 My name is Megan Kemp, and I'm the Executive Editor for DataVercity. We would like to thank you for joining today's DataVercity webinar, Data-Centric Strategy and Roadmap. This year's February edition in a monthly series called DataEd Online with Dr. Peter Aiken brought to you in partnership with Data Blueprint. Now let's go back to Megan Jacobs, the webinar organizer from Data Blueprint, to introduce our speakers at today's webinar. Thanks, Shannon. Hello, everyone. Welcome. My name is Megan Jacobs, and I'm the webinar coordinator here at Data Blueprint. We hope you've found the time to join us for today's webinar on Building a Data-Centric Strategy and Roadmap. As a big thank you, goes out to Shannon and Data-Vercity for hosting us. We'll talk to you in a few moments after I let you know about some housekeeping items and introduce your presenters. We will have one hour presentation, followed by 30 minutes of Q&A. There are many questions as time allows, but feel free to submit questions as they come up throughout the session. If you have any questions, you will receive an email with links to download today's materials and any other information you request during the session within the next two business days. You can find us on Twitter and Facebook and LinkedIn. We've set up a hashtag, DataEd on Twitter, so if you're logged on, feel free to use it in your tweets and some questions and comments that way. We'll keep an eye on Twitter feed and include any answers to these questions in our post-session email. Now, for the presenter and also our special guest for today, Peter Akin is an internationally recognized data management thought leader. Many of you already know him or have seen him at conferences while applied. He has a lot of experience and has received many awards for his outstanding contributions to the profession. Peter is also the founding director of Data Blue's print. He has tons of articles and eight books. The most recent is Monetizing Data Management. For today is Lewis Broome. Lewis has been a practice thought leader in data management who has been creatively disrupting the approach to data management. He has a lot of experience successfully designing, implementing, and leading global data management and information technology solutions. His record is marked by strong leadership, coupled with a passion for driving death and technology solutions from a clear business value proposition. Lewis has been a solution to multiple industry periodicals and is currently the CEO of Data Blueprints. And this is actually a preview of a workshop that Peter and Lewis will be doing at UDW. This is Peter and my business partner Lewis, who we were just looking back over history and discovered that the first time that I had actually presented together was back in 1996, so it's been a long successful run of us. Yeah. So, we started and get us kicked off and go over the outline. So, the outline of the presentation. So, the Data Strategy Overview. This is the why, what, and how the data strategy with the emphasis on the why section, determining the business needs, is really getting on the same page with your business partners, getting a sense of understanding of the business model and then looking at specific business opportunities. So, the success criteria in business terms, right? So, this is a critically challenging and somewhat elusive activity. Also, this is the outcome, though, that really matters most to the business. So, once we understand the business needs and measurements of success, then we're going to talk about developing the strategy solution that fits in that box. And then finally, we'll give you an introduction on developing a roadmap and a plan. This time, along with others, will be covered in more depth at EDW. So, the overview of the Data Strategy. Some of you may have seen this video before, this TED Talk, and it's basically a presentation on the power of understanding why we do things. It's why you do it. So, we've got to overemphasize how important it is that the better we understand the why, the more it is to drive a solution. So, again, we've come back to this a lot. We see this as a problem really quite often at all levels. So, it's more common than you can imagine. So, you know, it's how we saw a $30 million data warehouse get built and end up with a single user, and there's more to come on that. So, why do we need a data strategy? You have to understand data strategy in context of strategy in general. And if we go back to Michael Porter, the part I'm showing you here on the screen, the Y-axis, the one that's pointing upwards, is an innovation dimension. The Y-axis, the one that's pointing to the right on your screen, is improving operations. We generally talk about effectiveness and efficiency. And like all consultants, we have a four quadrant diagram on this. Where quadrant one is where the vast majority of the companies are. They're just trying to keep the door open with little or no focus around data as an asset. If you go into quadrant two, you're looking at then increasing efficiency and effectiveness. And if you go into quadrant three, you're using data to create strategic opportunities. However, the capabilities of the team that can successfully implement something in quadrant two are completely different from the knowledge, skills, and abilities that somebody is trying to do in quadrant three. Now most organizations, when we talk to them about data strategy, and since we do that kind of work, we get a lot of questions around this. Because they really do want to do something around analytics, which is their word for saying innovation around data. The challenge