Jonas Christensen 2:14
Bill Schmarzo, welcome to Leaders of Analytics. It is fantastic to have you on the show today.
Bill Schmarzo 2:23
Jonas, thank you very much, right. Yo, Jonas, thank you for having me on the show. This is great. I'm looking forward to this. Like, I know you wanted to send me questions. And I said ''No, no, no. No questions. Let's do this straight up and have some fun with it''.
Jonas Christensen 2:36
Yeah, all listeners out there. Bill is a true expert. And you will know because he's the first person who has ever said ''No'' to getting my questions before the interview. So everything here is completely off the cuff, which is brilliant, Bill. And we are going to learn a lot today. I will reveal Bill to you that we're going to be talking about the economics of data and analytics and how to get value out of your data, which I think is dear to the hearts of all the listeners here because that is a daily challenge for many people. Before we get to that, Bill, you have a very long and very interesting career across many domains. Could you start telling us a little bit about yourself, your career background and what you do?
Bill Schmarzo 3:14
Sure. So I've been in this industry a long time, over 40 years. I like to say I have a lot of Forrest Gump moments in my life. So the right place, right time, not because I'm tall or good looking or from Iowa. Sometimes in life, you just get lucky and opportunities come up. And when opportunities come to you, the best thing you need to do is take them. And that's what I've done. I've very fortunate opportunities have shown themselves. I've jumped on. I was involved in one of the very first BI data warehouse projects, back in the 1987-1988 timeframe. When I was working with Procter & Gamble, and we were looking at Walmart's electronic point of sales data, which was really the first time we had electronic point of sales data. Spent 20-25 years of my life in the BI data warehouse space. Taught at the Data Warehousing Institute. I've always been in my heart an educator because I've always felt like I've been so fortunate throughout my entire career to have worked with so many marvellous companies and people that it's up to me to give back and share. So I did that for many years and I was recruited by Yahoo!, where I was their vice president of advertiser analytics. And another Forrest Gump moment because this is one of those points in life when you realise that a lot of the things that you learned before were wrong and that the world had changed. And for me to stay relevant, I had to let go of a lot of things I thought were sort of gospel, were the way things had to be done. That was hard. I struggled. I mean, I personally struggled. But when I got through that knothole, I was better off because I now understood the transformation that was taking place in the industry because I actually lived personally through it.
Jonas Christensen 4:49
Could you elaborate on that? What are those aha moments, Forrest Gump moments that made you think differently about the industry?
Bill Schmarzo 4:56
So in the BI data warehouse space, we always try to capture every question that our users are going to ask, so we could build a dimensional model to handle that. So a lot of our mindset was focused on: Capture questions. Build a dimensional model. So my mind was always star schema. I start talking to customers, I'd be building a star schema in my head. When I went to Yahoo!, it was no longer the questions. Instead, what we focused on were the decisions the users were trying to make and how to deliver predictions and recommendations that helped them make better decisions. So if you think about the digital marketing space, you have campaign managers and media planners and buyers, who are trying to make decisions regarding how much money to allocate, what kind of campaigns, what ads to show, what audiences to target, what sites to put it on. There's a handful, pretty broad handful of decisions they are trying to make. And what I learned is that instead of focusing on questions and building out the schema and then letting people sort of discover data, that we can become very prescriptive and directive. That the more we knew about the decision they're trying to make, sort of the intent of those decisions, the KPIs and metrics around which they're gonna measure decision effectiveness, drove all of our, what then became, data science work. So that was a transition from being a business analyst, BI analyst to being a data scientist. We're actually trying to figure out ''Okay, what are those variables and metrics across all the ones we have that are actually better predictors of performance?''. And so star schemas gave way to decisions and focusing on really direct value creation. I guess it was - maybe Jonas never thought about this before but maybe for the first time in my data analytics career, I could see a direct line of sight from the data and analytics to how value is created in the organisation, because we could help make better decisions. And these decisions were huge. Some of them worth, you know, 10, 50, 100 million dollars, and being able to prove these certain decisions we're making. So that was the knothole of me. It's that instead of thinking about data, questions and schemas first, my transition was for ''What's the organisation trying to accomplish? What are the KPIs and metrics against which you're gonna measure success? And what are the decisions the stakeholders need to make and optimise in order to make this initiative successful?'' And then voila, everything else I've done since then it's just, it's all built on that Aha, epiphany moment for me.
Jonas Christensen 7:23
Yeah. And we're going to elaborate a lot on that, because I think it's actually a really, really interesting point and viewpoint that you have. Because it's sort of - you think about - maybe the people listening to this, maybe think that Bill's talking about semantics, decisions or questions. What? Isn't that kind of the same thing? But we're going to dig into exactly why that's not.
Bill Schmarzo 7:44
Good. Because they're not the same thing. Questions and decisions are very complementary, but they serve very different purposes.
Jonas Christensen 7:50
But Bill, before we get to that, I'm really interested in a part of your background, which is what you talked about: You as an educator, and you are affiliated with quite a few universities all at the same time. Could you tell us about your role or roles at all these different educational institutions? And why you do that?
