Speaking of Service

Value Engineering and the Economic Value of Data

October 12, 2022 PTC Season 2 Episode 8
Speaking of Service
Value Engineering and the Economic Value of Data
Show Notes Transcript

Read Danny´s Perspective on Machine Monitoring and the Journey to Service Performance

How to think differently about the economic value of data and investing in appreciable data assets. In this episode, Chris MacDonald speaks with Danny Jackson, Digital Transformation Director and Brian Allred, Global Director, Data Analytics and Digital Technologies at Autoliv about their digital transformation journey and how they've evaluated the strategic value of data.

Announcer:

Welcome to Speaking of Service, the podcast that uncovers practical ways to grow service revenue, control costs, and improve customer satisfaction. If you're looking to innovate, gain a competitive edge, or just learn about the latest service trends, you've come to the right place. In today's episode, Chris McDonald, head of AI and analytics sits down with Danny Jackson. Digital Transformation director and Brian Allred, Global Director, Data Analytics and Digital Technologies at AutoLive to discuss auto Liv's digital transformation journey and how they've evaluated that strategic value of data.

Chris MacDonald:

Welcome to the show. Today I'm excited to talk about the economic value of data and what it means for organizations pursuing a digital transformation strategy. We often hear adages like data is the new oil, or whatever the phrase of the moment is. Yet when. Start working in an organization, you often see it's seen as a discrete cost one in terms of how to connect to that data, how to curate, how to store it. The ROI or the strategic value of that data can be identified in terms of a discrete use case where the combination of technologies and data unlock specific business value as described and measured by a business. Yet the strategic value of data has a multiplier effect in an organization, not just for one use cases, but many use cases. So how to treat data like a strategic asset is going to be the topic of today's discussion, and I'm incredibly excited to welcome our friends remodel Leave Danny Jackson and Brian Allred. Danny Jackson's responsible for manufacturing digital technologies and data capture. And Brian is from the data Analytics and Digital Technology Teams. Danny, please introduce yourself and your role and then we'll go over to Brian for an introduction. Thank you,

Danny Jackson:

Chris. Hello everyone. I'm Danny Jackson. Uh, Digital Transformation Director at Auto Leave. Have responsibility globally, uh, for what we're doing in our manufacturing. And also with the collaboration with the other business functions into engineering, supply chain, et cetera. So happy to be here today. Thank you for having me and

Chris MacDonald:

Brian.

Brian Allred:

Yeah. Yeah. Thanks for letting us join the show. We're really excited on the topic today. Uh, I'm Brian All. I am the IT director, uh, for our data analytics and digital technologies fraud leave. Uh, so my team focuses on everything from the data sciences, uh, to the data manipulation, data engineering, and, uh, the other end of the spectrum, which is the reporting back to the business of that data.

Chris MacDonald:

Awesome. Well, well thank you both for joining through our working relationship with Oly. One thing, um, I'm highly aware of and I find to be very true about your brand is a focus on quality and safety. It's the core of what, um, you know, your company brings to market and what your, you know, your employee base is passionate about as, as part of your work. I think data. And the amount that you capture and curate and store around your manufacturing process and operations is frankly ahead of the curve and impressive in many ways. So there is something about how you think about data as a strategic asset, the value that data can place in terms of driving efficient operations. Danny, can you speak a little bit about how the business views this? What sort of changes you're driving to enable that? . Brian Allred: Yeah. I'd

Danny Jackson:

say the, the way we typically approach that is first we take a look at what, what are the business outcomes that we're trying to drive? Uh, before we start digging into the data sources and the types of data and the structure of the data, we really try to tie it back to the business outcome specifically, Like, you know, can we affect the quality of our product? Or if we wanted to affect the quality of our product, how would we want to affect it? What things. Would we want to control? Uh, or what are some sources of inefficiencies in our business? What are the things that cause us to have to do rework? What are the things that cause equipment to go down? And then we back into the data that we need in order to influence that outcome? So with that, we're able to really put a dollar value onto that and say, Hey, this data, if we could capture this data, And do something with it, we can really affect this or that business outcome.