is, of course, going directly from quadrant one to quadrant three without going through quadrant two, you can move directly from Q1 to Q3, but it takes longer, costs more, and delivers less at greater risk to the organization than if you learn to crawl, walk, and then run by going Q1, Q2, and by the way, Q2 will give you enough tangible money that you can use to fund Q3 without having to ask for additional funds along the way. Very few of the organizations that we've worked with have mastered both Q2 and Q3, putting them in a quadrant four, very specialized category. And if your organization is thinking of doing this at all, consider yourself very fortunate that only one in 10 organizations actually has a thought-around data strategy. So, in our view, right now you believe as most of you would probably agree, the marketplace for most every industry is driving fast and furious to be information-based. So, while I acknowledge this, very few actually live it or operate as if it was true, our view is that the future will compete on being competent, agile, and innovative with data and information. We have found there's still a wide gap between that particular place and where people are today. Right? Many people have thought of data as a direct impact on business value by determining which products to sell to which customers and which contacts. But it's evolved, I think, to me much more than that when we talk about real information economy. And here's the case that we look at it. First, when data and information is becoming part of what we buy, whether it's products or services, people really are willing to pay a premium for information. So, it becomes part of the thing that we're actually buying, and that is one way that companies can compete. This is that, obviously, information is power in the marketplace. So, having a 360 view and situational awareness of operating, whether it's suppliers or customers or regulation or whatever it may be, this is important. The last one I think is really kind of interesting is that organizations are really being put at greater operating and reputational risk, just a bigger amount of processing that they have to do with data. So, again, this is sort of talking about in our current sort of process and not even where we're trying to get to in the future. I mean, this is really good in companies. I think a lot of problems in putting them at risk, and we've seen some of that recently. So, this in our mind highlights why data strategy is important. And Lewis won't say it, but he's got a great blog post out there at the bottom of the screen. So, check that out. He's got a little bit more on this particular topic. Thanks. So, putting the strategy together. So, this again sort of highlights the breakdown of the presentation. So, we're going to have our understanding of the business needs, both foundationally and specifically, right? Without really understanding your competitive advantage for your company, because that is where the data strategy needs to align, right? So, you can really understand the goals and the strategy of your organization and of your business if you don't understand how they're going to be measured. And then again, we're going to get into this holistic solution and approach. You'll notice at the bottom, and this is mainly the most important point on the slide right here, is that we find organizational changes the biggest challenge because we have to think differently about their data and it just cannot be overstated. I think this is our biggest challenge we've encountered as we've worked through these. Yeah. So, getting into determining the business needs, so we're going to have these competitive advantages, right? It's important to have this fundamental understanding of your organization's positioning in the marketplace. This is the foundation your company has built its success on. There are a lot of ways to look at this and there are a lot of business gurus out there that have come up with all kind of frameworks to do that. But you can bet that this is the fundamental concept that management consultants apply to a business. They're talking about this. So, we're talking about thinking like a management consultant. So, again, we often talk about we want to be the management consultants of the data management world. But you might notice there's confusion around this quite often about understanding really what the competitive advantage is. That may be confusion within your own organization. The biggest problem we're aware of is the credit card company that they think they're a data company. At some point of orange, discount retailer tried to compete by introducing some more expensive products and it was just a complete flop and it was clear to the customers. So, the idea of competitive advantage that really everybody understands it but I think it's important to point out it's not being the best. So, some times we refer to it as a race to the bottom. Being the best means everybody is doing the exact same thing as being the one who does the cheapest. The idea really is to be different. So, being inexpensive is an increase in competitive advantage but not if that is what everybody else is doing. So, adding value in a way no one else is whether that is being good at providing low cost solutions and creating operational efficiencies or because you have specialized skills that no one else can mimic. So, again, the framework is a lot more some examples you can use. I'm going to get into this a little bit. So, this is one of those frameworks that we look at for competitive advantage and three components of this. There's lower cost versus the differentiated