A guide to big data

Hello and welcome to X Tech. I’m Debbie Forster MBE. I’m the CEO at Tech Talent Charter and an advocate and campaigner for diversity, inclusion and innovation in the tech industry. I’m delighted to be working with Fox Agency as the host of the all new X Tech podcast, and as a curator for the X Tech community. So, today I’m joined by Bill Schmarzo. He is the Customer Data Innovation Lead from Dell. Bill, thank you so much for joining me today.

Thanks, Debbie, for having me.

So, Bill, one of the things I love to do is for our listeners, it’s fascinating because we get to tech by different routes, many of us. Some of us born techies, others find their way by more roundabout methods. So could you tell me what was your path into tech? How did you find yourself here today?

So Debbie, I would say I’m one of those people who was born tech but then moved out of tech.

Oh, you were escapee and a returner?

I’m escapee. Yeah, so I mean, I’ve always been a tech person. I’ve always been interested in data, especially analytics. I got started in analytics when I was very young. I used to play a game called Strat-O-Matic Baseball, which was all about analytics and the forerunner to what we call sabermetrics or Moneyball stuff. I’ve always been in the area of data and analytics. I’ve always been intrigued by what we can do from the insights buried in the data to help make better decisions. I think what’s happened over the last probably 10 years, Debbie, is my transition from a technologist to an economist and understanding how we as organizations in society, first and foremost, how do we create value and how do we define value creation effectiveness? Then my technology background, especially around data and analytics, has a frame around which we can now have a conversation.

So did you go straight into the field pounding away at keyboards, or what else happened along the way?

Yeah, I got started. My undergrad degrees were in math, computer science and business administration. I got my MBA in information systems. Went from there to Arthur Andersen where I was coding mostly database routines. I built a database I/O module for our projects that we would use to optimize how we were querying databases to pull data off. So I started with tech, been with tech and more importantly, I’ve been with data because for the longest time, let’s be honest, Debbie, data wasn’t very cool.

Now Bill, I wasn’t going to be the person to break that to you, but yeah, it wouldn’t have been the best opening line in past times. Nowadays it’s hot, but back in the day, not so much.

Yeah. My kids think I’m cool now, but for the longest time I was just, data was the byproduct of all the operational systems. Everybody was concerned with ERP and operational systems. We got to dashboards and reports that basically told us how the operational systems were doing, but we really weren’t doing much in the area of really analytics. Yeah, I was an outcast.

But one of the things I was interested when I was looking across your career is rather than just doing, you’ve also been writing books along the way, so how did that start happening?

I’ve always been compelled to share what I’ve learned. It goes back to the influence my mom had on me. I’ve been very fortunate, very blessed in my life to have been in many different situations. I call them Forrest Gump moments, right place, right time. Not because I’m smart or tall or good looking or from Iowa. Sometimes in life you just get lucky, and I’ve had a lot of lucky moments where I was at the right place, right time, so I’ve always felt this need to share what I’ve learned. I started off by writing white papers for organizations, teaching. I like to teach. I taught at the Data Wealth Institute for a long time, and that eventually evolved into writing a book.

Actually, the story behind the book was, and I was at Dell, our Vice President of Marketing, her name was Barb Robidoux, knew that I was writing lots of blogs about big data because big data had just burst on the scene. I had a data background, so I was talking about it. She read all my blogs. She was a big supporter and she says, “Schmarz, you got to write a book. Come on dude. You got all this content, write a book.” She was really the motivation for writing my first book. Then the book led to some more teaching opportunities, especially at University of San Francisco where I met [inaudible 00:04:17] who was a professor there. He had me team teach with him, which ended up me being more teaching, so it all just cascaded around this idea that I’m in these situations. I’m working with some of the best customers in the world who have all these great ideas. What am I doing to take what they’re teaching me and bringing that out to the rest of the world?

I think that’s what distinguishes your books and your role because quite often if people are going from doing the white papers into books, they’re moving more into academia, so it’s clever people talking to clever people about how clever they are. Your books have a very different vibe in that respect. This is really how-to approaches on things, isn’t it?

