Building High-Performing Analytics Teams with John Thompson at CSL BehrinG

Data-Driven Podcast

Listen to this discussion on building high-performing analytics teams, the evolution of the enterprise data warehouse, and what’s in store for the future of analytical technologies.

See All Podcasts Button Arrow
Quote icon

“I’ve been at it for 37 years now in data and analytics. I built many, many, data warehouses. I worked with Ralph Kimball and Bill Inmon – and many founding members of the data warehousing movement from the business intelligence realm into the advanced analytics world.

The enterprise data warehouse is where it’s all at. I think the future for technology is really bright. I believe that we’re on the cusp of explainable AI. I think that the days of black box neural net models are pretty much over. We’re going to be able to have utilities to use our best analytical technologies and have clear, transparent explainability.”

A lot of people look at analytics teams and they’re like, oh, you know, it’s the data nerds, the math nerds. And, we are. I take that label and wear that proudly. The thing about an analytics team is that it’s a creative group. You’re not managing a bunch of IT professionals. I’m not being degrading about it, but they’re very cut and dry. In an analytics project, there’s failure. And we plan for failure. We plan for experimentation. We plan for trial and error. We want people to do things that are going to blow up in their face. And that happens when you’re doing creative kinds of things with data and algorithms and analytics. So, an analytics team is not an IT team. It is a creative team that is driving business value through synergistic and interesting ways to use data and math.


Dave Mariani: Hi everyone. And welcome to the data-driven podcast. And today I have a special guest, it’s John Thompson and John is a best-selling author, a keynote speaker. He’s an innovator. He’s been in the industry for data and analytics for quite some time. And I’m just so, so happy to get a chance to talk to John about what he’s seen and what he’s working on now, when it comes to, creating, data, using data science, data as product, to really drive change in, in enterprises. And, John’s a prolific writer and author he’s got, he’s got some great books for, for C-level executives on analytics. has he authored how to win with intelligence he’s got, another, a book for our director level sort of folks called building analytics teams, harnessing analytics and AI for business improvement. And I think he’s working on something new that you might tell us about. So John, welcome to the podcast.

John Thompson: Hey Dave, thanks for inviting me. I’m so excited to be here. I’ve seen some of the other episodes and some of the other, the guests that you’ve had in listen to something I’m, I’m excited to be a guest. So thank you so much for inviting me.

Dave Mariani: Yeah. You’ve got a story career, which we’ll get, which you know, which we’ll get into, but, just John, just tell the listeners a little bit about yourself and, and, and what you’ve been working on lately.

John Thompson: Sure, absolutely. Thanks, David. I always, always love being able to talk about myself.

Dave Mariani: We want to hear and we want to hear it. Don’t be don’t, don’t be too shy. Come on. Yeah.

John Thompson: well, I’ve been at it for 37 years now in data and analytics and I started out, you know, way back when it was part of the metaphor mafia with bill and those, those, those folks, when we first came around and it was, you know, I think we called it business or decision support or something like that, executive information systems or those kind of funny names that we all laugh at now. And I was a consultant. I traveled and built many, many, data warehouses. I worked with Ralph Kimball and Bill Inman and the founders of the movement moved up through the data warehousing business intelligence realm into the advanced analytics world. I was part of the software, the first, boom and bust of AI with companies like data, mind and magnify and all those companies. And then had a hand in creating some technologies, predictive modeling, markup language.

John Thompson: I was, I was a part of that movement for model portability. And then it was at Dell for a while and we created something called a N D native distributed analytics architecture for more model portability. and as of late, I’ve been on the user side building analytics for different companies, and I’ve been books, as you said, the first book was analytics, how to win with intelligence C-level primmer on how to create an analytics function and then building analytics teams. So as you said, it was for directors on who to, how to hire teams and manage those teams for high performance. And then my next book with the manuscript is complete and I’m talking to publishers right now. It’s called the future of data. So I’ve been at this for a while and excited to still be doing it.