Bill Schmarzo 8:11
Yeah. So what I focus on primarily in what I teach and reason why I do that at different universities, because they all have kind of the same fundamental need. It's around how can we help organisations become more effective at leveraging data and analytics to power their business and operational models? How do we take a - is it trying to become data driven and talking about data? How do we talk about value first? And what happens, especially when you're talking to business stakeholders, is that when you start by talking about data, you immediately lose them. They don't care. They have no interest in data. But when we talk about value and understanding how do they create value and how do they measure value, right? And who are the stakeholders impacted by that value creation? Now you're speaking their language, and they're all in, right. ''Oh, you're gonna help me do this. Oh, so data and analytics helped me do that'' and ''Oh, here's my role in helping the data science team to build the right kinds of analytics''. You've empowered them to be involved in a data science process. I call this ''The Art of Thinking Like A Data Scientist''. I've written a book and talk about it frequently. I got methodology designed templates for it. But it's really about talking the language of the business stakeholders. And while it's, you know, it's getting the business stakeholders to understand, you know, what data and analytics can do for them. What's also come out of this, as I've taught this set for many engineering organisations and such, is engineers want to know how to speak to the business too. Because they start by talking about data. They start talking about AI and ML, and they lose the business users. And you're like ''Well, this stuff is really powerful. It's great. How do I reframe the conversation in a language that they understand?''.
Jonas Christensen 9:52
Yes, just like most customers of most businesses aren't as excited about the products as those working in the businesses themselves. Most people in the business are not as excited by data as those working with the data themselves. So, right from data management to data science, and that is definitely something that we all need to think about. This is a challenge that I have every day with myself and my team as well, to get us to really step into the shoes of the rest of the business. I always say to my teams ''You have so much power at your fingertips with all this information. You can find anything about the organisation by searching and manipulating data. Your responsibility is to think like the CEO, like this is your organisation, like you own this thing. I'm not saying behave like the CEO. I'm saying, think like the CEO. How do you allocate resources and capital? And how do you use data to inform that?''
Bill Schmarzo 10:48
Yes, I was gonna add one other point too. So that is what I teach. Let me tell you why I teach, which is different than what I teach. I have found that for me to be able to articulate to students, - typically, they're juniors and seniors, a lot of them are MBA graduate students but I also do this a lot with businesses, business stakeholders, in particular - is if I'm going to teach something, I really have to understand it. And it isn't about making it longer. It's about making it more concise. I think it was Mark Twain, who famously said ''If I had more time, I'd write you a shorter letter''. And so one of my examples is: How do I describe data science to a business stakeholder, right? Here's a 600 page book about data science. No, no. Here's what data science is about. Data science is about identifying those variables and metrics that might be better predictors of performance. Period. That's all it is. They go, ''Oh, is that all it is? Well, that makes sense. Okay, I can understand that''. And then of course, when I write that down on a board, I circle many times that word ''might''. Because the data science process, like any sort of scientific problem/process, is full of lots of ''might'' moments. You're going to try different data sets that might be useful. Different types of transformation that might be useful. Different kind of out- There's a lot of ''might'' moments and as an organisation, if you don't have enough 'might'' moments, you'll never have any breakthrough moments. And so how do you not only bring data science in an explainable manner to the business stakeholders, but how do you unleash the single most important word, concept of data science, which is the word ''might''? You're going to try lots of different things. Most of the things you try are not going to work. With every try, you try something and you fail, you learn and it's about that learning process. So for me, as a teacher, I have to spend a lot of time trying to simplify things down. That's hard for me. I'd like to talk. I like to write a lot of stuff. It's like ''No Schmarzo. Get it concise''. If you can make it 16 words, that's better than 17 words, and it's certainly better than a 600 page book.
Jonas Christensen 12:56
Well, speaking of books, Bill, you've written - i counted five books.
Bill Schmarzo 13:01
Just four, just four. Well, if you count my comic book, my comic book...
Jonas Christensen 13:04
I did count that one. That's why.
Bill Schmarzo 13:06
Jonas Christensen 13:07
They were four textbooks and a comic book. And you've written those on similar topics. Could you tell us about those books? What's in those books? Who are the audiences? And why did you write those?
Bill Schmarzo 13:17
Alright, that's a good question. So the first book was written when I was at EMC. And one of our marketing lead, senior marketing lead came up to me and says - You know, I was writing blogs about data and analytics. I've written several white papers over the many years I've been doing it and she came to me and said ''Schmarzo, write a book on big data''. Big data had just become kind of this big craze. I just left Yahoo and was now at EMC and she said ''Write this book about big data''.
Jonas Christensen 13:43
What year are we, circa here?
Bill Schmarzo 13:45
I'd have to look the book up. I'm not even sure what year it is.
Jonas Christensen 13:47
Bill Schmarzo 13:48
12 years ago?
Jonas Christensen 13:49
Yeah, okay. So call it 2010.
Bill Schmarzo 13:53
Yeah, probably something around that. Yeah-ish. I could look up exactly.
Jonas Christensen 13:57
That's close enough. Yeah.
