Hello and welcome to Episode 2 of the first season of the "Your DataOps Advantage" podcast series by Hitachi Vantara! In this episode podcast host Bill Schmarzo (our CTO of IoT and Analytics) talks with Mike Foley (our Marketing Science Lead) and Joshua Siegel (our Global Co-Creation Lead, IoT and Analytics) as they share details around the Digital Value Enablement process, the upcoming Digital Envisioning Workshop, data culture transformation and how Hitachi Vantara is scaling data science across the organization. Tune in and catch a glimpse of how you can merge the art of the possible with the art of the practical.
Bill Schmarzo is regarded as one of the top Digital Transformation influencers on Big Data and Data Science. His career spans over 30 years in data warehousing, BI and advanced analytics. As the current CTO, Analytics and IoT for Hitachi Vantara, "The Dean of Big Data" guides the company's technology strategy and drives "co-creation" efforts with select customers to leverage IoT and analytics to power digital transformation.
Responsible for data science and machine learning marketing across Hitachi Vantara, Mike drives product innovation, customer success and marketing optimization.
IoT and analytics, Hitachi Vantara
Hello, welcome to the Hitachi Vantara DataOps advantage podcast. My name is Bill Schmarzo and I'm the chief technology officer of IOT and analytics here at Hitachi Vantara. The DataOps advantage podcasts is going to track the trials and tribulations of different organizations of different sizes across different industries as they wrestle with how do you get value from your data? What we're gonna do is we're gonna talk to these organizations about how they're leveraging DataOps to uncover value in their data and help them to figure out how to drive return on their data investments.
Hello everybody and welcome to the second podcast in our DataOps advantage series. Today I've got two worker bees. We're the day before the actual workshop and the two people who are going to be instrumental in making this thing work are Josh Siegel, who heads up our digital value enablement practice and Mike Foley who heads up our marketing data sciences organization. So today we're going to talk about what we expect is going to happen in the workshop tomorrow. And I start off with a little friendly question. I want you each to share. And Mike, I'm gonna start with you. Give me something, your first experience with data and analytics.
Well, that was a long time ago. I've been in this field for just since the first relational databases. That was my first marketing database. I was a marketing researcher and working with the, you probably never heard of R:BASE 5,000 those early, early days. I've also perpetual students. So I've just kind of recalibrated over and over and over again over the years until ultimately data science is my profession now.
Got it. Josh.
I got started a little bit of a different way. I have a professional services background. I used to work in the credit card industry and it used to be that the credit card companies spent a lot of money, spending millions, sending millions of mailers in the mail and getting less than a 1% response rate. Well that all shifted and they started to do electronic based, communications and use transaction history to base those offers on. That's when I became interested in analytics.
Good. Thank you. So tomorrow is actually the digital value enablement workshop day and you're both going to play very important roles. So tell us more about what your role is and your expectations for tomorrow's workshop. Josh?
Well, as you said, I run digital value enablement for Hitachi Vantara and the digital envisioning workshop is, is a cornerstone of our methodology. It's designed to build excitement and consensus around a use case that this organization can use to move forward. The idea is to quantify the benefit of that use case. The process works through everybody's thoughts and ideas about use cases, great ideas that we've interviewed out of people and we bring people together. It's a highly facilitated process to find that one use case that is that perfect combination of valuable and doable in the time that we have.
Got it. Great. Thank you. Josh. Mike?
Well as you know, I've, I've done this many times before where I come in as the first person to practice data science and marketing. And then I look at the infrastructure, which is usually in the very early stages and then bring in later on a team. So at first I'm hands on by myself, but I've never had the benefit of having a methodology like this to help me ramp. And so, what I'm really looking forward to is how this will help me ramp up the learning curve with adoption, getting input from the other marketers, looking at this current situation and level setting where before, as the sole practitioner, I've had to do all of that by myself, evaluate everything from the infrastructure to the ability to build predictive models. And so this I think will just help me in a lot of ways to increase the velocity of scaling a long term data science practice here.
Excellent. Well let's, let's build on that. I think what's, you know, you, we've been done a lot of pre-work, orals of meetings, email exchanges, going out to wazoo. Tell me up to this point, what have you seen that has impressed you by what you've learned so far in preparation for the workshop and what does scared you? What is kind of in the unknown? So the good surprises and the bad surprises?