Yeah, they really are how-to books because for most of my life I have been on the consulting side of the shop, so I’ve been working with customers trying to figure out how do we develop a data strategy? How do we build an architecture that supports our ability to get at data? What about data management? What about data governance? All these things, so as a consultant working with these customers, I had to do this stuff. So everything that’s in my books, everything, are things that I have done personally with customers. Again, it wouldn’t be very fun if I was doing them by myself, but again, I’ve had this opportunity to work with some really intelligent, bright, forward-thinking customers. Not always large customers. Some of my best and most innovative customers have been very small companies who really are trying to do things differently. Yeah, it’s a how-to book. You do this, you do this. Lots of design templates, how you do things. I share the design templates freely because I find that the more I share, the more I get back. People try them and say, “Hey, Schmarz. I tried this. It didn’t work very well. Here’s what I would suggest.” It constantly evolves and gets better.

So more and more conversations rather than just you spouting out information. Much more cyclical in that sense.

Yeah, actually I’m really not that smart. What I think I’m good at doing is listening and integrating and extrapolating, hearing what people are saying. If I hear three or four people say the same thing, I was like, that’s got to be more than coincidence. Then how do I solve that problem? So a process of methodology or framework. Somebody called me like I’m Mr. Framework. Yeah, it is. I have lots of frameworks that I use because that’s how we can help people become more effective at leveraging data and analytics

And frameworks are powerful because it’s not answers. Frameworks are adaptable. It’s input and output that become more valuable.

That is a marvelous point, Debbie. It’s not a checklist. It’s not railroad tracks. It’s guardrails and we’re going to let people bounce between those, but what works for one customer in the same insurance industry as customer number two may not work. Every company is different. I always like when a company says, “Well, you work with insurance companies. Tell us what we need.” I’m like, “No, every company is different. You all have a different culture. You all have different values. You all have different ways you create value for your customers. There is no cookie cutter.” The framework has provided these guidelines that people can explore and learn within those guidelines what’s most relevant to them.

For the audience, I think when we’re looking and growing people within our teams or looking to get people and teams, this is what we’re looking for. The people who want to develop or work within frameworks, don’t want an answer. Not our tick box people. Not how do I fit it into my spreadsheet and my set formula, but really looking at that human-centric. It’s business problems, isn’t it? It’s not business is here to serve tech. It’s very much the other way around.

Bingo. I think one of the reasons, Debbie, that I’ve gotten so much into economics the last decade is at the end of the day, it’s all about how do we create value? What are we doing with tech, data analytics, blockchain, quantum computing? Pick your favorite shiny silvery thing right now. It’s how do we create value from that? If we don’t as organizations understand how we create value, who our stakeholders are, how do they define value creation? What are the KPIs and metrics they’re going to use some measure effectiveness? If you don’t know that to start with, then you’ve got a really hard time. I’m going to say a really low probability of success by just throwing technology at problems.

Yeah, and that’s what happens. The people that don’t have it, it is about just give me that shiny thing and I’m going to fling it at the wall at the problem and hope something sticks. Things stick, but it’s not what we want sticking to that wall.

So let me give you a little story. So we’ve been running a bunch of research projects. I’ve been doing it for years, but formalized it in the last year and a half, looking at a customer’s journey from business need to business outcome using data and analytics. There’s a journey map we’ve created, a design thinking concept of what people go through, the outcomes they seek to go along that process. 90% of the people we talk to when we ask them, “Well, where’s the biggest problem on this,” it isn’t at the end of the journey. It’s at the beginning.

We poorly set up where we’re going to go. We don’t know exactly what we’re trying to do. We don’t know how we’re going to measure success. We don’t know enough about our stakeholders. It’s all the stuff upfront that has very little to do with technology and everything to do with customers, to do with culture and to do with economics. If you frame that up, then everything else falls in place, but like you said, we fall in. We get infatuated with like a neural network. “Wow, look what I can do with a neural network, a ChatGPT. I can ask it a question. It gives me an answer back. Wow, that’s really cool.”

