralization versus decentralization, can you just give us a quick summary, again, what is the challenge that organizations are facing and how does your Framework address those challenges? And, you know, just really high level, why does it matter, that decentralization versus centralization?

[00:03:55] Dr. Duru Ahanotu: Yeah, so at a very high level, what I set out to solve or help solve is the challenge of how to best organize data professionals, both along their career paths and how to organize them as teams and groups in a way that will facilitate the best mode of collaboration amongst themselves and collaboration with the rest of the company in the service of delivering value through products and services to a marketplace. And what I saw in my career is that there often wasn't a lot of structured thinking about how to do this. And so the Insight Supply Chain came about from my own career experience in what I thought was the best framework for doing all of these things in a very effective way and in an efficient way that would allow people, data professionals in particular, to best enjoy their work and also have strong and robust career paths. And so this came out of a lot of my experiences and it's been constantly updated because this field changes a lot. And so I've learned also to make this robust through flexibility. And so it's been timeless in that sense and I've been able to apply it in my own work and also in my theoretical outlook on what's to come in the future with AI.

[00:05:14] Michael Meyers: And what you've seen, you know, consistently and clearly is that this Framework has a big impact, right? Like making decisions arbitrarily, not thinking them through using a framework, you know, taking brute force approach, none of these models come close to enabling you to be anywhere as effective in taking the data and turning it into decisions as what this framework enables you to do.

[00:05:32] Dr. Duru Ahanotu: Right. Exactly. That's exactly right. Yeah, I mean, you've said it all right there. I don't have any much to add on that one.

[00:05:40] Michael Meyers: You know, today we're talking about the DOM, the Data Org Matrix. In my background, you know, as an engineer, the Document Object Model is what I think of, you know, when I think of DOM.

[00:05:50] Dr. Duru Ahanotu: Yeah, I thought about that. Sorry, quick sidebar. I thought about that because I first learned about the DOM way back when working with an internet company, and I said, ah, well, I'm just gonna claim it as my own domain specific acronym.

[00:06:02] Michael Meyers: No complaints. But, you know, I'm curious, I hadn't, until I had started digging into this framework with you and learning more about it and using it with our organizations, what is the Data Org Matrix? I don't think a lot of people are familiar with that term, and how does it play into this?

[00:06:17] Dr. Duru Ahanotu: Of course. So the running joke from business schools and folks in consulting is that to explain any concept you need a two by two matrix. And in working with actually a friend of mine a while back, in terms of figuring out how to best communicate these concepts, the two by two matrix just served the role. So that's where the M comes from, from the matrix. Then the data org part, of course, is I wanted to speak to organizational principles. The Insight Supply Chain started as a way of just describing a flow of data through an organization from raw data to the insights that again deliver value through products and services. But what was missing was, well, what are the implications for strategy in a company? What are the implications for organization? And so the Data Org Matrix came about to serve all of those roles, to put it all in a neat summary, two by two summary, using the, again, the decision of decentralized versus centralized as the core narrative, and then looking at how those two simple pivot points drive a very rich set of outcomes in your organization. So that's what my goal and intention was in building this matrix. And it just serves as a nice meme, in a sense, in terms of communicating what this is all about.

[00:07:20] Michael Meyers: It's matrix inception, because you have the original two by two, which we've talked a lot about. We're digging a level deeper here. The initial matrix, the fundamental thing is whether or not you are decentralized versus centralized, and whether or not you have specialized or generalized resources that you're working with across your organization. We're going a level deeper here. Walk me through this matrix. How does each section apply and how do I leverage this?

[00:07:46] Dr. Duru Ahanotu: Right. So with the two by two, you have the axes of decentralized and centralized at the bottom. Sorry, that's the X axis. So it's centralized on the left, decentralized on the right. Then on the vertical axis, you have generalized versus specialized. And so that's the skillset. And so this is where the interaction, intersection is: how are you going to hire or staff your people? Is it gonna be a broad, you know, people are gonna have broad skill sets or very specialized? And then again, how are you going to organize them? Are you gonna organize them in one place? Are you gonna distribute them across the organization? So now if we want to see how this plays out, we can pick any part of this matrix. So let's take one of the simpler ones, which is the upper left, which is the centralized interdependent business units. So what that means is that you're gonna make a centralized set of data professionals, and then the business units that they serve are gonna be interdependent through their need for the centralized data team’s services. Then the skillset, when it's generalized, well, that means what you have is a business that may be very mature. Maybe the business doesn't change that much. And so the standardization that you need is just making sure the data pipelines are running and they're robust, and when people are doing their weekly, monthly periodic reporting, the reports are ready. So that's like the basic bread and butter data service model, right? So again, centralized data team and you have generalized skill sets because you're just focused on a business that just needs the basics out of their data.