Dave Mariani: You know, it’s like, we, you know, we, I know a little bit about your background, cause we’ve talked before about, and, and really I’m a computer historian and I was, I love reading books about how things began, especially when it came to the PC and the personal computer and the like, and, and so, and so, you know, John, you were there for metaphor. and I think that, I think metaphor came out of Xerox Park. If I, if I remember right. talk to us, talk to us a little bit about those early heady days, where, you know, where the graphical user interface was still sort of, theoretical at the time.

John Thompson: Yeah, it was, it was, it was a great time, you know, I, I was working in a bank in the IT department and I heard about this crazy company called metaphor. And at first they were like, yeah, kid, you can’t, you can’t be on the team. You’re not good enough. And you know, it wasn’t consultant or anything like that. So I had lunch every week for a year with the hiring manager until he hired me. He actually lead a job for me. So he was like, you’re not going away. Are you And I’m like, Nope, we’re not going away. So, yeah. Because of the star workstation coming at a Xerox Park, Don Massaro and Dave Liddle went over and licensed the technology. It was an engineering workstation with a graphical user interface connected to an ethernet backbone, which was an experimental IP network at the time to a Britton Lee hardware, software database, backend predating Oracle.

John Thompson: So, you know, it was one of those things that I was just a kid and I didn’t know what was not possible. So they dropped me into places like miles lab and Pillsbury and Coke, Cadbury Schweppes. And I lived in London for a while and, you know, business users would come and say, Hey, we want to understand how to predict a target market, where to put these new products in. And I built an application that predicted the best target market for a shelf-stable vegetables for Pillsbury. And I think they used it for 15, 20 years afterwards. So it was nested down about 17 levels. It did all sorts of crazy calculations. It ran about 24 hours each time you hit run and it worked. So it was, it was great fun.

Dave Mariani: That’s insane. You were doing, you were doing sort of prescriptive data science back in, back in the day before anybody knew what the, what it was called. But even before business intelligence really, was really invented. It came onto the market. If you guys, if the audience, if you guys want to read about all this, you know, Walter Isaacson did it, his innovators book that he wrote is a really good sort of description of, of, of, of all this history. And, and this is one of the, what, this is one of the chapters that’s prominent in the development of the personal computer and, and really analytics as we know it today. So it’s pretty awesome that you were a part of that.

John Thompson: Yeah, it was fun. I remember going to miles labs and not only did you have, you know, in your head CQL and all the languages you were working on, but you had a toolkit too, cause you had needed a screwdriver and a wrench and all this kind of stuff, because you were taking workstations apart and putting in circuit boards and things like that. So it was, you were hardware, software technician.

Dave Mariani: Yeah. I love it. I love it. It was like, yeah. When hardware and software were equally important, we’ve lost, we’ve lost our sensibility of hardware. It’s all locked away and sealed and we never get to see it unless, you know, you really, you really want to try hard, but anyway, miss those days. So, so let’s like, okay, let’s, let’s fast forward to today and the challenges it’s up today, but back then, you know, we didn’t have, we didn’t have the ability really well. We didn’t have the sensors, we weren’t generating the kind of quality quantity of data that we do now. I mean, everything is tracked. and so there’s so much more opportunity to collect data and then also, it’s a challenge to actually figure out what to do with it and how to make sense out of it. so, when you think of, when you think about, you know, when you think about what, what analytics leaders, need to do or need to be prepared for, you know, you’ve written about it in, you know, in, in your, in your building analytics teams, book, what, what can you say What can you summarize, John for the audience in terms of some of the things, some of the takeaways for how analytical analytics leaders, can make sense of, of, of all this data that’s out there

John Thompson: You know, it’s a, it’s a great question. And it’s really challenging for everybody. The, the business users, you know, the C level executives and see SVPs and EDPs are flying around on planes and reading books and saying, well, it’s, you know, it’s, it’s just do it. It’s all there is to it. You know, you have, you have the data for everything, just get at it. And you know, when you really delved into it, you know, there are some problems that we don’t have data for. So data scientists, data analytics, leaders need to be clear with the business leaders, what you can solve, what you can’t solve. You know, I often say we’re data scientists. We’re not magicians. You know, if we don’t have the data, we can’t do it. You would just can’t make it up. the other thing that I always liked to talk to people about is, yes, we do have data for almost every phenomenon out there, but the thing about analytics that we’re all trying to get to is what we’re trying to do is we’re trying to build models that represent reality.