and low cost is just focusing on things like operational efficiency, low cost suppliers. The differentiated is just how to focus on innovation, introducing new ideas and new products, product features. We're going the other way. It's narrow versus broad. So, you get the four quadrants. You can be narrow broad. You could be low cost or you can be based on being different. It's very rare for someone to be in the center but occasionally some people can sort of be a little bit above. So, a great example of this is the Walmart. They're probably a leader in low cost and their market is very broad. Walmart is being very different. So, they're offering very different products. You can't find anywhere else. They put together a unique combination of goods but they're still being fairly broad. Trader Joe's is in that center spot where they're low cost but yet they're providing high quality goods. So, Target could also potentially be. The key here is to get this concept and understand where your company is positioned because again, it's probably somewhere in one of these quadrants, one of the four quadrants. So, Walmart is the point. It's unlikely to be broaching two of the markets. Exactly. So, now that we understand how your company is positioned, the next thing is really to understand how does it specifically create that competitive advantage in your particular positioning that it's taken along, right? So, everyone has a unique value proposition that gives them an advantage. That's why they've been successful. They want to focus on what is important to the business. So, again, the reason for doing this is we get everybody, and you hear people say a lot, the data strategy needs to align to the business strategy but I think for us, the piece that's been missing is how do I understand the business strategy in such a way that I can align my data strategy. So, this is one way to do it here at Data Blueprint. So, this is going through this quickly. It's really talking about who has leverage in your particular situation. Sometimes the buyers have leverage over your organization and sometimes you have leverage over your buyers. Suppliers, OPEC was a great example of someone who clearly had a supplier that had huge influence over the world markets but specialized skills such as transplant doctors is a type of supplier. So, and it also could be expensive to change suppliers. So, you know, who has leverage over who in the relationship between your company and the customers, the suppliers. Another way that drives competitive advantage is new entrants. So, how easy is it for a new entrant into a marketplace in which you're operating? Is it capital intensive because you're not in the industry and, well, it's expensive to go out and get airlines. So, are there many other options for customers if they become price sensitive to what you have to offer? And the bottom one is, or the center one is, is more subjective is really about, you know, competitive as the marketplace, like code for this policy. The point is to have some sort of framework to get your organization's competitive advantage because more than likely, their business strategy will be focused on either showing up their strong position or trying to push back where it doesn't happen. So, we're just looking at Hyundai and Porsche. Hyundai is a, their customers are cost sensitive. It's a broad range of buyers. Porsche, clearly, very much a niche market, very differentiated. So, looking at their positions, looking at the five forces in the table, you can see how they sort of fall out. But the key point here is that you know, the Porsche, you know, customer data is really important to them because retention and repeat customers is really how they maintain their profits. And so, they won't have very centralized or individualized relationships with their customers. So, how do they manage that data? We're really into R&B, you know, a different, a differentiated product, right? Research and development constantly developing new product features. It's kind of the opposite, right? They're very price sensitive customers. So, they're really into operational efficiency. So, if you're thinking about your data strategy, you know, there are areas you want to dive into at these companies to go look for because, and it's not to say that they don't care about, you know, the Porsche doesn't care about operational efficiencies. But if they're going to look at operational efficiency versus R&D, they're going to go with R&D every time. All right, so, the takeaway is that it's important to your business partner and focus there. It makes for a smaller element to try and eat when you're looking at the volume of your work list. So, here's an excerpt from our article, first article, I'll update. And I just thought it was interesting how their proprietary propositions for leveraging data matched up nicely to the five forces. It drives insight and innovation, right? That's really about being differentiated as you establish barriers to competition, right? That's making it hard for new people to enter operational efficiency. Again, that's around low cost providers. So, you know, it's a repeating theme that when you're looking at business value, these are the areas you want to look. Just to quickly summarize, you know, this is critical for a couple reasons because it sets a vision for why we practice data management. I think that's really important. And I think that's something that, you know, I struggle to do. And it's not that many people skip this step. But in our case, it's because people look at data sometimes as having IT-driven goals and not just goals. So, we need the