If I had a pound for every time somebody has come to me and said, “I need a fill in blank, an app, a neural network.” You ask them, “Why? What do you want to do?” “I don’t know. I just know we need one. The bosses are we’ve got to get one.” Well, depending on which side of the transaction you’re on, that’s a good way to waste money or make money. I’m not sure which. So, that gold thread through what you’ve been doing is data. Let’s unpack one of these big shiny things that people are playing with. [inaudible 00:09:52] without hitting something about big data. What I’d love to know, it’s clearly here to stay. Can you unpack what we mean by big data? Then walk me through what you find most frustrating, most exciting, so let’s unpack one of these shiny things that people are throwing about right now.

That’s a good one. So first off, big data’s not about volume. It’s about granularity. So we’ve always had data, but we’ve always had data at this aggregated level. Sales within a store over the last week or number of incidents by carrier over the last month. We’ve always had data. What big data is about is not about volume. It’s about granularity. I can get down to individual data usage patterns. From those individual data usage patterns, I can create and identify their predicted behavioral performance propensities. This is a concept I call nanoeconomics, right? You got macroeconomics. You got micro. Big data introduces as concept of nanoeconomics where I can understand on an individual level whether it’s a human, a customer, a doctor, a nurse, a patient or a device, a wind turbine, a chiller, a compressor, whatever it might be. I can start capturing data at that individual granular level to start understanding their predicted behavioral performance propensities. That’s what big data is about, and that’s why it’s so much more powerful and scary than just regular data because I can get down to the individual level.

One of my sons said, “Dad, you’re Big Brother.” I said, “Yeah, I kind of am. That’s not what I wanted to be, but okay.” I can get down to that level, so what excites me, Debbie, about this is we can develop programs, policies, procedures and products that deliver individualized benefits. For example, hospitals could create individualized wellness curriculums just for you. Universities and colleges and high schools could create curriculums just for individual students knowing their predictive behavioral performance propensities and such. We can get down to that granular level to drive precision decisions that we could actually as a society do more with less because we have so much waste in what we do today. When we get down to that granular level, it has impact across education and housing and healthcare, and every part of society could benefit from this concept of individualized nanoeconomics. See, the flip side is that same data can be used for very nefarious reasons, too.

Let’s look at the good case scenario for a moment. Let’s say I’m a business, and I want to do that. That’s exactly what I want to do for my business. What’s stopping me? When you are working with companies that are beginning to really grapple with how they can use big data, what are they getting right? What are they getting wrong?

The first thing that they get wrong is they don’t thoroughly understand and triage how they create value and how they measure value creation effectiveness. It starts there. One of my favorite stories is having a client walk to me and say, “Hey Schmarz. I got this data set. Tell me what’s valuable in it.” Of course, my reply back to them is, “Well, what’s valuable to you? How do you define value?” What organizations struggle with I think is they want to jump into the data right away. Data management, data economics doesn’t start with data. It starts with value. There was a great study done by Tom Davenport and Randy Bean that was looking at companies that are trying to become data-driven. Shockingly, what they found out is over the past three years, the number of companies that are trying to become data-driven has declined, substantially in some cases.

You don’t want to be data-driven. Who gives a crap about data? You want to be value-driven, so if we can refocus this conversation on value, then all this money you’ve spent to capture data, to store it, to back it up, all this plethora of tools you’ve got, it’s like giving everybody a bunch of hammers without any blueprints. Just start knocking nails at this. I mean, come on. So we lacked that blueprint upfront around how we create value. When organizations do that, it’s printing money. Their eyes open. They start to see. It’s like they couldn’t see before. They had shades on their eyes. You just pull one and go, “Wow, this is actually pretty easy because now I know what I need to do, and I know what I don’t need to do.” Here comes this framework.