Chris MacDonald:

Yeah. And if we take some of those examples, um, that, or successful use cases that you've had in manufacturing and I think about Brian's role in the organization, how does the nature of feasibility and reality of that data tie into an identified or prioritized business case? And Brian, that's, that's really for you. Yeah, that's a

Brian Allred:

great question. So back in the, uh, back in the eighties, uh, we coined the phrase, uh, called data mining. So we used this phrase, and it was, it's a great analogy. It's, uh, uh, taking the time to pick through all of the data and try to find a negative gold. Uh, and that's how we worked for many years. Uh, just again, back through the, a, uh, several decades of different technologies to try to find the nugget of gold in all of the lead. And modern tools and technologies, uh, now act more like a philosopher stone. It's much more about, uh, let's not sift through our data. Let's transform our data. Uh, let's move away from the traditional, uh, ETL where we would try to, to polish that data as much as we could as we moved it, and then do a report on it, and then hundreds of man hours of scrubbing. Uh, that data to, to hopefully improve the quality of it, um, to now moving it to more modern ways of working with elt, with things like the Lambda architecture where we can now take what used to be garbage data and start building models and, and really transforming that data. Uh, again, kind of acting like the philosopher stone. Instead of picking through all your lead to find a piece of gold, let's go transform the lead into.

Chris MacDonald:

Absolutely. And I'm sure you know, you experience what many do. That data is diverse even across an operation, especially when you're first getting started, you know, on your digital transformation strategy. So as you think about capturing the right data or standardizing how you capture, um, that data, not pulling that data in a one-off manner, but in sustainable, sustainable, consistent manner, what are some elements of, of how you govern all that? . Yeah,

Brian Allred:

that's a great question. So defining what the data is and where it is, where to go get it, that that's the easy part, right? But in many cases, you run into these challenges because the data is coming from multiple data sources, uh, with different formats, different relationships, and very often completely unstandardized. Um, so the first step is to really. Bring some standardization to, to your data. Uh, and that starts with building a model. I'll give you a great example. I, I've asked in the past a group of managers, To define a simple data term, let's say supplier. Uh, if you ask a finance person what a supplier is, it's gonna be a different answer from what logistics or purchasing might give you. But all of the answers are correct. So the key is coming back and building a standard model that when we say supplier, this is what we mean. This is a. Uh, and you build that data model and you build that, uh, reusable pattern for using that data. Uh, and if we go back to our metal analogy, it's really about building a mold. You build a mold that says, this mold is a supplier. This, when we say supplier data, this is it. Now that you have that, it doesn't matter whether your data comes from one source system or a hundred source systems, you, you're gonna pour the data into that mold, you're gonna scrape off the garbage, and what you're gonna have is, Nice golden nugget of truth, uh, that says this is our supplier data. Um, and so here's the, the focus is on the technologies and the methodologies that help us to build those master records and build the tools that help us to enforce the quality of that. Uh, and these things are critical to ensure. The delivery is given in a consistent manner with consistent quality. And, and honestly, it all starts with business governance. There there's gotta be clearly defined data stewards who own that business relationship with that data and can drive, uh, that

Chris MacDonald:

mold. Thanks Brian. That was really, um, insightful and I want to go back, switch a little bit back to how the business views, how to take advantage of some of these elements we've been talking about, understanding the importance of beta, how we capture it, how we stored, how we model, how we govern it. There's elements where on one side the business needs to contribute to that, right? It can't just be a data driven or an IT driven exercise. The, the business has to understand the importance of, of capturing this data, of contributing to it, of understanding it, of leveraging it. And I think, uh, Danny, you probably have experience on some of the hurdles and challenges of, of making that happen in partnering with the business. I'd like to hear about that. And also, uh, if you have an example of where you really overcame that hurdle and, and what that specific success looked. . Yeah,

Brian Allred:

that's a, that's

Danny Jackson:

a great question. I, I think we do have, like every business, we, we tend to fall in the trap of being data rich and information poor, right? Mm-hmm. , uh, and we have to come to the recognition, the realization that part of a digital transformation. Is much more than throwing iot devices out on the shop floor or just pulling data from a plc. We need to actually take that data and transform it into information, actionable information, uh, and the way you, the only way we've found to do that is to build this really strong partnership between the business executives and the IT team. Uh, and when the business executives start to realize that they are the owners and the stewards of the data, it is really there to provide infrastructure, to provide systems and processes, But they need to govern the data. They need to say, This is why the data is valuable, and here's the context to it, and here's the output that we can get from it. So if, if I think about some examples we've been through, one of the things that's critical to any manufacturing businesses, operational availability, um, and when you think about operational availability, it's really, the inverse of that would be machine downtime, right? So if you have. 10% downtime, then your operational availability is 90%. Um, a lot of people who do OEE calculations will recognize that as one portion of the overall equipment effectiveness, but, but for us, it was really critical to understand. , what are the causes of those downtime? What are the things that are one offs? What are the things that are really systemic? How does that look line to line within a plant? Plant to plant within a country and even country to country? Do we have this same type of downtime in China that we have in the United States? What about in Europe? Is it the same? Um, so we, we started off with kind of a, a process approach or a, a system approach saying, Okay, everybody collect your data in this system, Right? So everybody ran out, they collected the data in the system. They said, Okay, we got the data, it give us a bunch of insights. And we went, uh, none of the data matches. Mm-hmm. , right? So we had to come back on that loop and say, Okay, business partner. , what do you want to call this? When the machine goes down for reason X, what's the name of Reason X? When it goes down for reason Y? And can we have every plant in the globe put it in under the same reason? So that that governance and that structure became very apparent, quick, very quickly to say we can get the insights we need. If we have that structure