business to be more, to have more ownership and accountability for how we leverage data in the organization. So, that's a summary of what essentially is an entire semester's course on how to do strategy. Some of you may have encountered something similar in business school, but let's talk about the real world now as we encounter it. And when we go in to do data strategy and work for organizations, we generally find that we end up with one of two conclusions that comes away. One, the organization has a great business strategy. And what that really means is that they've got a great business strategy as an organization. Then they also have a really wonderful IT strategy. And underneath that, they have a data strategy. And all three are aligned perfectly and working really, really well. And we see that, what, maybe one in 10 customers? The other part of it is, the other nine-tenths of them are, we've found out that the strategy they have actually either could be improved or is really not well expressed at that level. So, then we have to apply some analysis to the process and to figure out what is their strategy. We use a lot of what's called reverse engineering analysis so that we can tell them what is the most important parts of their processing world that they're doing and how doing a lot of it seems to indicate what their strategic needs are in certain areas. And if it's not, it's a good hotel for them to come back and react to. So, the key, if you don't have a business strategy that you can start to work towards, is to find out what changes would be seen as useful or important. And plan to do something that is useful or probably even more so than useful and noticeable in the organizations. Okay. So, in terms of career, the organization's business goals and objectives have to be measurable or that it's not a full definition. So, and from a data management perspective, I've often struggled with this myself, getting at how to measure the business value. Like, how do you quantify the business value of data quality or the enterprise data architecture? So, I think we found that it takes a clear business need that can be measured to derive the value of the underlying data strategy activities. And in the absence of a clear value proposition, the business value of a data strategy is unknown. So, we would be looking at one side of the equation when we'd be looking at the call side. If we can't measure the value on the business side. So, to measure the value of a business operator is critical in the development of this successful data strategy. Again, we look at another framework here on understanding really what we mean by measurement. Again, there's some doubts that needs to be looked at to really get that concept. But this object of measurement, I think, is really an important part. The concept of understanding why. It's too often assumed that we understand what we mean when we're going to say we're going to measure customer satisfaction. We're going to measure customer retention. These end up being ambiguous once you start to get into it. So, we're interested in measuring repeat customers. And Hyundai is very interested in measuring or counting the number of parcels. But if someone was looking at customer satisfaction, we may get into the conversation, well, what's a customer? So, a customer may be someone who actually hasn't bought from us before in that particular context. And if you don't define that, it becomes difficult, or you will not really be able to get it the measure that the business wants. So, that's important. And in methods of measurement, it really is, this is your tool bag. Just the sampling would be an example, Bayesian theory. One of the inspirations around this area is a book by Helen and Douglas Hubbard. I just love giving a shout out on this. It's a fantastic book. It's $25 at Amazon. And, you know, to push everybody else's books instead of ours, but this is really worth reading if you're in faced with something as Lewis was just describing. If these things seem vague to you, it's probably only because we aren't practiced at it. And one of the things that he talks about specifically is, when you start doing this analysis, you start to formalize stuff. And Lewis, we've had many of these discussions as we're introducing art during your associates to talk. One of the things Lewis emphasizes over and over again is, rate. I hope that didn't blow your speech to everybody, but it's in my attic there. Because the minute you have a picture in your mind, you can have some things worth criticizing. Whereas the ideas are vague and they just don't help a whole lot. A wonderful excerpt from this, Chapter 7 of this book that I'm showing on the screen here now, talks specifically about measuring the value of additional information to a decision. And one of the things that he goes in and proves pretty emphatically is that most people put too many variables into their equations. It turns out he's simplifying to actually get better results in order to do that. And we're going to talk to you about how we did that on one particular customer's approach at this point. So I'll tee up the example. And this is a data blueprint example. One of ours is an international chemical company. They do a lot of research and development and creating chemicals to go into oil and gasoline as added. So fairly large company, a billion plus. Basically, what happens is a customer will come to them and they sell oil on the market for people to put into their cars and they'll come to this company and say, well, we want to sell oil for high-performance cars that will keep the engines running cooler. And these guys will