Well, I've done this as I mentioned probably four or five times before as the first person when I landed. And when I got here, what I was really impressed by was the technology that's in place for early days of a data lake. What will ultimately become a data lake. I think all the hardware's there. And on our IT side, we have a lot of professionals that are very comfortable with the latest technology like Pentaho, Kubernetes, Kafka, you know, the latest big data technology either. So they're able to stand up tools like R, Python, Jupiter notebook for me in a way that I can start working right away. Second thing was just you as analysts always do exploratory data analysis. And I had some consultants that came that work on actually your team that are data scientists that could work with me right away to start analysing the data. So we found a lot of really good data there and then some gaps that we were able to remediate right away in preparation for the workshop. So now we have a good handle on the technology and the data that will be there to address all the use cases so we can kind of get a feel for the feasibility of anything that comes up. On the other hand, because this is all new area, I don't think there's a lot of process in place. There is some legacy relational database data warehousing kinds of perceptions about what data science is about. And I think that that kind of thinking and processes need to be to evolve to accommodate the more freewheeling ad hoc nature of data science and also the capabilities there though that many people have never experienced before. If that makes sense.
Yeah. I think what happens in an exercise like this, if there is a, a lot of unlearning and relearning for a lot of people involved and even the user requirements for marketers now, if they've never experienced data science in a marketing, uh, uh, forum, they really can envision what the possibilities aren't necessarily. Right?
Yep. Great. Josh what have been the pleasant surprises and the not so pleasant surprises?
Sure. So you're correcting that. We've been preparing for this digital envisioning workshop for quite some time. And part of that preparation is a lot of interviews and we interviewed probably 30 people across, uh, both the marketing organization and the it organization sales. And one of the great surprises that we've gotten is how pumped everybody is for this project, for this initiative. People are bringing forward great ideas, uh, both on the technical side as well as the, the use case side. They're excited to be working together to be focused on this and about the potential that exists in our software and our solutions. And also this methodology. On the, on the flip side, the not so great surprises are probably familiar to, to many of our customers because we unfortunately suffer through some of those same challenges around data silos, around organizational silos and you know, historically probably less communication across those silos than we would have liked. And those are things that we are actively looking at and addressing and mitigating. And that's been a great benefit of this process as well.
Let's just talk about cultural change. I think we all know, we've all been in this industry for probably too long. The technology isn't what gets you, it's the cultural change. So what have you been doing in order to prepare the teams for tomorrow? Get prepared for what could be a culturally changing kind of experience?
Sure, sure. We do a number of things. We, we try to pre message everything that's gonna happen in the workshop. We talked to everybody individually. We confirm back with them what we are going to share in the workshop. And we tried to make it as comfortable a process and an environment as we can to share ideas. All ideas are worthy of consideration as someone I know likes to say. What's important in addition to that is for digital value enablement as the facilitators of this process to push a little bit to propose ideas to connect the dots. We are very focused on doing our jobs and doing them well, but to connect a solution like a Data Lake, you know, all of the great data that we have to an outcome that is relevant for the business, we need to connect both of those organizations. The facilitation technique, perhaps a little bit more art than science, but that is a part of the process that we call digital value enablement and a very important process.
Now, Mike, you, you joined us recently, um, but you and I knew each other from a previous life and cultural change has been one of these things that has dogged a lot of organizations. Share with us what you think you've learned in the past and how that will, you can apply that here regarding some of the challenges we're gonna face from a cultural change perspective.
I think in this space, getting adoption is really the hardest part. It's key. And if you go to academic conferences, for practitioners like myself, that is the biggest theme. It's not what's the latest machine learning technique out there. It's because there's just a huge range of those, but it's getting the organization, what they call legacy organizations and there's a huge market for that, transforming these legacy organizations. So, I think there's a lot of, a need for change with marketers who've been very successful doing things a certain way and don't necessarily understand data science, which is a very new field, you know, a hybrid of applied mathematics and statistics and computer science. Just, a melange of many things. And so, for many people that's very black box and on the it side we're not talking about one system, but a Data Lake that is 20 or 30 systems. So to deal with that as a whole different framework, then uh, then one might be accustomed to. And I think in organizations where even at there has been maybe six months of just spinning your wheels, just waiting for people to come around to that aha moment that Hey, we don't have a year in roadmap. Hey, we're not dealing with one system. Hey, we are doing things with marketing analytics that are transformational, that have never been done before. Don't be afraid to try it, but we can't say that it's been done for 20 years successfully. You're trying, you're breaking new ground. All of those are different perspectives on the state of marketing and the state of the intersection between marketing, mathematics, computer science and DataOps. Then people typically have come up within their careers. Does that make sense?
Totally makes sense. And I think one of the challenges we see consistently across all of our clients is this, this need to sometimes unlearn traditional ways how to do things. Which brings me to my next question. So we know that these workshops, lot of them explore, will explore a lot of the realm of what's possible. Part of the, the opportunity for us to overcome this on learning is to show people what's possible with these new technologies, new data sources. But let me ask you a really tough question. How do you balance the user exuberance from all of this new things we can do with the pragmatic realities of what you can actually do given a time frame for us in particular of October in the NEXT conference, Josh?