[00:09:01] Michael Meyers: I don't know if we should go in increasing complexity or, you know, is this like part of the data maturity curve? Is there a logical next place you would be, or is it just dependent upon where you're at as an organization?

[00:09:10] Dr. Duru Ahanotu: Yeah. Well, let's take the next step in the complexity. And yeah, so it's not in the graphic. You know, I will refer to the data maturity curve occasionally. It's not directly integrated in here because the data maturity curve is kind of like this environmental variable in the background, and you could be in any one of these quadrants and be going up the data maturity curve. So it's not a direct dependency in a sense. But for instance, if I'm taking the next step in the complexity, this is where this question of, hey, is data gonna be a core competence for your business comes in. So the next stage of complexity would be, yes, I'm centralized in terms of my data professionals, but now I need specialists. I need specialized data professionals because data is going to be, must be, a core competence of my business. And it's because either I'm delivering data as a service or product, or data is an integral part of making my services and products successful. And what does specialized mean? So if we think about the Insight Supply Chain and the very basic roles, the three-part roles – the data engineering, the data analysts and the insights analysts – when I say specialize, I'm saying I'm gonna hire people who specifically are experts, deep experts in data engineering. They're deep experts in data analytics, and they're deep experts in insights analytics. And let me just talk a little about the insights analytics because people, this is a term that came out of my work some time ago, so people probably haven't heard this term before. But think of insights analysts as folks who are really expert in the business and know how to take the work of data analysts, whether it's their own or someone else's, and really evangelize a particular decision based on the data and the evidence that's been produced. And so they can talk directly to executives. They can make the data plain and clear. They don't use data professional jargon to get their point across. They can do that translation layer very quickly and they're very good at promoting results. And they can do visualizations in dramatic and coherent, clear and crisp ways.

[00:11:06] Michael Meyers: I'm sitting here wondering, how is data a core competence a question in today's world, right? Like, isn't the answer yes?

[00:11:15] Dr. Duru Ahanotu: Well, okay. So, you know, there's a long history to this term, even “core competence.” I remember it was really big in the 1990s that, hey, every company needs to have a core competence and they need to focus in on what that is and then cut the rest of the organization off, right? So I think now we have a more nuanced understanding of what core competence means. So for instance, back in the nineties, if I told you my company's core comp, data is a core competence in my company, you might advise a company, okay, get rid of all the other functions and then focus on creating data or data products and data service. But that's not quite what we mean today. What we mean is that data as a core competence means that if you don't take, if you don't professionalize in the sense, your data functions, then you are not gonna be successful in the marketplace. And you may be undercut by competitors who do take data seriously as a strategic lever for the business. That's another way of thinking about it: data as a core competence doesn't mean it's the only thing you do. It means that it is a very important strategic lever for the success of your business. So yes, now today we say, well, duh, of course data is a core competence because data is everywhere. We have big data. We went through big data, and now it's AI. I think AI in particular is going to make this question very salient and, more than ever, just about every organization will have to say yes, because you're not gonna just be able to sprinkle AI on your business and then make magic happen. You actually have to make sure that your data's well governed, high quality, it's structured or unstructured with intention such that the AI can make good sense of it. And that's a whole discipline in and of itself, but still the Insight Supply Chain still speaks to that necessity, with the different roles and functions along the path.

[00:13:19] Michael Meyers: I mean, the unfortunate reality is, joking aside, data should be a core competency for a lot of organization and it isn't because of everything from where your professionals are and your inability to properly address and structure that and what we're talking about here, to related problems like the fact that you adopted all of these SaaS tools and they're sprinkled across your organization and they don't talk to each other, and you can't really leverage your data in a way that you want. And AI isn't yet fixing that and may not for some time. So it's not just… and I guess that brings me to something else that's in the back of my mind with respect to this. You know, I'm a eat‑my‑cake‑and‑have‑it‑too kind of personality. I want it all. I'm looking at this saying, well, that upper left quadrant, that seems like table stakes. I've gotta have those pipelines in place for my business to enable people to function. But I also wanna move more towards that lower left if I'm up there. Is it one or the other? Certainly you might shift over time, but can you exist in a world where you have both of these in place?