John Thompson: You’re trying to understand a phenomena, a customer’s mindset of the optimal price, how oil flows through a pipe in a mayonnaise factory. You’re trying to model reality. Reality is never just one set of data. So it’s not a billion images of school buses. It’s probably five or six or seven different data sources, and it may be four or five or 10 models working in conjunction. So people want to think, oh, it’s a set of data and it’s a mob. That’s not where we are today. It’s a collection of data, an ensemble of data that may be integrated together. And it’s a set of models that are taking the insights from one treat, you know, sending to another and another, and you have pipelines of data and you have pipelines of mops. So if you were at that point, you were at the apex of the market. If your analytical team is not having those conversations with you yet, you’re still coming up the curve of sophistication.

Dave Mariani: Yeah. You know, John, when I talked to other smart people like yourself, when it comes to data science, there’s a, what I, the common theme I hear is involving the business, the business right in the beginning of the process. And so what can you say about that? What’s your opinion on just how good data, how it, how a team, how you put together a team to get a good outcome and good ROI out of an investment with data science.

John Thompson: That’s a great question, Dave. You know, and, and, you know, I talk about this a lot in my books and in podcasts and speaking engagements and keynote addresses is that the business is the key to everything. So, you know, the first question I have with, executive is, do you want a number, or do you want sustainable change? And if they want a number, that’s a project and I get a data scientist, and then we put together some information, we give them a number and we’re done. That’s it. If that’s all you want is a number, then that’s a project and we’ll do it. And sometimes three hours, sometimes three months just depends on the sophistication. Now, if you want sustainable change, that’s a program. Now we’ve got to have a conversation. Me as an analytics leader, you as a business leader on how much change you can plan for and ingest and tolerate, because we can drive analytics that come out instantaneously.

John Thompson: Now, can you change your processes to take advantage of those insights? If you can’t, then there’s no sense in us doing the program, because it’ll just be data that ends up on the floor or in someone’s email box. And no one will take action on it. So we generally don’t engage in those projects. And when we say no, politely say no, and you know, we walk away and we do other things, cause there’s many other things we can be working on in generally then the business leaders will come back some months later, four months later, six months later and say, you know what I thought about what you said, and I think that’s a great idea. Can we collaborate on figuring out how we can be more data-driven and how models can help us understand what’s really going on in the business. So the first time you have those conversations with a new business leader, be ready for them to say, no, I’m not up for that. I don’t have the stomach for it, but they’ll come back around.

Dave Mariani: Yeah. And so, you know, I think about that and it’s like, it sounds kind of organic. Like it’s a, you know, it’s as opposed to like a, I don’t know, what I think is something like it’s more tangible and more, you know, actionable. I think that there should be a process where these things can happen. Right. But is it, so John, is there, like what can enterprises do to make sure that this happens not organically and by accident, but I don’t know by w w with a little more direction. I don’t know if you know what I mean. Yeah.

John Thompson: Yeah. I know exactly what you mean, Dave, and, and, and it needs to come from the top, you know, it’s, it’s one of those things. We have this conversation pretty regularly. People say, you know, Hey, where should the analytics leader be in the organization? And I say, as close to the leadership, and it’s close to change agents as possible. You know, if it’s an, it that’s a problem, CFO, that’s a problem. you know, other functions, operations that’s the problem, you know, because the people who have the, the power to make that, the make that a, an initiative are generally the CEO and the COO. If you’re not reporting into them, you know, then you’re, you’re going to have problems driving that change. Cause you really wanting to have data drive the leading edge of the development of the organization. And the CEO has to make that an initiative that people are thinking all the time, how am I going to continuously improve How am I going to get a competitive advantage How am I going to leverage the data that I know we have that we’re not using It’s gotta be an initiative coming from the top.