Right. Well, I think you put the, the nail on the head really, we'd like to call this, uh, a really kind of merging the art of the possible with the reality of the practical. And that is something that is so, so critical in these digital envisioning workshops, but also in the step that follows that which is what we call a digital proof of value. And it's important really to get people to focus on the fact that this is an interim step. It's been great working with Mike because Mike has such a great vision for marketing science, some of the data sources we need, the techniques we need, the resources and, and people we need to do that are longer term, you know, efforts. But what we need to do is build incremental successes. We need to demonstrate that this great data that we have and these great solutions that we have can affect an outcome. And getting people to focus on that interim goal is a key to that.
Mike, what do you think? How do we balance this.
Hopefully achieved user exuberance with the pragmatics of what we can actually do in October timeframe?
I think early on we'll be able to give them a good feel for what is possible. Show them some very concrete models, concrete results, the ability to execute. But I really would rather have them focus on it as a long term transformational way of doing things. I remember you mentioned the company where we work before and people will say, well, Mike, how accurate are these? propensity to buy models. You use random forest or you use a support vector machines and how accurate are they? Well, we look at many, many diagnostics and it's not really an easy to explain how accurate they are, although we do have a feel for that. But I would say don't even worry about that because these models, even if they weren't accurate, are going to learn over the course of their use for many years because every time we go out there and it makes an error, it will learn from that error. So just start using it and don't think about a one shot marketing execution like in the old days, but think about it as a continuum over time of targeting or segmenting in a more precise way. And those models will improve through use because there are going to learn.
It's actually an interesting point because one of the conversations we have a lot of our customers is this realization that the economies of learning are more powerful than economies of scale. Yes, you can put in place an environment where you're driving, sharing and reuse and refinement that the models may start off, not mean great, right? But they get smarter over, over use and, and um, and they get more valuable. Which really brings me to one of the key points. So we understand we're on this journey. We're, we're even an event. NEXT isn't the end point. It's one more step on this. You said, Michael is this digital transformation journey. Here's the hard question. How do you scale this? How do you scale this across the organization? Like I'll give you that, right first.
I mean, that is ultimately the key is not to build a model and that, I mean, that really is not the objective. The objective is to build the infrastructure, a platform and a team that allows for repeatability and scale across a many, many different quantitative approaches to marketing. And through a company that's called, uh, pervasive use of analytics or a company becomes an analytic competitor by scaling across everything. So that entails not just the infrastructure, which is a big deal. The team which understands the discipline and can do many things, not just one thing. Well, but then of course you have to have all around you a transformational change among the users and also the ability to execute so that model is not on a PowerPoint, but you're trying to execute through systems. So that in itself to be able to push out, not just bring data in from a, and this goes to the DataOps standpoint, but it's not just integration, but also integrating back out into the systems for execution. So all of those things, uh, that, that is the actual analytic team could be very small, right? But it's all of the, all of the things around it that are huge and all of that requires transformation, if that makes sense. And that is not trivial.
Yeah. I feel like I'm playing straight man a little bit here. There is the technology component and the resource component to scaling this for sure. In addition to that, I would add a couple of things. There is a, there is a process component. There's the institutionalization of this methodology. Getting an incremental success within the marketing organization will be huge. Turning that into longer term success for marketing and marketing, science, all of that perpetuates itself and can be scaled to other parts of this organization. Once that quantifiable benefit, that real organizational benefit is there, there is a real people are, people gravitate to this because they know it works. The second, the second point I'd like to make is there, there is um, uh, there the requirement really for, for an ongoing commitment to focus on outcomes. Uh, we've worked with customers before who, started on this journey and they have been focused on those outcomes and then they've, they've, they've lost that focus. The technology and the models, we're no longer as effective as they could be without that. I think those four components that the technology, the resources, the process and the commitment all, are helpful to scaling this.
Thank you, Mike. Josh, thank you very much. This is the wrap up here on episode two. And the one thing I wanna mention before we leave is that both Mike and Josh talked about, I'm going to call summarize management fortitude that is the ability for the management organization to stay the course, to not chase the low hanging fruit all the time, but actually have a vision in place of what they can actually drive towards. You heard in the last podcast from both Renee and Jonathan, that level of management, fortitude, that sort of belief that we can actually deliver something, not just once, but deliver it over and over again and literally transform the culture of the organization. So I hope you tune back in to episode number three where we're going to have finished up the workshop and hopefully we're still talking to each other, but we'll share with you the learnings that we have from that workshop. So see you soon. Thank you. Thank you.
I hope you enjoyed this podcast and you certainly want to come back to the next one as we talk again to more organizations about how they're leveraging dataOps to drive value out of their data. If you want to learn more about Hitachi Vantara, track us on Twitter @HitachiVantara, or if you wanna follow me, follow me at @Schmarzo. I'm the only one on Twitter. Thanks for your time. Until next time, cheers.
Stay connected with updates from Hitachi.
© Hitachi Vantara LLC 2020. All Rights Reserved.