Chris MacDonald:

properly. I and Brian, I don't know if that Go ahead. That, that absolutely answered. Uh, you know, the question, and it, um, made me think, and Brian, you can probably relate to this. As someone who thinks about data and analytics day to day, um, there's consistency that you can have in, okay, what is, um, a failure or a downtime or an event or a reason code, um, that can exist. But even once you get there, oftentimes. A leader or a plant manager frames a question right about asking of that data. How do they understand, you know, operational efficiency and throughput in relation to a phenomena? And that means that you, as someone who has to look back at that data, Maybe it's collected standard, you know now, but how can I relate that question and the phenomena that may or may not be represented in the data appropriately? Can you speak to some of those that, that subtlety in, in translating those business questions against the data that you have to manage and govern? . Brian Allred: Sure, absolutely. And, and I think this goes back to the earlier question, uh, and ties in very closely with that, where again, we come back to having that standard definition where we have that standard, uh, model and that standard glossary of terms even, uh, where we can standardize on not only the data model, but the definitions of that data and what it is, and. Also how the relationship between the different pieces of data work, the relationship is very, um, important and oftentimes very, very tricky, uh, to build. And again, this comes back to needing those business experts, working with the technical experts, uh, to build, uh, an information model. Uh, I wouldn't even call it a data model. It's very much an information model that builds their relationships, uh, and builds that structure so that everybody is talking, uh, to the same thing, to the same sheet of music as it were. Yeah, I like to say that it's, it's all semantics, but it really is technically right. It's a, it's a semantic layer. And how do I translate these, this understanding of an organizational wide representation of data and the problems we're trying to solve with that data, The fact that you are leaning into it, right? That you, you are able to, to think about these concepts and the business is a sign of great progress that I'm sure many of our listeners and their organizations are not quite there yet. Right. Um, and trying to think about that. And it begs the question for Danny. What's the right level of business organization relationship between data ownership, data governance, um, like with what Brian does, um, and driving business transformational value?

Danny Jackson:

Yeah, I, I think it's a tricky balance and I think it's one that's a little bit dynamic. There has to be a little bit of give and take on both sides. Uh, I don't know that there's any hard, fast rules. But I think what we've seen is once the business executives kind of started to take the ownership of this, and they were the ones driving it, from a value perspective, uh, it really made a difference in how quickly we were able to make progress on some of the topics. So from, from the executive level, them taking that ownership of becoming the stewards of the data and all of the other functions are supporting functions in achieving a a given outcome or a specific vision. That started to make a big difference in how quickly we could standardize the data. How quickly. Brian could put the governance and the models around it, and even when it comes time for, for budgeting for additional, you know, devices to capture the data, the infrastructure to transport it and store it, all of that very quickly becomes aligned at

Chris MacDonald:

that point. Brian, any additional thoughts there?

Brian Allred:

Yeah, that's spot on. Um, uh, no additional thoughts specific, excuse me, specifically to Danny's points. Uh, but I will say one of the other challenges, uh, if we talk about challenges that comes from this is building these models in these relationships is often a very time consuming process, especially when you've got to have, uh, the business integration as part of that. So one of the hurdles that we often face is how do we balance. Building that golden source of truth versus the desire to get insights very quickly. Uh, and these can often be mitigated by starting to build some early insights on less than pure data. Right? If we go back to the Lambda architecture, in fact, the analogy I use about the, the metals, uh, is still used in the modern land architecture used by the big companies for big data management, uh, where you have your bronze level, which is, you know, Raw, unfiltered, um, un unproved data, uh, and then you have silver, gold, even platinum. Uh, and so one of the, the challenges we have is what level and at what level, or, or, or what cleanliness level mm-hmm. , uh, do we need before we can start connecting the business to that, uh, because there are opportunities to connect. Business user to sometimes even the raw level of data and let them start building some of their own insights to a self-service tool, uh, or even uh, our data scientists using AI and machine learning to better filter and, and generate insights out of that raw data. Um, That sometimes mitigates the time consuming process to get to the gold level, but it doesn't come without its risks, right? If you have somebody building reports on, uh, raw data and they're doing the cleansing, uh, you now are on the risk of making decisions off of data that's, you know, where the quality is questionable.