Bill Schmarzo 13:59
So I wrote the book and I got a lot of feedback from universities, who wanted to use it. And then I was asked to teach at University of San Francisco by Professor Mawafa Sadawi, who really he and I clicked and became dear friends going forward. We always collaborate together. He said ''Hey, come in and teach one of my life classes''. I said ''Sure''. So I used that book and much of other stuff. And I taught the class and I hated all the textbooks. And even my book wasn't written as a textbook. It was written as a book. I mean, just kind of threw up all these things I had heard. It wasn't really well organised. It didn't have - lots of great ideas, I guess, but it didn't hold together. So the book I wrote as a textbook was my second book, which probably came about two or three years later, maybe two years, called the ''Big Data MBA''. And that book was written, targeting my students to give them something that I could use to teach a class. And I have since found that many universities, not just MBA programmes, but in even in their engineering programmes, now use that book as part of their textbooks. Because it was written as a - not a dry kind of textbook, because it's got my sick kind of sense of humour throughout it. But it was written as a book where you could go through exercises and do things. That was my second book. So I was great. Got the book done. And then I realised in my classes that I was always making the students create these design canvases. So you probably know, I'm a big, big, big believer in design thinking. And I think design thinking and data science go like this. And I think design thinking is the secret sauce for any organisation. And so I had created this concept, this methodology called ''The Art of Thinking Like a Data Scientist''. And it always started with class, with getting a bunch of flip charts and writing things down, and then, you know, walking through the class and handed up PDFs, my slides, I said ''This is crap''. So I wrote a third book called ''The Art of Thinking Like a Data Scientist''. Unlike my other books, I self published. And the reason why I self publish is because the minute you publish a book, you lose all distribution and pricing control. They're gonna price it where they want. Yeah, you can get group discounts. But I didn't want a book is going to cost 35 or 40 bucks. I wanted a book that was a handbook and it folded out like a handbook, and you could use it. So I created an e-book. It's on my site deanofbigdata.com. I charge $9.95 for it. So people still have - students still have money for beer and pizza afterwards. But I wanted to have a real cheap price. And I've done some site licences for a couple of companies who have bought, - you know, everybody a company gets one, they pay me, you know, a couple 1000, 3000, 4000 bucks. Everybody gets a copy of it. I'm not gonna get rich from the book. But I wanted a book that students and businesses could walk through as they went through the exercise. So it was an eBook, a handbook that complemented the ''Big Data MBA''. So that was my third book. Comic book was actually my fourth book. So I thought ''Eh, let's do something kind of fun''. And it kind of took a life of its own. It ended being a lot harder than I thought. I thought I'd just crank this thing out and be done with it. My graphics skills really suck. So it took a while to get that one up. But it was kind of fun. But my fifth book ''The Economics of Data, Analytics, and Digital Transformation'' was really driven by the pandemic. I was so frustrated by the decisions that countries and governments and organisations were making. They were making blanket decisions based on generalised data. They were making decisions based on averages, not on, you know, precision insights. And we did everything from a data and analytics perspective wrong with the pandemic. Everything we did from a data analytics perspective is wrong. And it just pissed me off. And so if you read the preface of my book, the opening paragraph is me just doing a complete rant on why I was so mad. And so, in that book I actually - I kicked out really fast because I was so furious. I was so pissed and just everything sort of came together. And I got eight chapters done felt pretty good. Really talk about the the economics of data and analytics, how we need to think about a different, how do we need to make better decisions, how we need to reframe the conversation, how economics is a very powerful enabler, etc, blah, blah, blah. And I realised at the end, I said ''There's, there's a chapter missing''. This is great. Talks about, you know, economics and data and analytics. Weaves in a lot of design concepts. But it didn't really talk about culture and creating an empowering culture. And so chapter nine, which actually ended up being my favourite chapter. I'm not sure if it's my best chapter, but it's my favourite chapter in the book. It was really about: What are we doing to empowering the humans? We think about AI and ML. They're great at optimising those decision, very tactical decisions on the forefront. AI and ML are great. Data scientists use those tools and they can optimise decisions, right. They can optimise the cow path. What if you need to reinvent it? The reinvention is not going to be found in historical data. You need to be able to empower people. You need to unleash their natural curiosity that fuels creativity, that fuels innovation. And so chapter nine was like ''Folks, all the AI, ML, economics stuff is really nice. But if you don't empower people, you're not going to be successful. You're not gonna be able to scale it. Not gonna be able to sustain it''. Because in the end, it really is all about the people. That was a really long answer. Sorry about that.
Jonas Christensen 19:00
It was a great answer. And I think I am correct in saying that your last book was published in November 2020. You would have written that very quickly, because you would have gotten angry about the misuse of data around March or April that year, and you've pumped out a book then in six months, which is pretty impressive for a book of 220 pages or whatever it is.
Bill Schmarzo 19:24
It's just I was mad as hell. I wake up every day, just pissed that I was locked into a room. Again it was just like: I could understand why organisations were making decisions, if this was the 1980s. I couldn't understand why we're making this decision given we're in the 2020s. It was like everything about it was just maybe - made my stomach turn, So yeah, kicked it out really fast. So get me mad and I write really fast.
Jonas Christensen 19:50
That's it? Nothing like anger to fuel some progress.
Bill Schmarzo 19:53
Jonas Christensen 19:57
That is actually how most inventions come about. Someone's annoyed with the status quo and the book is an invention of sorts or an innovation, you may say. That's a great way to fuel yourself, really.
Bill Schmarzo 20:11
Jonas Christensen 20:13
Now, Bill, in this book, you talk about how it is actually CEOs mandate to become value driven, not data driven. You've already touched on this. We talked a bit about the questions versus the decisions. So let's dive right into that now and really elaborate on that. What do you mean when the CEO needs to be value driven, not data driven? And how does that relate to your decisions versus questions?