Dave Mariani: So, and that’s, so that’s a good point. So when it comes to the organization, so you’ve got to, it’s either coming from CEO or the COO, is there a, is there a chief data officer or chief analytics officer What are you seeing, John And like in terms of enterprises, how, how to sort of operationalize that sort of, that, that direction from the top is who’s responsible for that. Is there an office of the CDO What what’s, what’s it look like

John Thompson: Oh, wait, it looks like, it looks like everything right now, Dave. I mean, you, you know, this as well as I do is that CDOs and CAOs and CDI owes they’re, they’re all over the place. And it’s, it’s an evolution. Some is an evolution of CIO title. Some are brand new titles, you know, the CDOs that I see a lot of them are defensive, not personally defensive, but they’re doing defensive kind of things like

Dave Mariani: Reactive. Yeah,

John Thompson: Yeah. Data governance and, and integrating master data management and stuff like that. The CAO I see as more of a, a, an offensive role, you know, it dry for a change, you know, someone who’s looking at it and saying, okay, the data is there. Whether it’s well taken care of or not, I can get at it. And I can do things that are actually gonna improve revenue, improve customer retention, improve manufacturing, efficiency, improve the supply chain. So I see a lot of that going on. And if you ask me where those roles are, they’re everywhere. You know, I see a lot of openings right now for CDOs that are reporting to the CFO, which is kind of weird in my opinion, that’s

Dave Mariani: Weird.

John Thompson: And then you’ve got a CIO that’s reporting into the CIO. That’s weird. you know, and, and you’ve got things I saw one the other day, someone clipped it and put it on LinkedIn. It was a C, D O with a salary of like a hundred thousand dollars. I’m like, well, I’m like, well, you’re only a hundred thousand dollars level. Okay. Go ahead.

Dave Mariani: So, John, what would you, what would you do? It’s, it’s the it’s, it’s a John Kay Thompson corporation. So, so how would you organize that top team

John Thompson: Yeah, it would be the CEO and the, a reporting to the, to the CEO and on a peer level with the, the COO and all the other C level executives. And they would have a mandate to go and work with all their peers about having a data-driven initiative that is continually improvement, that the mandate is continuous improvement through data. And that’s, if I were the CEO, that’s exactly, it would be one of the top three priorities that I had.

Dave Mariani: I love it. I completely agree with you on that, that needs to be a need that that role needs to be, at the table, with the rest of the C execs reporting to the CEO. I think that’s the only way it can happen. Cause it’s cross-functional right, because you can use data to drive improvements in every, in, in, in, in, in every part of the company. whether it’s, you know, whether it’s, you know, finance, sales, whatever, whatever have you, it’s like, every discipline can, can use, data and data science to improve performance. so it needs to be, it needs to be cross-functional that way. okay. So, so you’ve written a lot about, sticking on the, sort of the, the, the teams and organization kind of, concepts here. You’ve written a lot about this, so that’s why I want to drill down and get your take on this. So when you’re hiring and, you know, if you’re an analytics leader and you’re looking on building out a team, like, what do you think an analytics leader should be looking for when they’re sort of building out that analytics team and on operation

John Thompson: You know, it’s one of the things that people need to do is get their job description straight. you know, I reworked all the job descriptions in my current job to have all the way from intern to a principal data scientist. You need to make sure that those people can go up that track and they can jump over to management if they want, or they can continue all the way up and have a fulfilling career as a data scientist, making good money. That’s number one, number two, is you on a hire for intelligence I mean, every data scientist is going to be intelligent. There’s no doubt about that. It’s kind of a prerequisite. And then you want curiosity and you want honesty and you want, you know, hard work is what you’re looking for. That’s what I look for anyway. And, you know, and I get these people that come in and when I’m hiring and they try to explain to me that they know and can do every type of model known demand, neural networks, natural language processing, natural language, understanding classification, clustering, explainable AI.