go away and customize a formulation to measure product specifications. Over the years, these researchers have run hundreds of thousands of tests at 25,000, 250,000 each to create these formulations for customer requests. And it's really a very exciting environment because you walk through the halls and there are all these labs, and their labs are like engine blocks sitting on a stand with a thousand probes on it taking all kind of measures to make sure that the formulation is doing what it's supposed to be doing. So in this case, the objects are the things that we're really focused on. Through this exercise that we've been talking about, it's just execution, customer satisfaction, and research and productivity. So again, what they were figuring out was we have all these tests that we've run previously. We have to leverage that data in such a way that we can reduce the number of tests that we actually have to run because we understand if we have a little bit more phosphorus given the other variables in this formulation that it will run cooler at higher RPMs. And that's the point where they can almost really formulate variations for their customers. And because they were finding that they were actually end up running essentially the same test over and over at the very expensive cost per test. So we talked to them about the test execution, the customer satisfaction, then if they could virtually formulate the solution, they were getting a quicker time to market from initial request of certification. And then of course you have these Ph.D. researchers who don't have to repeat something that they've done before or that they could do much quicker. So again, the number of tests that they get and the number of simulations per measure were the objects and measures. So that's how this all worked out. One of the things that our team did was go into a map, a process map of what was going on in the organization. And I want you to imagine the map you see on the screen now without all the circles and lines drawn around it. And each coming up with that process map, they considered that we had done a superb job because they'd never been able to figure out what their actual process was. So here we are in an exact instance of what Lewis was saying before, write it down, and now we have something that's improvable and we can start to make sense out of it. Now the other interesting part that happened on here though from a data strategy perspective is that while we have the process map done, we then add the dimension of time. And so what you see down the right-hand side there are six major categories of non-value added work that these Ph.D.s were doing. And these Ph.D.s did a wonderful job of actually doing this. But as data scientists, they were only about 20% productive. They're time taken up by literally transferring digital data between different machines, moving it around manually, manipulating it, doing what we call thinning and reconciliation. There were tribal knowledge requirements and at the whole thing, the piece of red icons that you see there, it was all being run on FoxPro, which hasn't been a product for many years. So here was a way where we could add a person to their group who helped them with data management practices and the entire group got to be more productive as a result, allowing them to get back to their strategic strategy that they had, which was, again, when it was run over before, through thesis. So I just want to take one other comment about that, Peter, and make what we've been talking about. So again, here's an organization that's very focused on innovation. While the process here was very inefficient for the researchers, while they found value in that, it was just that they did more research not that they were necessarily more efficient. Again, it seems like a subtle difference, but if you were to come in and tell them, you know what, I can make this really efficient for you so you don't have to do all these things. That's not where their head is focused. Their head is focused around R&D and results and being innovative. So again, I think it's very subtle, but it's important in that that's how you understand that that business operates and therefore how to bring opposition to them. Cool. In a data strategy that included elements of process improvement, architectural components that needed to be developed, aspects of data quality that they needed to be educated about, and giving them a more integrated system development process, it reduced the number of tests that were needed to get them to product. It increased the number of tests available for researcher, and it reduced the time to market for new product development, and it worked. In this case, I kind of looked at Lewis and said they told us we saved them $25 million. We didn't charge them an eye. We left some money on the table there for sure. So just to summarize this section on measuring business value, this concept really is a skill unto itself. So if you're looking to bring someone onto your team, it has background in statistics, operations, research, finance, someone with similar experience, it's not a step that can't be skipped. As I said earlier, you need to become familiar with these concepts. You need to practice these skills. Again, I think the important point is that you can't really develop the data strategy and the solution, and drive business value if you're not sure how the business is measuring that value. It's important that if people can understand these concepts you can't endear you to your business partners. All of a sudden, you're speaking a common language and you're understanding how to measure that