Bill Schmarzo 20:38
Okay, so if we are going to create organisations that know how to use data and analytics to drive value, then there must be a mandate from the CEO. This has got to be a pincer effect. So first off, it needs to be a mandate from the CEO that says 'We're going to instrument and measure all the decisions we make. We're going to actually measure our decision effectiveness. We're going to do that right up front''. If you don't have that mandate, and it becomes optional, then the only people who participate are those who probably don't need help in the first place. Right? So it takes this executive mandate that says ''We are going to use data and analytics to make better decisions, and we're going to measure the effectiveness, and we're going to change how we pay people''. So John Smale, who was the CEO of Procter & Gamble back when I was working with them back in 1980s, used to always say ''You are what you measure, but you measure what you reward''. Think about that. You are what you measure. So if we are putting measurements in place around the decision effectiveness and we have the KPIs and metrics that tell us how effective we are, that tells us a lot about the organisation. In fact, I'll tell you right now that - let me go the next step. So we are what you measure, but what you really measure are the things you pay people for. So if you're paying people for profitability and that's your only measurement, when things get tough, just lay people off. Screw it. That'll drive up profitability. So if your only measure is profitability, then employees, customers, partners, environment, society, all the other variables that drive value, right, get pushed aside. So if you're serious, for example, if we're serious as a country around diversity, it'd be worked into people's comp plans. If we were serious about the environment, it would be worked into people's comp plans. If we want to change the world, you want the world to change how we think about the environmental situation, change how they get paid. Anyway so, you first need that mandate that we're going to measure. We're going to define the KPIs and metrics that measure success, and we're going to change the reward system. That's great. But at the same time, you have to start building - I'm gonna call a Catalyst Network. A catalyst of people who are inspired to make change. Don't know why they're inspired. Maybe they just have - maybe they're pissed. Maybe they see an opportunity. Maybe they're just good people and want to make the changes. But you can't mandate that everybody in the organisation do it. Because a lot of people are going to lock their arms and say ''Uh uh''. You just find those people who want to do it and you slowly build that support. This is kind of one of the hearts of design thinking and cultural transformation. It's how you start empowering people at the lower levels of the organisation. So you create this pincer effect with the CEO, who's mandating ''We're going to measure decisions and here's the KPIs''. And here is the catalyst network who's actually trying to put in place the recommendations a CEO is talking about. So that's why I think that the CEO needs to become - the conversation must be around value. But that also means as an organisation, you need to have a thoughtful conversation and broadly communicate: How does your organisation create value and what are the KPIs and metrics against which we're going to measure value creation effectiveness? If you don't do that, you will never be value driven. Let me give you a little story. I'm sure every head of a data science team can share this story with you, right? So I had my data science team, and one of our executives came to me and says ''Hey, I just acquired this dataset. Tell me what's valuable in the dataset?'' And of course, my answer back was ''You tell me what's valuable. What's valuable to you and your organisation? Then I'll tell you what's valuable in the dataset. If you can't tell me what's valuable, I can't distinguish signal from noise in the data''. If you think about a point of sale system, for example: You know, at a retail store, information about the person running the checkout, the clerk there, right, isn't very useful. I'm trying to do customer acquisition and retention and stuff, right? Not really important. That information is noise for the signal about customer acquisition. But if my use case is around satisfaction and productivity of my clerks, that becomes a signal and information about my customers becomes the noise. So if I don't know what it is you're trying to do, if I don't know how you're going to measure effectiveness, I have no way to distinguish signal from noise in the data.
Jonas Christensen 25:02
So how does that relate to a question versus a decision?
Bill Schmarzo 25:08
Yes. So first of, let's clarify the difference between question and decision and when to use what. So questions are great for provoking thought. For getting people to think outside the box. What if we did this, right? Ask a question about this. You know, asking people questions, and probing is a great way to get people thinking, right. But ultimately, I need to take an action. And the beauty of decisions: Decisions are actionable. Questions may not be. I can ask you lots of questions about what you think about these customers, what you think about these programmes. That's nice to know. But ultimately, I've got to boil it down to a decision. Which programmes are we going to fund? Which ones am I going to cut off? Which customers... I mean, think about all the decision you have to make. So first four things about decisions that I love. Four things about decisions. First off, like I said, before, they're actionable. The fact that I'm going to make a decision is actionable. It's taking an action. The second thing is every business stakeholder I've ever talked to in my 40+ years of industry knows what decisions they're trying to make. Because they're already trying to make them. You know, think about: What price should this be? What products should I fix? You think about all the decisions they're making. Right now, they're making decisions based on a coin flip and some gut and some natural intuition. Not all bad, by the way, but they're making decisions today. So number one is action. Well, number two: Every stakeholder knows a decision or trying to make. Number three: I can attribute value to making better decisions. Improving my decisions around inventory management, how to stock in a store, reducing hospital acquired infections, right. Who's most likely to catch a staph infection? Making decisions, right. So if I can improve that decision, it's quantifiable. There's a value attached to it. And the fourth thing is that data science teams know how to optimise decisions. Decisions is the lowest level of value in the stack and the data scientists know how to optimise decisions. ''Oh, you're trying to improve customer retention?'' Boom, off they go. They're gonna bring in datasets. They are going to try this and that. ''Oh, you're trying to reduce unplanned hospital readmissions?''. Boom, off, they go. ''Oh, you're trying to figure out, do some predictive maintenance''. Boom, right. They know how to optimise decisions. So decisions provide this very natural linkage point between the business stakeholders who know the decision they're trying to make, understand the value of those decisions and the data science team, we have all these great tools and data tools and analytic tools and know how to optimise those. Decisions is the easiest thing in the world. And it's, again, for the reason I just said, piece of cake. But different than questions. Both have their role, but they are different.
Jonas Christensen 27:48
Great. I'm going to come in on this, Bill. Because you provoked a few thoughts in me and it's going to end up with a question at the end. So what you're talking about is really why I've started this show. It's called Leaders of Analytics because as analytics leaders, we are not just people who can get data out of databases and tell you facts. We actually have the opportunity to really shape and change the paths of our organisations in the right way. And that is actually a huge leadership responsibility. So all the reports that you might generate, the machine learning models that you might operationalize and so on, that's the technical operational bit of the job. But before that, there is a opportunity and therefore responsibility to move the organisation towards asking the right questions. And you're talking about here. See, I use the word ''questions''. So it's a questions pertaining to value decisions, right?