Chris MacDonald:

Is there any sort of, uh, guidelines that I, from my experience, there's. There's not a perfect system here because ultimately, like you're saying, it'd be great to have a perfect world where all data is of quality. Everyone's using it the same way. It's governed perfectly, but oftentimes, yeah, you want to get to the insights. You want to get to that business value, and you don't wanna let perfection be the enemy of the good. At the same time. So striking that balance of how do I ensure data quality, It's hard, especially from, you know, my perspective as an analytics professional. I don't want to, if some, a business user wants to start exploring and doing their own analysis, I think that's fantastic. Right? How do I govern that though, so that they can have that freedom, that they can be more data driven, which is ultimately what we're driving at while creating some guardrails. Have you found any antidotes or, or things in the organization that help strike that balance? . Brian Allred: Yeah. There, there are some things that do, Uh, one is, uh, being a little bit more strict about who are the people able to go access those lower level qualities of data. Uh, you need to make sure that they are the business experts who know the data that you, you can trust with the data. That are able to build their own insights. Uh, we also have some governance around reports that are built on data that's of lower quality, uh, where we have a certification process for the data, um, where the people looking at that report have an indication of where the data for that report comes from, uh, so that they can trust in that level of data. Whether it's certified or not. So there are some frameworks we put around that. And of course we, we manage the security and access to that data carefully, uh, to make sure that it's the right people driving those earlier insights, uh, off of less than the pure data. Awesome. And Danny, for question for you, and I like to ask this, um, of many of our guests, but if you were to look back from a Yeah. Business and a technical perspective, if you were to restart the transformation journey and what, what you've been a part of, you know, at Auto Live, what would you change if you had the ability to go back? What are a few things that you wish if you could have gotten that right to start with, um, you would've saved a lot of time, a lot of energy. I, I think

Danny Jackson:

it's an interesting question because. , I would say I would go back and put more time into the planning and the structuring of the data up front. I say that and say it's tricky because you also don't want to get into analysis paralysis, right? And like you said, banking, perfection, the enemy of good. At some point you need to move forward, but you know, like the, the, the example I gave, we rushed out with a software package saying, put all your downtime in this software package. But we didn't really have a good structure for the data, and so we had to go back and do some rework in order to get the insights we were looking for. So striking that balance of a plan, that's good. Maybe it's not perfect, but it's good. Have a plan and then go forward. . I would say in hindsight, we could have done a better

Chris MacDonald:

job at that. So Brian, let me flip the question, um, into one thing that you wish that you would have known and you wish that you could explain to executives and business owners and plant managers about the importance of thinking about data in a certain way. If you could have done that earlier on, how would it have made your life easier and overall digital transformation more successful across the.

Brian Allred:

Yeah, great question. And without a doubt, if, if we'd have taken a more business focused approach on the data earlier on, I think, uh, it would've made everybody's lives a little easier. Uh, but to your question directly, it's, it's really, again, taking the time to educate the business on the criticality of building that data model, uh, and building one consistent source of truth for a data set for the company, uh, that you can then build that, that mold. For your golden records and the importance of that versus, um, the traditional ways of just dumping it all in a bucket and trying to sort it.

Chris MacDonald:

As we wrap things up and as I thank Danny and Brian for their time participation and, and honesty and insights of how auto leave, uh, treated data and its economic and strategic value, I wanna recap some of the things we've heard. We've heard a lot about what it means to understand those business use cases up front, the business strategy up front, and to align data. In a unified manner as much as possible, but not to the extent where we are preventing people from accessing and deriving insights or getting into analysis paralysis. All of that speaks to what it means to have data as a strategic asset at an organization. If we can unify around what data means, what it means for the business and operations, what it means to model and abstract that data out, suddenly it became can become a center. A business strategy rather than an asset that's costly or that someone doesn't understand the ROI on. If we can align business, technical, um, and operational folks together around data as a strategic asset, it can empower and fuel a digital transformation strategy rather than burden it. So I want to thank Danny and Brian for joining the show today. I always appreciate your insight.

Brian Allred:

Thank you for

Danny Jackson:

having. Thanks for inviting us.

Announcer:

Thanks for listening to the Speaking of Service podcast, brought to you by ptc. If you enjoyed this episode, please subscribe wherever you get your podcasts and leave a rating or review. And be sure to check out other episodes to hear new perspectives on improving life for after market professionals, service teams, and the customers they support. If you have a topic of interest or want to provide feedback, email us at speaking of service at ptc dot. Or visit us at ptc.com/speaking of service.