That is almost more powerful, what we don’t need to do, what we don’t need to keep. All right, what else is it? Let’s say I’ve bought me some data scientists, so I’m excited. I’m starting to figure out what value chain is.

Strike one.

I’ve got some data scientists.

Strike one.

Talk to me about why.

Don’t start with data scientists. You’re going to need data scientists, but if you really want to be effective as an organization driving value to your customers and your stakeholders or your constituents, the data scientists are important, but the people who are going to uncover those variables and metrics that might be better predictors of performance, the whole feature engineering component, the people who really understand what drives the business aren’t your data scientists. It’s your frontline people. It’s the front line of people who are engaging with your customers and your partners. It’s the front line of people who are running the operations, driving the trucks, running the theme parks, whatever. It’s the frontline, and the starting point has to be how do you empower those people and bring them into this ideation data science and visioning process?

When you empower the front lines and create a process that brings them into this area where I’m trying to use data and analytics to drive value, now you have a group of people who have intuition about what happens out in the field with customers in operations, with data scientists who know how to codify that. Now you’ve created this really powerful lock. It isn’t just a data science team trying to grab stuff. You’ve got the people who really know what’s going on and you latch them with the data scientists and data engineers. Now you create things that are not only more effective but are more relevant and will actually get used by the front lines because it was their idea in the first place. That’s the whole premise behind one of my favorite methodologies, which is the art of thinking like a data scientist. How do you bring everybody into the process? It isn’t just the data scientist.

I kept hearing you when you were talking about this. It’s a process. It’s a process. We do this, it goes on. I’m not hearing a nice tidy little project.

Wow. That’s a good observation. It’s not. What we have found is as organizations are moving through that data management journey from business need to business outcome, it’s a process they follow. It starts as a project, but it eventually evolves into a product if they’re successful as they go through this process. Of course, as they go through that journey, they’re learning certain things don’t work. They’re failing often, restarting back, coming back, but at a certain point, there’s an inflection point they realize, “We’ve got something. This model really works. It is more effective.” Now we got to turn that exploration project into a continuous learning and adapting product. We got to basically build a product that integrates AI and ML that can continuously learn and adapt. So you’re right. There is a project aspect at the front, but it’s only in support of what happens in the early part of that journey. Eventually what we see is customers realizing when they finally have something they know is valuable, turning that process or that project into a product. Does that make sense?

Okay, so let me reflect that back to you because the things that I’m hearing that I’m not sometimes hearing in tech businesses or in businesses trying to use tech and big data, exploration projects leading to a product, but you were most often talking about processes and journeys. All right? I think that’s a powerful point, so it is not we bring in and sprinkle a data scientist across something. This is a changing way of thinking on both data scientists and others in the team that that journey will become products, but it’s much more than just a program or a product. This is you use frequently that idea of journey or process.

Debbie, you nailed it. You’re spot on, and here’s the other key aspect of this. Your product is never done. Your journey is never done. If you are building products and a culture, not just products but a culture of continuous learning and adapting so that every customer interaction, every operational interaction, transaction is an opportunity to learn more because we are now in the age where the economies of learning are more powerful than the economies of scale. So that journey doesn’t start and stop. It basically keeps cycling. As an organization, your ability to accelerate that economies of learning process, is what’s going to distinguish winners from losers.

This is what I’m hearing again and again from guests on the show. This shifting away from just products and projects, whatever they’re working on, it’s around the people in the learning and to start thinking about not training but learning and seeing that as an investment, not a cost and the benefits that are coming through. I’ve not used that. I’m going to quote you again and again on the economies of learning, not just scale. I think is really, really powerful in that respect. Right. Okay, so big data. You’ve fixed that for me. Thank you very much. I’ll put that on. We’ve got that sorted. You mentioned that you’re writing another book. All right, so why? What’s led to this one and what’s boiling away in your head now?