Bill Schmarzo 28:47
Jonas Christensen 28:48
They are questions because they're interesting. That is the challenge of many analytics teams. That they get out of the 2010 approach to analytics, which is ''Here's our PowerPoint presentation full of facts. Now business, you can go and do what you need to do. We've told you how the world is and you can go and make decisions'' to actually co-creating with the business, right. And you talked about design thinking a bit. And one of the elements of design thinking that we use a lot increasingly and our team is forcing ourselves more and more to do it, to actually bring the design thinking into the analytics process, is user stories, right? So a user story is framed as an X, I need Y so that I can do Z. So that means the end of it is actually your decision. So as a product manager, I need to know what are the customers that are most likely to buy this type of product, so that I can make sure that the value is right for them in the product. I'm making this up on the fly. That's sort of a poorly constructed user story, but...
Bill Schmarzo 29:54
I love this. Yeah.
Jonas Christensen 29:55
But the point is, it's the ''so that...'' which is the decision, right? So, here's the question, Bill. How do analytics teams get good at this stuff? What are the behaviours and habits and skills that they necessarily need to create within themselves as individuals and as a team and organism that works with the business to make this happen?
Bill Schmarzo 30:19
Great skill? Great question. And you raise up lots of points I want to make. Let me talk about the skills one. When I think about the best data scientists I've ever worked with, the one skill that's jumped out at me time and time again, is they're humble. They are humble people. They're collaborative, right, by their nature. They're very quick to give credit to other people, that they themselves should take the credit. My lead data scientist in my last two jobs, Waylynn, - I'll say his name - was the most humble person I've ever met. And every time he came up with a patent application. I think, six or seven times. He always included my name on it, because he said ''Bill, this was your idea to start with''. I said ''Yeah, I had the idea. But you made it happen''. He said ''Yeah, but you had the idea''. And I found when I worked with him, I wanted to work with him more. And I've had a lot of people like this. So I've been very blessed in my life. I've worked with people who are very humble, who are very quick to give credit to other people. And when they do that, you want to contribute more. You give more, they give more. So I think the number one skill is you have to be humble. You have to be willing to listen and be willing to collaborate. And you have to be willing to unlearn. It's to realise that you may not know everything. In fact, I personally hate the term ''expert'', because an expert infers that you already know all the answers, and there's no experts in our space. Let's be honest, there are no experts. Because we don't know how this game ends. We don't know where the story is going to end up, right. So there are no experts. So we have to be willing to unlearn and let go, sometimes. We got to take ideas. And it's like climbing a ladder, right. You climb a ladder at some point, you got to let go that rung below you because it's not true anymore and you got to let go. And it's scary. There's a moment in time when your hand - you only have one hand hanging onto the ladder and it's rattling like this, right. Put your second hand up there, but you can't be afraid to unlearn. So again, that to me is the key things. Humble, collaborative, very willing to unlearn and learn from other people.
Jonas Christensen 32:19
Great tips. So often, when I ask questions in Israel, no one really talks about the technical aspects of what people need to know. They don't answer. ''Oh, you need to know these Python packages and you need to be able to have such proficiency in R'' or ''You need to use this or that system'' or ''Understand how a star schema is constructed'' and all that stuff. It's often or not often, it's always interpersonal skill, and interpersonal relationship building. Could you elaborate on why that is so key for analysts?
Bill Schmarzo 32:55
I'm teaching a class this next semester on how to future proof your career. I think there's three things that everybody needs to master, if they want to survive in a world that's constantly under transformation from a technology perspective, from a social perspective, from an environmental perspective, from any - We're just being bombarded, right. So number one: You need to be able to talk about economics and economics is the branch of knowledge about the creation and distribution of value. So economics is about value creation. By the way, value is not just financial. There are other dimensions of value. Customers, employees, partners, society, environmental, ethical. So you need to understand how value is created because that's the language of business and that's how you're going to be able to converse with CEOs on that level. The second thing is you need to understand data and analytics, data management, data science, data engineering, and what you can do with it, not how it works. Right? I don't expect you to have to learn Python, or learn TensorFlow or learn something, but you better know: What is it that I can do with it? What do those tools helped me do? So then when I get into an ideation situation, when I'm sitting between a data science team who really knows this stuff and the business stakeholders, I always see myself as that bridge. Can I help ideate to show them what's the realm of what's possible? We can do with neural networks to solve this problem or reinforcement learning to solve this problem or an unsupervised - So I think you need to understand what analytics can do and what are some of the uniqueness of data.Especially from an economics perspective, that I can take advantage. So there's knowledge there. And the third skill is design thinking, because I think it's up to every one of us to help unleash the greatness in each of our team members. It isn't just the boss, whose job is to unleash the greatness. It's my job as a member of a team to help unleash the greatness. I work with a great team right now, just phenomenal people and they teach me a ton every day. And I feel like my job is to bring things to them. Provocative, make them feel uncomfortable. You know, maybe sway their thinking forward. But I'm doing that because I I'm seeking to help unleash their greatness. Everybody has a natural greatness into them. So I think design thinking is one of those skillsets, that not only does design thinking help us think about how do customers create value, but how do we ideate around those value creation? So it isn't just about optimising, like I said earlier, right? I can use AI and ML to optimise. Sometimes, I need to reinvent. And that's where the whole design thinking comes in and this idea of empowerment. Makes sense?