John Thompson: And, and I just look at them and I say, look, you know, you can either start being honest with me, or we can end this interview right now because nobody knows all that. No knows it. So, you know, what you’re looking for is an ensemble of people. You know, someone who’s good at neural networks, someone who’s good at NLP, someone who’s good at classification and clustering, someone who is an expert at feature engineering, and you want them to work together as it is a harmonious ensemble of people. I really liked that word ensemble. You know, you want them to be focused on projects that the person, the lady that’s a neural network expert can work with the feature engineering expert and they can do a great project. And then the kid that’s coming in at college, who wants to learn neural networks will be a second on a project with her.

John Thompson: And it just everybody’s learning all the time from everybody else, you know And if you get someone in there, who’s, you know, a fly in the ointment, get them out as soon as possible. Cause that is, that’s just terrible for a data science team. So I’m looking for a people, a group of people who are real about their skills, real about what they want to learn about and really want to solve business problems. You get some people that are like, Hey, I’m a data scientist. I only do the modeling. No, not on my team. You’re going to do data acquisition. You’re going to do feature engineering. You’re going to do model building. You’re going to do subject matter expert engagement. You’re going to talk to executives. You’re going to present your findings. You’re going to do it all. And if you’re a pre-Madonna, that’s only once one step in that process, I got no time for you.

Dave Mariani: Yeah. You know, I couldn’t agree more. It’s like, I think a lot of, a lot of leaders end up, like over rotating on, on, experience and looking for somebody who has the exact experience that they’re looking for in spades. And, and really it’s like part of the fun and energy comes from learning new things. So like you, I like to hire, you gotta make sure they have, you have the right mix of skill sets to be able to be successful in doing what you’re doing, but it’s okay to allow people to grow into those roles and to be mentored by people who know the subject better than they do, because it’s exciting for them. They’re going to be engaged. They’re going to be learning and they’re going to be happy and contributing, and you’re going to be building your, your, your bench, and for those next set of leaders.

Dave Mariani: And so I completely agree with you. The other thing you said, which I think, I just want to emphasize, for the audiences, like is I know what sounds like big corporation and it Yahoo, you know, I’ve always been a startup guy. And so when I got acquired into Yahoo, I learned about job leveling and PR and job titles and the like, and it’s like, oh, here it comes. Right. I can tell you that, having the job levels where you have specific skills and achievements to be at that level with a pay grade, that’s at that level, it get charts are really nice course for career development. and, and by doing that, it’s like, you know, when you sit down with somebody, you say, well, these are the things you need to, develop skill-wise to get to this next level. It makes it very prescriptive versus just being a being subjective.

Dave Mariani: so I mean, I, like, I really liked that. and I did that at that scale as well. And the other thing you said was, you know, it’s okay to have, I think it’s important to have two tracks and then an individual contributor track and management track. And those you can mix in between if you want, but you gotta be able to reward people equally, whether they’re an individual contributor or whether they’re a manager. I think the, I think the other mistake people make is that they spend so much time that they focus on, oh, if you’re a manager, that means you’re worth more to us. So you should be paid more than an individual contributor. And that’s just so far from the truth. And, and it’s wrong and you gotta have the right mix. And if somebody is a good manager, somebody is a good individual contributor. It doesn’t mean you forced them into a management role. because that may not be what they want to do, or if they’re good at, and I think we tend to, we tend to do that too much, especially in sales, you see that happening all the time. Yeah. So

John Thompson: I’ve had data scientists that have come to me and said, you know, Hey, I’ve been here for a while. I think I want to try management, you know, and you start, you know, giving them management like duties and, and giving them a taste of it. And some people are like, yeah, I like it. You know, get me on a track to move over to the management side and other people go back and go, you know, I really hate that. I’m like now, you know

Dave Mariani: Yeah. And you know what, it shouldn’t be a financial decision for them either, because I think that creates the wrong kind of incentives.

John Thompson: It’s gotta be cash, it’s gotta be cash and compensation neutral. Yeah. You can stay on either side and still make the money.