[00:14:19] Dr. Duru Ahanotu: Yeah. So actually I have two thoughts that come to mind. One is that we're making an allusion to the data maturity curve. As I said, it's always this sort of environmental variable that comes up when you're talking about what kind of states exist in the organization at any one time. And yes, you could have different parts of the organization at different levels of maturity because you just have different needs that are across the business. So your business may not be a big monolith. You may have the cash cow. In fact, let's talk about it in terms of a cash cow, a legacy business that does move slowly. Everyone understands exactly what needs to be done. You just turn the crank, turn the products out, and boom, you make sales. And then you have other parts of the business – and companies like Google of course, and Facebook or Meta understand this – you have other parts of business that are much more innovative and they are moving fast. They have to create and destroy products and ideas at a rapid pace. That can exist in the same company as, let's just say, in Google's case, the cash cow would be the search business, right? And not to cast any aspersions. I'm not calling Google search slow and plodding. I don't work there, so I don't know, we're just using it as example. But you know that Google, in fact, they went through that restructure and called themselves Alphabet because they recognized that they had this cash cow. But then they have to be purposeful and intentional about all these other businesses and opportunities that they're incubating. And I can imagine in those other smaller organizations that they're incubating that they have a lot of specialists, data specialists, that are working hard to advance up the data maturity curve as fast as possible and making data a core competence of these new emerging businesses. And so, yeah, you can have different parts of the business moving at different speeds because you have different product lines and different services and different marketplaces that you're serving. It's not a one size fits all when you have a very diverse portfolio.

[00:16:23] Michael Meyers: We do a bunch of work with Google, but not with the search team, so you can say whatever you want. You can see a distinct difference between our personalities. You see. The data maturity curve, you called it an environmental variable a few times now. I would call it an existential crisis.

[00:16:39] Dr. Duru Ahanotu: That's fine too.

[00:16:41] Michael Meyers: And as an existential crisis, that's where our opportunity is to help organizations succeed and do better, when we can see where those crises are and the pressure points that are creating those crises. I'm looking forward to our episodes on the data maturity curve. I think it's a really interesting model. If people aren't familiar with it, it would really help them to understand it better. So we've covered the top left, the bottom left. What's the next step or where do you wanna go from here?

[00:17:06] Dr. Duru Ahanotu: Yeah. So let's talk, or let's get another level of complexity. And this is the case where we're in the bottom right, and this is the case where we are now decentralizing and we have specialists and data is, we've decided, a core competence. Now, in this case, what you may have is a lot of data scientists, for instance, in various parts of the company, and these data scientists, think of them again in terms of the Insight Supply Chain, are serving like the insights analysts who speak the business but also can organize the data, who know how to analyze the data and build models and algorithms and so on, but they do it in the language of whatever particular vertical or product and service offerings that they're working on. And a term that I've come to learn recently that I really love is this artisanal aspect to the work. And for those who aren't familiar, an artisan is very specialized. You could think about in the old days they were the craftspeople who had very specialized skills that took years to hone. They may have served as an apprentice and they learned by doing. And so they have very specific, their skill sets are very specifically tuned to solve problems in a particular domain. And in this case, when we're talking about the business, they're solving problems specific to a set of products or services. And so in this case, they are specialized because they're data scientists. But then we also may have specialists with data engineers, because we don't want the data scientists to spend too much time building pipelines and doing the data governance and all that kind of stuff. Now, the fantastic thing about this new world is that there are more and more tools that are blurring some of these lines and enabling data engineers to function as data scientists and data scientists to function more as data engineers. So that's how we come up with terms now like analytics engineers who are kind of bridges along this line. But anyway, long story short, this is where you're specifically trying to monetize your data. So data's a core competence. You're trying to monetize your data in a specific way by creating data products. Again, all of our big tech firms, we know them well. We know that they are creating, using, collecting data as a specific product. And their customers are very specialized sometimes, in non‑overlapping verticals. And they have to move rapidly. Individual verticals have to move fast because they have competitors that are constantly innovating. And when you have that fast‑paced environment, having people have generalized skills may not help because innovation and creativity moves fastest when you have deep set of skills in a particular domain, and that's what we call specialists. But you need to have a structure around them such that they can communicate their innovative and creative ideas effectively so that they make it out into the organization. So it's no good if you're very good at a particular thing, but you can't integrate it into the rest of the business. And so that's why in this case, the decentralized model exists because you need to integrate those specialists with the business folks and they can work well together and move fast. So, sorry, that was a long answer to your original question, but there's a lot here. We stepped up complexity, and the more and more complex you get, the more fluidly you have to think about how you're organizing the teams, the data professionals.