Jonas Christensen 35:28
Yeah, it does. And the design thinking part, for me, it's really something to guide you as you structure your thinking. But it's also to actually - this my interpretation - it's kind of to break the mould a bit and get you out of your chair, because you have to interact with the rest of the business. We're not just there to answer questions. We're there to actually shape the organisation. And if you want to be a really impactful analyst or a leader in an organisation, you have to do that and feel that discomfort that at times that brings about.
Bill Schmarzo 36:03
Jonas, you said something earlier on and you didn't build on it. This idea around co-creation. I think co creation is a very powerful way, a powerful concept and a powerful approach. And sometimes you co-create - fact, you co-create with your customers, right. How do you co create with customers? I love to co-creation process. And here's something kind of weird. I love to co-create with people who don't like me, who don't believe what I say, who think I'm full of BS, who don't share the same beliefs as me. Because if there is a friction in the organisation, it isn't to be avoided. It's to be embraced. To understand, why do they feel the way they do? So I'll give you an example. So, we had an election, you know, two years ago, a presidential election, and it was a very divisive election. And there were some people who felt very strong.
Jonas Christensen 36:55
I've heard of that. Yeah, there was a...
Bill Schmarzo 36:57
There were some people who carried arms and stuff on January 6th. So, they didn't like the result. But people voted for who they voted for, sometimes out of emotional stuff. But if you drill down into them and you started asking them ''What was your rationale? Why did you support that person?''. What you'll get is some really valuable nuggets about it, right. ''Well, I liked the fact that he was an outsider''. ''I liked that he was charismatic''. ''I liked the fact that he was a strong individual''. There were characteristics in those - when you had those conversations, you go ''That's good to know'', right. Yeah, there are other things about that person that may not have been ethical or whatever else. There were nuggets there. And if you have a conflict with somebody and you pause ''Tell me, Help me understand your rationale for position'' and drill down. Once you can find those individual nuggets, you can, you might be able to combine those and blend them together to create something even more powerful. I believe that friction and conflict is how you drive innovation. If we all felt the same way and all thought the same way, we would never innovate. It's only when we have those colliding heads. Now you you can't let people have pitchforks and throw them at each other. You need to have an arena that says ''Okay, I'm gonna listen to what you have to say. I'm going to hear you out. I may not agree with 80% of what you say. But that 20% I do agree with''. By the way, it's the other way around typically. It's usually the 20% you don't agree with, which the stuff you see and hear. It's the 80% underneath the base of the iceberg, where you're like ''Wow, you and I are both seeking the same thing, right? We're seeking these kinds of characteristics. We're seeking these kinds of outcomes. We're seeking these kinds of results''.
Jonas Christensen 38:32
Yeah, it's a really interesting point, because you talked about earlier - we were talking about analytics teams here and how they behave as individuals and as an organism. You talked about people being humble and seeking knowledge and all that stuff. The archetype we're describing there of someone who works in analytics is someone who doesn't typically - and I'm painting broad-brush here - who doesn't typically lean into conflict. When you don't agree on something, it can feel like conflict and it's actually a really important point that I want to make to listeners that you have to lean into conflict in your career. It's just a really uncomfortable, unavoidable thing that we have to do. The more you do it, the better you get at it. And I think your Jedi mind trick that you're playing on yourself, Bill, is exactly the way to do it, which is: How do we become comfortable with that? How do we see the perceived conflict as a sign of positive momentum? That's actually the difference. It's not about conflict for sake of conflict. It's to drive that positive momentum for the organisation. You made me think of a project that we did a couple of years ago in an organisation that I worked in, where we had this quite valuable analytics opportunity to give hundreds of people information on a daily basis that they could do to perform their individual jobs very effectively. At a low level, they were sort of flying blind as individuals in the organisation. And we started the usual way with the ''Oh, let's make a prototype of this tool using the data'' and what we realised really quickly was the first thing we needed to do was actually to get everyone in that business on board with what we did. And we set up an agile project. We got stakeholders to take ownership in the project, so that we could roll it out. And the first three months or so was just getting the business on board with our vision and to sort of uplift their analytics literacy to a place where we could talk the same language about the solution. And we're a bit frustrated with that process. ''Can we just get to the building?'' throughout it, but once we've gone through it, everyone, including ourselves, not least, had learned so much. And they had changed, I had changed as analysts and leaders, and we'd seen firsthand ourselves how we could impact the business. So for those analysts, it was a fork in the road, their career and for the business as well. We became more data driven, not just because we delivered the project, but because the approach to getting there.
Bill Schmarzo 40:54
That to me - your story is perfect. It's that we know data and analytics so well, that we want to jump right into solution mode and we haven't given people a voice in the solution. We've laid it on them. And if we don't give people a voice, not only is it not likely they're going to adopt what we do but maybe more importantly, the model we build will be less effective. If we think about, for example, in the data science process, Feature Engineering, identifying features and such to make the ML models more effective, my experience is that the people who have the best insights into that feature engineering process, aren't the data scientist. They are the people at the frontline of the organisation. They're the people who are working with the nurses, the doctors, the teachers, the technicians, the operators, right. It's the people at the frontlines of customer engagement and operational execution. If you can bring them in, if you can empower their intuition, you will build better models. And as a result, because they're part of the process, they're likely to own the resulting solution.