Dave Mariani: And that’s where those levels come into play because those levels then intersect there and you can make sure that you’re paying people equally for, you know, or not equally, but you’re paying people, equivalents, equivalency for X, you know, experience and value to the organization. So, I love that. So I it’s, I, I love that topic. It’s often not very well understood. And especially in the startup land where I am in, you know, there’s, there’s, there’s usually little concept of, of doing that. It seems like bureaucracy, but it just, it, it really does. It really is good for all the team members and the employees to know, you know, where they’re going and where they can achieve and how they improve, how they improve their own and build their careers. So

John Thompson: We’re all competing for talent. You know what I mean If there’s a young person that’s, you know, 2, 3, 5 years experience and, and they’re excited by the startup environment, I’ve been part of seven different startups. I love startups. I think they’re fantastic. But if you can’t explain to that person where they’re going to go and how they’re going to make money and how they’re going to develop, they’re going to go to a bigger company and, you know, you can call it bureaucracy, you can call it overhead, you can call it whatever you want to call it, but it’s clarity in you’re giving people clarity. That’s what it is.

Dave Mariani: Amen. Amen, brother. That’s great. All right. So, we’ve talked to, any, any other, I guess, any other, sort of, crystals of wisdom that John, that you may have on the, on just on team development before we move into technology.

John Thompson: Yeah. One last thing that I’d like to say is that, you know, a lot of people look at the analytics teams and they’re like, oh, you know, it’s, it’s the, the data nerds, the math nerds. And we are, I, I take that label, like wear that proudly. I am, I’m one of those people, but the, the thing about an analytics team is it’s a creative group. You’re not managing a bunch of it professionals, and I’m not being denigrated or pejorative about it. Professionals, you know, they do a good job, but they’re very cut and dry, you know, it’s, it’s a, B, C, D E F. And, you know, we’re gonna, we’re gonna sorry about that. We’re gonna get, you know, from, we’re going to get this project from start to finish and people do that, and that’s great, but in an analytics project, there’s failure. And we plan for failure. We plan for experimentation, we plan for trial and effort, and we want people to do things that are going to blow up in their face. And that happens when you’re doing creative kind of things with data and algorithms and analytics. So, you know, an analytics team is not an it team. It is a creative team that is driving business value through synergistic and interesting ways to use data and math.

Dave Mariani: Yeah. I love that. I love that. That’s a great, that’s great. And, you know, it’s, you know, one of my sort of, sort of passionate pleas just to, to, to, to kids in school and especially young girls, is that, you know, it’s, you know, engineering, software engineering, computer science, it’s, it’s, it’s, it’s not a rote subject, right. It’s, it’s as creative as can be. You’re just using a different canvas to create your, to create your masterpiece. And, you know, it happens to be maybe it’s code that you’re using to create a masterpiece, but it’s no less artistic or, or, you know, creative then, you know, then, then, then painting them, painting a picture. So, I just want to say like, yeah, it’s, I’m right with you there that, really anything across, technology, if you D if it’s done right. It’s a creative endeavor. It’s not manufacturing whatsoever. Yeah.

John Thompson: Yep. Yep. Couldn’t agree. More.

Dave Mariani: So, so let’s, let’s, let’s talk a little bit about the state of technology. I want to talk a little bit now about the modern data stack, and, and because, you know, there’s, there’s so much change that the cloud sort of brought into that, that catalyzed people to finally look at what they’re doing when it comes to their data and analytics infrastructure and rethink it all. So, so, so, you know, so John, when, when, when you hear that term, the modern data stack or modern data and analytics stack, what’s that really mean to you What do you, what do you think that means

John Thompson: Yeah. You know, it’s, it’s one of those things that, people, there’s a lot of evangelism going on, and there’s a lot of people that feel very strongly about, Hey, it’s gotta be in the cloud, or it’s gotta be here. It’s gotta be there. It’s gotta be a hybrid. And, and those are all good things, you know, and, it all depends on your corporate profile and your asset management, your depreciation, many of these things are financial decisions. They don’t really have much to do with technology. So it doesn’t really matter to me. If you put the data in your data center, you put it in Jeff Bezos’s data center, you know, or, you know, you know, Google cloud services that’s kind of irrelevant to me. but I do start to differentiate when you get to, you know, the data mash and the model ops and the dev ops and those kinds of things.