[00:20:39] Michael Meyers: I mean, we could do an entire episode just on this one quadrant. We're trying to get an overview. But I love the way you described it. My background in startups, I think about how you drive innovation in environments and this quadrant speaks directly to that. You want to have specialists throughout your organization to drive innovation, not to be held back by the organization as a whole. It makes a lot of sense to me why this model, to me, helps illustrate from my background and context, why it's so important to follow a model like this and how it can clearly define success and failure, or you need to brute force things. You really do fit into an aspect of this for parts of your business at least.

[00:21:19] Dr. Duru Ahanotu: Exactly.

[00:21:20] Michael Meyers: I'm gonna start referring to, like, I'm gonna share my artisanal spreadsheet with you. I gotta find ways to work that into my vocabulary. I love that.

[00:21:28] Dr. Duru Ahanotu: I know, I just love that word and it's a great way of honoring the past in terms of how people develop skills in the workplace.

[00:21:36] Michael Meyers: I love it. So that leaves us with the upper right?

[00:21:40] Dr. Duru Ahanotu: Yeah. So the upper right, you notice I was going, what I was saying is sort of this linear increase in complexity. And so the upper right is not necessarily the increase in complexity, because again, the folks in this model are generalized, but they're decentralized in business units and data is not a core competency. But the complexity here is getting generalists to move fast enough with the business. So the emphasis for these individual business units will be they need to address certain business problems, marketplace problems as fast as possible. They don't need to innovate in the data realm per se, but they do need folks who know data and can offload the work of data from the other people in the businesses. You know, we talked about earlier the sales, marketing, operations, finance, HR and whatnot. So the generalists know enough to help folks, but they don't have to be specialists in a particular data domain because data's not a core competence in this particular case. Now, as we discussed earlier or inferred earlier, this upper right quadrant is probably gonna become less and less relevant over time, just because, in some form or fashion, data is going to be a core competence of any fast moving organization.

[00:23:02] Michael Meyers: I wonder, you know, you go back to how everyone's becoming an analyst. Perhaps some portion of your organization might fit into that need. Because another thing we talked about is you can just as easily extrapolate and make bad decisions from data. The idea that empowering everybody to make decisions based on data is both great but something that needs to be checked, and having data professionals, generalized support, to help a broader army of analysts, that could be the mainstay, the bedrock of every organization that has to have that. I don't know.

[00:23:30] Dr. Duru Ahanotu: Yeah. I mean, I can envision a world where there's a role called the data coach. Let's just call that person the data coach. They could act like a consultant for all the folks that are using data. We earlier talked about data users. My favorite example is finance, because in a startup situation and in a big company situation, I often found the folks in finance to be the most savvy in the data user class who were not themselves data professionals. But because they're all about the books and making sure the money's flowing and all that kind of stuff, they don't have time to build out pipelines and make data a big part of their job. But they are savvy enough because they deal with numbers all the time. They understand the concepts of trends and correlations and all that kind of stuff, statistics, all that kind of stuff that data professionals deal with. And so in that case, you have a very savvy data user, so you could just serve as a data coach, trust that they can run with the tools that you give them that make their life easier. Whereas, again, I'm speaking from my own experience, I won't mention any specific companies, but whereas in marketing for instance, you have data users who, they don't have time, they don't have the compunction to crunch numbers and to go through and figure out… and I'm not talking about marketing analysts, so all you marketing analysts out there, I'm not talking about you of course, because that's again a very specialized role. I'm talking about the general marketing organization. So the general marketing organization may have marketing analysts, you've seen that role before, because they are offloading a lot of the data user stuff that marketing has. Or if they don't have marketing analysts, then again they're relying on the data professionals in a centralized team, whether it's within marketing or whether it's in the organization. Anyway, all that to say is in this data coach example, you can dial up or dial down the amount of collaboration that you need to do depending upon the savviness of the user you're dealing with. Always cognizant though that the organization should always be cognizant of what, again, career path are you creating. So are you incentivizing your marketing, your sales, your finance folks in their performance reviews? Are you evaluating how effectively they use data? Then the data coach will need to dial their support accordingly because this person's being judged on how well they use data, and not just how well they use me, the data professional, as a resource.