Jonas Christensen 42:01
Great, you're actually making me change tack a bit here, Bill, because I wouldn't mind talking to you about data management. I think it is very, very important for the next part of our evolution as data analytics leaders, but also the whole field of data science. Actually so much so that I'm writing a book about data centric machine learning at the moment, because that is about analysts and data scientists becoming more involved in the data engineering aspects of the pipeline or the journey of data through an organisation, from raw material to the decision. So, Bill, earlier in my career, I used to think ''Data management, it's kind of boring. You guys over here. Can you just put it in the data warehouse for me? Put it into tables and I'll analyse it''. And I can see now that was a gross stupidity, really, and...
Bill Schmarzo 42:58
You didn't know any better at the point. Now you are smarter.
Jonas Christensen 43:00
Very ignorant approach. You have said that you called, I suppose, data management, one of the single most important business disciplines in the 21st century. So you obviously, in a very different camp to who I used to be. I'm on board with that now. But can you elaborate on why that is?
Bill Schmarzo 43:21
Yeah. So if we believe that AI and ML are generational capabilities, that have huge amounts of value to organisation, and we've, - if you've read Gartner or PwC or any of these analyst firms out there, they're all talking about the economic impact that AI and ML is going to have on society. Okay, it's great. Sounds too good to be true. We're talking about autonomous cars and self learning health programmes and etc, etc. What we also know that AI and ML, the results that they produce are only as good as the data that feeds them. I think Andrew Yang, who's sort of the modern godfather of AI, a Stanford professor, came out recently and said ''You know, instead of spending so much time trying to fine tune the algorithms, maybe we need to fix the data''. Because algorithms won't overcome bad data, incomplete data, and a low latency. Look, I mean, there's so much stuff on the data side that we need to get right, if it's going to become the fuel that feeds AI, that is the fuel that's going to feed business innovation. It's kind of fun to get the slide here, right? We think about the role of data management is to make sure we've got the right high quality, low latency, low granularity, enrich data that fuels the data science process, that looks for those customer product service and operational Insights and we're going to basically drive around business innovation that feeds back around here. So data management provides us fundamental foundational capability that is needed to fuel that. But to be successful, the data management process doesn't actually start with data management. It starts by understanding: How does the organisation create value? What are the KPIs and metrics? And the data science team has to go through a process to figure out ''Well, what are the data elements, the calculations and the variables in the algorithm we need to do to actually figure out that value?''. And then data management has to provide the data, the quality of data, the management of data, to be able to access it, all the things required and data to do that. So while Data Management supports like this, the value creation process goes this way. That's a huge disconnect. Because we always think about data management as a series of features and functions. And no, we just talked about design thinking. I think you've seen, I did a blog on the data management candyland, journey map, right. I created a journey map that really says that this is actually a journey. It's a journey from a business need to a business outcome, driven by data and analytics. And data management needs to be this foundational capability that enables data science, enables AI. So to me, for an organisation that believes that AI and ML is the source of untold amount of value, we need business leaders to think about data management, not as a series of features and functions, but as a series of outcomes. Not outputs, but outcomes. And how do I drive outcomes? How do I accelerate that process from a business need to a business outcome, learning from that and cycling back through again? So to me, it isn't I'm going to have business leaders who are going to be cleansing data. But they are certainly the leaders that tell me which of my data sets are most valuable, and what I can do to make that even more valuable, given the kinds of decisions I'm trying to drive.
Jonas Christensen 46:41
And all this is very logical that business should help firm up what the data should be. But data science teams should also validate the dose, the way that we capture that information and those pieces of information. Yeah, features can become features in a model, and so on, that are actually captured in the right format, and all the technical aspects of it. It's totally logical. How many organisations actually have that habit and approach right now? it's very few that actually, fundamentally connect business and data well, and through the data management discipline, the data science discipline, and the operational discipline of how that data is collected, stored, etc, validated.
Bill Schmarzo 47:25
So now we've come full circle. I love this. I love ending here. Because the reason why there's not more organisations that have done this, it is our fault. It is our fault. Because we start every stinking conversation by talking about data, when our stakeholders, they care about value creation. So instead of talking about ''Here's what we're gonna do for you''. Like, it's like walking into the hospital or to a doctor and doctors just say ''Well, I'm gonna fix your toes, and I'm gonna do this with your ankles. I'm gonna do this and etc''. Like, ''Sorry, I got cancer. I'm just sent a sample and I got cancer''. You know, taking care of my ingrown toenails and my flat heels, you know, doesn't matter. We have to diagnose before we prescribe. But yet we go right into the prescription mode, right. We're going to tell you what you need to do. ''I don't care what your problem is. Here's what you need''. So it's on us, because we always start the conversation talking about data, and pardon my bluntness, but no one gives a crap about data. You know this idea and I always tease Doug Laney on this, right. The three V's of big data: volume, variety and velocity, right? It's great, really great concept, right? Volume, variety and velocity. Not a single business user I've ever talked to gives a poop about the three V's of big data. What they're focused on are the four M's of big data: Make Me More Money
Jonas Christensen 48:48
Haha, I guess.
Bill Schmarzo 48:50
So it's on us, we focus on the wrong things. So we need as a discipline in data science to not start the conversations talking about data but instead let's frame it by talking about value. Understanding what the data science team is going to need to do to be able to create that value and then how we provide the data in the right formats, in the right location, whether it's a data warehouse, a data lake, data lake house, data mesh, data fabric, data Bs. Who cares, right? That we provide this capabilities in light of what's trying to happen here?