It’s funny, Debbie, that each of my books have had something that’s really motivating me, usually through frustration. This book really is an attempt to try to open the doors to everybody about the power of data science, so I’m thinking how do we convert everybody into citizens of data science? Not citizen data scientists. I don’t need to turn everybody to a data scientist. I need to have people understand what can data science do. As a citizen, how do I do that? Underpinning the citizen of data science is a desperate need around data and AI ethics. We need to have a cultural awareness and transformation, so I have a AI and data literacy educational framework. There’s five stages in it, I think when I get done with it, that really these are the areas when we start talking about data, AI, literacy, we need to make sure we’re covering all these areas.

You need to understand how your personal data is being captured and used to influence you. You need to understand some of the basics of analytics. You need to understand how we make predictions and how we apply predictions, how we need to apply critical thinking. We need to understand how we build models. We need to understand the basics of ethics, so one of the things I’m working on right now is how do we codify ethics because if I can’t codify it, then I’m going to have a hard time putting it into my AI utility function, which means my AI models aren’t going to have an ethical foundation.

I actually believe, and I’ve got a series of blogs I’m testing with this, that you can actually create the economics of ethics. You can actually apply economic concepts to measure ethics. I’ve tested it with a few people, a few customers. It needs work, but it’s a step in the right direction because if we don’t figure out how to codify it and work it into our systems, then it is a difference between having Terminators who are wiping out humans or having a Yoda on our shoulder whispering good advice in our ears about, “You don’t want to do that. You might want to do this instead.” My motivation is given where I am in my career, this may be my last hurrah. Let’s make this a book that’s available for anybody. I’m not talking about executives. I’m talking about not just college students but high school students, maybe even middle schoolers who can understand that in this new world, here’s how I need to think about these things and understand where and how my data is being used and how can I be aware so I’m making more informed decisions in an imperfect world.

We were talking about shiny things in tech, and you’re seeing people and you’re seeing research coming out, seeing that there’s really people falling into these two camps. One that AI is the great thing that’s going to fix everything. Then there’s the other, as you say, the Terminator. AI is the end of everything and those are quite passive ways of viewing it. A lot of times when we’re talking about AI and data ethics, we are often talking about bringing in someone to save us or protect us or to stop things, but there is an element at the individual level and at the business level that we need to see that this is not a passive process. This is something in progress that we shape.

Debbie, I love that. Ethics is not a passive activity. It’s a proactive activity. When people say, “Well, tell me the difference,” I say, “Remember the story of the Good Samaritan? The Parable of the Good Samaritan is an example of the difference between do no harm, which is passive ethics and do good, which is proactive ethics.” I won’t bore you with the story. We all know it. If you haven’t, you should pick up the Bible and read the Parable of Good Samaritan. We need to become the Samaritan. We don’t need to be the other people who left the stranger on the side of the road bleeding and didn’t do anything because they weren’t going to do any harm. They did no harm. No, that’s passive. We need to be proactive with our ethics.

That means that people need to be brought into the process. They can’t wait for somebody on a white horse to come in and save the day. The shiny knight isn’t going to save them, and the government isn’t going to save you either. By the time the government reacts to this thing, we’re already going to have Terminators knocking on our doors, delivering packages on one hand and God knows what with the other hand. We as humans, as humans need to understand the importance of proactive ethics and how we need to be taking charge of that, everybody. It isn’t just the elitists at the top. It’s got to be everybody involved in this process, which is why maybe my most interesting chapter in my book is a chapter about empowerment. How do we empower everybody in the organization to help get this thing done?

So how can I do that, Bill? If I’m in tech, so thinking about our audience, we’re in there. I might own a company. I might be a CIO, a CTO. How do I start boiling ethics into what I’m building, what I’m doing with my services, my products, et cetera? What should I be asking or doing today?

John Smale, who used to be CEO at Procter & Gamble used to always say, “You are what you measure, and you measure what you reward.” At first, I didn’t understand that, and I thought, “Wow, that’s really powerful.” People always say you are what you measure. If you measure things, you can optimize them, right? If you can measure, you can predict. You can optimize, but he went one step further that you measure what you reward, which is really his way of saying that the only things that are really important are the things I’m paying you to do. If you’re an organization and your paycheck is entirely built on selling products to customers, then that’s what I say is important. Anything else I say about customer service is BS, right? If you are an organization and really believe that ethics is important, you better find a way to work it into your compensation. You better figure out what are the metrics around that, right?