[00:26:10] Michael Meyers: Very fluid. As more and more of your organization becomes analysts in their day‑to‑day function, we talked about this a little in the previous episode, it's critical that you complement and support them with data professionals. And so as the analysis component grows, it's not replacing professionals. If anything, it's increasing the need for professionals to make sure that your, quote unquote, army of analysts are effectively utilizing that data, making decisions with it that are going to lead to value and not serious problems.

[00:26:37] Dr. Duru Ahanotu: Yeah.

[00:26:38] Michael Meyers: So, I mean, one of the things that I'm wondering is, I'm an organization that wants to use this framework. Is it aspirational? And by that I mean, do I look at where I want to be or do I look at where I am today when I try and figure out how to organize my organization?

[00:26:55] Dr. Duru Ahanotu: Yeah. So this comes down to organizational strategy. You should always understand where you want to be. You should always understand where you are and then be able to assess the gap between those two. And I think, again, as we were alluding to before, almost no matter where you think you are in terms of data as a core competency, you can look ahead in the future and anticipate that to some extent, data is going to be a core competency. And then this entire matrix, the bottom of this matrix, is what's going to apply to you. So you may be at the top part of the matrix, but you should be thinking about, over time, you're gonna have to move yourself to the bottom. Again, whether it's the centralized organizational model or the decentralized model, you are going to need some level of specialization when data's the core competence. And again, I know we'll talk about this in the future episode. AI adds another interesting layer to this question because what people are going to try to do is make AI the data specialist, and there are all sorts of trap doors if that's the approach you're going to make, in terms of creating a core competency in data. And there'll be all sorts of blind spots because you're not taking advantage of the experience and the training of data professionals who can spot gaps. They can do verification, they can do validation of errors. And they can also help people use, again, use the tool effectively given the context of what the data ecosystem itself looks like. Yeah, so that's the way I think about it.

[00:28:34] Michael Meyers: One of the things that stuck with me from the first episode, you talked about how SQL was to be the holy grail, to enable non‑technical people.

[00:28:41] Dr. Duru Ahanotu: Oh, I love that one.

[00:28:43] Michael Meyers: You reach out, like 50 years later, here we are, you have to be an engineer. But it was a meaningful step forward, and AI, as it is today and will be over the next couple of years, is a meaningful step forward. And it does enable organizations and especially people at smaller organizations. They're able to do more than they were before. But it just keeps coming back to the need to have real data professionals, to think about the way you structure your organization to empower your larger teams, to make sense of your data and your systems. That's another thing that we haven't talked too much about, but like all these different systems and enabling you to have them talk to each other, be able to leverage your data. There are all sorts of problems that organizations face that data professionals are core to solving.

[00:29:24] Dr. Duru Ahanotu: Yeah, and I think the interesting thing is, you know, we're going to get into, we are in this now era of agentic AI. I mean, every year people are pushing these new faddish terms and soon there'll be armies of agents, AI agents, and who in the world is actually going to be on top of managing how and whether these things are all collaborating effectively? I contend it's gonna be some flavor of a data professional that's gonna be tasked with making sure that these AI agents are collaborating effectively for the organization. I could see someone right now thinking, well, couldn't I just build the AI agent of all agents to manage all of the army? I was like, no, don't do that. I think we're still, you can't just hand over the keys to the kingdom and then go away and expect profits to be piled up at the end of the year. No, that's not. Again, if we talk about the holy grail, that's I think what is in the back of some people's minds, but be very careful.

[00:30:22] Michael Meyers: I think a human-in‑the‑loop strategy is really important. I'm looking forward to our upcoming episode on AI as it relates to all of this. And I really wanna dig into the data maturity curve. I think that's gonna turn into more than one episode because there's so much going on there. But I really appreciate your digging more into the Data Org Matrix and for talking more about the framework.

[00:30:45] Michael Meyers: For everybody who tuned in, thank you so much for listening. You can check out past episodes on the framework as well as other topics at Tag1.com/podcast. We'd love your input and feedback. If you have questions about the framework, we'd be happy to answer them. You can reach us at [email protected]. And of course, please subscribe so you don't miss out on future conversations. Special thanks to Tracy Cooper and June Gregg for producing today's episode with input from Hank Vanzile and Cassey Bowden. Until next time, take care.