John Thompson: Now, we’re, now we’re actually talking about productivity tools. You know, it used to be just, you know, here’s an IDE and, and, you know, hopefully you do better than a text editor with it, but, you know, those things make a difference. And those are tools that developers and advanced analytic professionals and data scientists and data engineers can do their job and do their job well. So I think there’s a lot of options in those, in those various stacks. And, it’s really cool. And there’s a lot of really pay as you go interesting tools that you can get your hands on. You can learn from, you can walk away from, you know, there’s, there’s really no, you know, you got no skin in the game until you’ve decided this is where I want to invest, and this is where I want to learn, and this is where I want to be. And I want to be part of this community. And I think that’s a really interesting, great development for the technology world.

Dave Mariani: Okay. You know, one of the, sort of, one of the, sort of the, debates going on, and this is like, it’s, it’s a debate about data mesh, which to me is more about decentralization of, of, of, of owning data and analytics and the analytics process. You know, we sort of saw like, in the early days of, of, of, of business intelligence, you know, it typically own that you have business intelligence managers and they were responsible for the whole sort of, continuum everything from the data acquisition to how you, the data to actually delivering reports to the business. And, and that model really, you know, that model, I think Tableau blew up that model when they gave a easy to use powerful tool that allowed the business to do their own analytics and build their own reports and dashboards, and all of a sudden, you know, it was sort of scrambling to, to, to, to keep up.

Dave Mariani: but, but that also had its own sort of consequences because, like at Yahoo, when I was running analytics at Yahoo, you know, it’s like everybody had a different definition of what a, an impression was, you know, it’s like, whether it be an ad impression or a click or, or, or a purchase, you know, and so, that didn’t do any, do us any good. So, so, but we can’t go back to it or a central team owning the whole factory. So, you know, so John, what do you see out there Like what’s the right model to allow the business to be able to create and do their, and, and self-serve, but with some sort of sense of, of governance and consistency, what do you, what do you think is the right sort of balance there

John Thompson: Yeah, it’s a, it’s a great question. And it’s, you know, it’s the age old, you know, centralization, decentralization, tug of war that we’ve had since you and I, you know, came into the business world. It’s, you know, the centralized model is the only way to be. And you know, that the data warehouse, the enterprise data warehouse is where it’s all at and, you know, it’s the data lake and instead of data mash, and, you know, it really comes down to, you know, thoughtful architecture is really what it is. And, you know, if most companies, not all companies, most companies have pretty good control over their transactional environments. They have to go out of business. and then the real question becomes what happens with your systems, your analytical systems, your systems of insight, your advanced analytic systems. How do you build an architecture on top of that, where that data comes out of those transactional environments and comes from external places and runs through some kind of standardization engine and ends up in the right end points could be in Google.

John Thompson: It could be an Amazon could be, you know, in on-premise could be, you know, wherever you happen to want it to be. But that standardization engine, that’s usually where the problems occur, you know, because people are not good at it. And what we’ve been doing, what I’ve been doing over the last couple of years, 10 years is that, you know, we have the business take ownership and point in this one, the, the, my day job right now, you know, we’re building something that is a personalization engine and environment for our donors and the, the plasma operations team owns that. So all the data that’s flowing through from the transactional systems through the standardization engine, in, into that end point, that is all owned by plasma operations, and they care about that. And they have it built in a way that it’s purpose-built for them to personalize and understand donors. And I think you almost have to have a partnership, it analytics business for each of those streams. So the people that care about it take care of it. You know, if it ends up in some, like you said, centralized it function, nobody really owns it. And nobody really cares about it. And it grows weeds and gets old and falls apart at some point.