Customer satisfaction, employee satisfaction, environmentalism, society, diversity, all these other variables, so don’t tell me as a company you care about diversity, but no one gets paid on it because that’s BS then. So you are what you measure, but you measure what you reward. My advice is organizations need to think more holistically around the KPIs and metrics around which they’re going to measure their value creation effectiveness. Guess what that does? That brings us right back to the concept of economics. Here we go again, right back to it, baby. All roads lead to economics, so that’s what I would say to an organization. You got a process. Thinking like a data science methodology, the first step in the process is a very labored process to make sure as an organization you have thought through from the perspective of all your different stakeholders and constituents, how do we measure value? How do we know we actually created value for everybody that’s involved in our ecosystem? Hard.

I think within that, isn’t it also because I think if we are going to build in the data ethics, measuring our value, but it’s also recognizing our cost impact negative aspects. That is a scary thing for companies to begin grappling with. What is the cost when things go wrong? What is the human cost? What is the impact on the environment and those things? Measuring value, but measuring … I’m not sure if I’m getting the right word, but the cost to other things. Those “intangibles.”

You’re hitting on a key point, Debbie. When we start thinking about making these kinds of decisions, we need to go through the second and third level and understanding what might be the unintended consequences. Really to go through a process and say, “Okay, we’re going to do this.” Now our government is great at making all kinds of policies without ever thinking through the second and third-level ramifications of this. “Well, we did this.” The policies were made with the best of intentions, but they didn’t think through about, “Well, what happens if this happens?” One of the exercises we go through in our workshop is what are the ramifications and cost of failure? We’re going to do this project. What does it mean if you fail? How do you know if you fail, and what are the costs, right? Unemployment, people being laid off, customers maybe dying from the product? Who knows, right, but you need to think through these unintended consequences so you can really understand … we’re going to get techy in a second … the cost of the false positives and the false negatives.

If I’m dependent upon a data science team who’s going to build a model and they’re going to measure the effectiveness of that model based on the cost of the false positive and false negatives, the data science team does not. It’s not the responsibility to figure out the cost of the false positives and false negatives. It’s the responsibility of the organization to think through what are the costs of the false positives and false negatives, and it gets very hairy. For example, using AI to make hiring decisions. Okay? Think about a false positive. A false positive in hiring is hiring somebody you thought was going to work out and they didn’t work out. You had to let them go, right? Your models, your AI models need to learn what a false positive is so they can become better at that.

The good news on a false positive in that example is we have all their performance data. We know what they came in. We know what failed, so we got that data. That’s great. Perfect. Yeah, my models can learn. So I want to build models that continuous learn and adapt, but here’s the gotcha, the false negative. What of the person you didn’t hire that you should have? Yeah, it’s easy to say, “Oh, we can’t track that.” Bullshit. Yes, you can. Yeah, you could follow that person on LinkedIn. You could call them and do surveys. You could follow up because your model, if it’s not also instrumenting for the false negatives, isn’t going to continuously learn and adapt. It’s an important conversation to think through these unattended consequences and the results that it has on the impact of the false positives and false negatives if we truly want to build AI models that are unbiased and help us to continuously grow and prosper.

Listen, Bill, I think we’re going to have to get you back another time. I want to definitely hear when the book comes out. You’ve covered so much. You’ve left us with so much on thinking about the data, thinking about AI. I know how busy you are. Thank you so much for joining me today.

Thanks, Debbie. It was a great, fun conversation.

Really, really enjoyed it. We’ll look out and we’ll be aiming to build our Yodas, not our Terminators. That’s part of what we’ll be resolving to do. We’d love to get your comments, your thoughts on what you’ve heard today, and you can share them with us at fox.agency