Jonas Christensen 49:25
Yes. And it's actually a new paradigm we have to step into, both in the way we function operationally, but also the way that we think about these problems, right. So to do this, in practice, the business has to be involved and the business has to take ownership of that same data as well. They are the original creators of it. So I always say to my business stakeholders ''We as a data science team, if we were a bakery, we'd make the bread and our data management team, they'll mill the flour and all that stuff. But you guys, you eat the bread at the end, but you're also the farmers who are out there fertilising the grants, making sure that we harvest at the right time of the year. And there's no mouldy crops, no myosin in the fields and all that stuff'', right. The data has to be accurate, otherwise, the bread will not exist in the first place. And that ownership has to come across the organisation through that collaboration and something that we talked about
Bill Schmarzo 50:25
That co-creation, right? We all have to own it. Everybody in this cycle up here has to own value creation. We all play different roles in that value creation process but it is around - I call it the ''Unleashing Economic Value Of Our Data'', that is everybody's job. It isn't just their job or their job or their job. It's everybody's job, which is why I keep coming back saying ''It's a business discipline''. It's a business discipline. Just like marketing is a business discipline, just like logistics, just like inventory management. Data Management is a business discipline.
Jonas Christensen 51:02
Hi there, dear listener, I just want to quickly let you know that I have recently published a book with six other authors, called 'Demystifying AI For The Enterprise, A Playbook For Digital Transformation''. If you'd like to learn more about the book, then head over to www.leadersofanalytics.com/ai. Now back to the show.
Bill, are there any data management principles and processes that need to be rethought then and or adjusted to this world that wants to take advantage of machine learning and AI?
Bill Schmarzo 51:43
There's a lot of things that we need to - if we're going to make business management, I'm sorry, data management a business discipline, then you think about what comprises a discipline. Well, you need theorems. You need laws. You need frameworks. You need underlying products and tools. And so we don't have a lot of those things yet. You know, I just wrote a blog this week that talked about data products is going to be a very critical output from this data management cycle. We're going to create data products that are going to continuously learn and adapt. But we need to be thinking about data subassemblies, that the data craftsmen can use to build these data products more quickly. Right now, data is just scattered out there. You know, I want to bring together data, its metadata, its access methods, its governance, everything should be packaged together. So it's easier for the data craftsmen or crafts people to able to bring these things together. So there's a lot of work we need to do in this space yet but we're moving in the right direction. We've got theorems. I wrote a whole blog about theorems. In fact, I even created a set of playing cards that has the data management theorems in them, right. There, you got playing cards, right. So, we can talk about all the things that we need to do and even have fun by saying ''Hey, do you do this very well? Do you do this very well?''. And so we're starting to head to this direction. And there's a lot of people out there who are all locking arms and trying to make all this happen. So I think it'll happen. Hope it happens before I pass away, because I ain't got no time for green bananas. But I do think as an industry, more and more people, like yourself are not only feeling a sense of responsibility to help drive data management as a business discipline, but a sense of obligation. That ''With great power, comes great responsibility'' to quote Uncle Ben from Spider Man. And we have great powers. Our data science team and our data management teams, our data engineers, they're the modern superheroes, but we don't do enough. And it's up to us. It's an obligation. It's our obligation to take this next step.
Jonas Christensen 53:47
Great. Bill, we're kind of at the end here. So that was a great final statement. You didn't plan it to be that but it was kind of that. And I have a couple of questions before we round off.
Bill Schmarzo 53:58
Jonas Christensen 53:59
One is one that I always ask for guests on the show. And that is:
Bill Schmarzo 54:03
Jonas Christensen 54:04
To pay it forward by recommending who you think should be the next guest on Leaders of Analytics and why?
Bill Schmarzo 54:09
Oh my god. Well, have you had Kirk Borne on yet?
Jonas Christensen 54:12
Yes. He was one of the early ones.
Bill Schmarzo 54:13
Yeah. Kirk is phenomenal. John Thompson also.
Jonas Christensen 54:17
Yep, he recommended you, so there you go.
Bill Schmarzo 54:19
We all pay each other. John Cook. Malcolm Harker. These are people who I've had some just incredible conversations. Sunil Gupta, another person. But these are people - Kirk and John, I've met in person. John I've worked together. But a lot of these people and names I'm giving you are people I've never met. Only on social media. And their sense of contribution is great. Mark Stross, another great person, Marks Stross. So if you send me a note, I will introduce you to some of these people because they're also great thinkers in this space. And I find that when I talk to them, when I engage with them on LinkedIn, I always learned so much from them.
Jonas Christensen 55:00
Brilliant. That is definitely what we're going to do right after this show. Now, Bill, last question. Where can people find out more about you, get a hold of your books and your playing cards?
Bill Schmarzo 55:12
I hang around LinkedIn. I write a blog every week. I publish it on Data Science Central. That to me is kind of the watercooler for the data scientists and data engineers. I post it on LinkedIn and there's usually a pretty good conversation that happens from that, because I'm very fortunate. I have a lot of really, - a lot of my friends are smarter than me and they can take ideas and they can add them, they enhance them. So if you really want to get involved in this space, follow me not because you're following me, but because you're following a community that includes me. And there are some brilliant people in that community who are throwing out all kinds of really great ideas, who aren't afraid to be wrong, who aren't afraid to say something. Because we're all trying to solve this problem together. #Bettertogether.
Jonas Christensen 55:54
#Brilliantrecommendations. Bill, it's been such a pleasure to have you on the show. I have really enjoyed this conversation. And you have made an enormous contribution to this field. And by doing that, not just to the field, but to the business world in general, through your interactions in businesses, but also your teachings and education, by books and in educational institutions. So on behalf of everyone who has been exposed, including the listeners today, thank you so much for your contribution and all the best for your future.
Bill Schmarzo 56:29
Jonas, thank you very much. Fun conversation. Went all over the place. Those are always the best kind.