Dave Mariani: Yeah. You know, I, I, I, I liked that model. Basically you architecture is king and everybody using the same tools and leveraging the same tools, it just creates, that, that synergy, that common knowledge and, and also, you know, obviously standardization. So I do like, you know, I do like sort of a team owning the architectural decisions, obviously working in conjunction with the business, but the business they’re the domain experts. So, and they know, they know what’s what it means. It’s really hard. I think to teach a data engineer about the business, it’s easier to teach the business about, about, about the data than vice versa. so, so I think that, you know, if we can empower the business to be able to give, deliver them tools and tools that work and work for them that speak their language, and we can be responsible as technology leaders of delivering those tools. Then we can maybe have the best of both worlds. I don’t know.

John Thompson: Okay. I agree. And we don’t try to foist on the subject matter experts in the business then making decisions about technology or architecture, but do absolutely defer to them on what an impression is. You know, you know, everybody comes to the table and says, okay, well, we’ve defined it as impression as, okay, well now we know what we’re talking about. That’s great. So, you know, the definitional aspects of it totally in the business subject matter experts, all the infrastructure and algorithms, data movement, we, and it take care of that.

Dave Mariani: I love that. That’s great. That’s that’s great. So, so, we’ve been talking for quite some time now, John, and I could talk to you forever, but, I would, I always love to sort of, sort of to wrap up with, with the future. So I know you’re writing a book about it, so I don’t want you to have to give away any goodies here, but what do you think, that, what do you think we should be keeping an eye on, in terms of trends when it comes to data analysts and analytics, for the, for the future

John Thompson: Well, I, I think, you know, I get the question quite often is when is this stuff going to be done And it usually comes from the business professionals. And sadly, when I’m asked that

John Thompson: And they say, what are you talking about You know, what, we have the ultimate database and what we have the great networks and what we have the right mash and the right tools. And, and I said, yeah, they’re always going to get better. But the, at the core of it, it’s data and math and does data and math ever go away. No, it never goes away. So I think the future for technology is really bright. I really believe that we’re on the cusp of explainable AI. I think that we are the days of, of black box neural net models are pretty much over, you know, we’re going to be able to have utilities from, I’ve seen two or three startups doing really cool stuff, where we are going to be able to use our best analytical technologies and have clear, transparent explainability. You know, right now we’re kind of handicapped, you know, we can’t in finance and healthcare.

John Thompson: We can’t use our best tools because we can’t explain how the, in, why they do what they do. But I think that’s a two year window. And then on the data side of it, you know, I think we’re about three years away from you and I, and everybody else in the world owning and monetizing our own data. So the EU with GDPR, it’s been around for five years, the day they act is going into, you know, into effect in the next six months, the data governance act is going to be, you know, it passed and come into a force in 20, 23 or 2025, you know, it’s going to be, you know, Hey, I, I don’t really like the airlines or I don’t like the oil companies. And if they want to use any of my data, they got to pay me, you know, $5 million literally for every piece of my data they touch. And I think that’s a pretty strong message that you don’t want to collaborate with them. So I think technology is going to March on math is going to March on, and we are going to really see a sea change in data in the next few years.

Dave Mariani: Yeah. You know, it’s like, we’re already giving it away. Right. I mean, it’s like, and, and not being cognizant that we’re giving them, we’re giving it away for using these, these free services. but these free services are generating massive quantities, mass quantities of, of dollars. I know that John, cause I I’ve been in advertising, digital advertising, you know, from, for most of my career. So, believe me, it’s the data, that’s the value, for sure. It’s, it’s it’s and so that’s, that’s something that I think that people are now realizing a little bit more, but still not to the extent of just how that data is used. So yeah,

John Thompson: That’s exactly where I wrote the book. You know what I mean It’s a little bit a plug here it’s called the future of data. I hopefully it will get it out there in the September timeframe. And that’s really what it is. It’s an expo say of, you know, the last hundred years of why the data environment we have today came, how it came to be and what the data environment of the future will be.

Dave Mariani: Well, I can’t wait to read it and I’ll be the first in line. so John, thanks. Thanks so much. This has been an amazing conversation. and, and to the audience out there, stay and be data-driven. Thanks everyone.

John Thompson: Thanks, David.

Be Data-Driven At Scale