value so you can help them measure their own success and you'll trust that the data strategy is really important to the business model, which ultimately, you're going to need. So moving into the solution to address needs, we're really going to outline the structure of this section now that we understand that the business needs and the measures of success. So rethinking that as BLP, this is really saying that you've got to think about this difference. And I really love how Peter has sort of formulated this idea. But then we look at the holistic part, right? This is the data strategy. I think this is the piece that most people are probably familiar with, to some degree, is okay, well, we have a holistic solution to be able to better leverage data. And so we're going to go through that. We're going to go through it kind of at a high level of what all the components are in the description and examples. But at EDW, we're going to go into these in much more depth and they're all integrated together. And then the last one is just a quick point about matching your organization's abilities to deliver. This leads into the roadmap concept. So to be successful, you can't over-promise. So the first piece around this is that we've been teaching everybody in IT how to hold things in correctly. And that's a pretty brave statement for me to make, but after studying this for 30 years, I'm really convinced I hope you'll agree with me after these two slides. It makes sense that we should develop some specific goals and objectives so that we can see whether we are on track with that strategy or not and whether that strategy is in fact achieving the results that we want. And it does make sense to the next level of thinking to say, let's talk about systems and applications that will help support the achievement of those goals and objectives. Network and infrastructure becomes a next step on that, and data and information usually become an afterthought at that point. For example, if it's step three, we decide that the answer is an ERP, then every discussion in the network in the data part becomes about the data for that ERP and the network for that ERP. That absolutely guarantees that the data is going to be formed to the applications instead of around organization-wide requirements. We spend so much time reformatting our version of customer for the 16 different definitions that we have around the organization that the processes are narrowly formed around those applications, and hence very little data reuse is possible. Actually, to the old days, the object orientation was supposed to be about reusing data. Realistically now we're simply reusing data, actually reusing code. Realistically, every measurement that we have shows we are not reusing code, which would be easier to reuse a piece of data as well to find or a piece of code that's rather poorly described, and the answer of course is data. So, here, it's a very subtle one again, as Liz mentioned before. If you, with your strategy, both gives perfectly fine, but next let's develop organization-wide information requirements. These are assets that the organization needs to manage. They've put a lot of time and money into getting them to that stage. Let's make sure they actually can. Then we can do network and infrastructure, and then only then should we talk about systems and applications. This is a cleaner approach, less complicated approach to systems development than the way we talked about it before. It means that the data assets can be developed from an organization-wide perspective that the system supports the organizational needs instead of the other way around, and we can maximize our data and information reuse. If we don't do it this way, the only part of the traditional systems development lifecycle is getting to produce more small piles of data about what we don't want to have. So, Peter, again, I think these slides because, you know, in this information, this approach, you know, I talk all the time when we're talking to new clients and existing clients, and it's just hugely powerful, but in the context of the data strategy, it clearly, you know, getting information and being clear to the goals and objectives and the way, you know, now it makes more sense. The data strategy now makes more sense, right, because it is the thing that falls out of the business piece that sits at the top. Look at this from that perspective. We also have to talk about who's involved, and I think the leadership should be there. Some sort of reports of the business lines that talks about how data as an asset should be used to achieve the organizational objectives. In addition to this, this is a relatively new development. I say that in the sense that we haven't been having data governance conversations around this. We used to call it something different, but governance has actually gotten some very good traffic there in the field. We also have a bunch of different stakeholders, our CX levels, CEO, CFO, COO, et cetera, et cetera, lines of business, senior managers and team leads. All of these are going to be stakeholders that are involved in the development of your strategy. Finally, of course, you need to have the data team. Again, architects, modelers, developers, analysts and stewards. Stewards in particular here, because stewards are the ones that are going to be charged with implementation of the strategy in the very near future. We've seen recently a need for a new individual called a chief data officer here. The chief data officer is the person who can guarantee that the applications group doesn't need to be developing things until they have organization-wide definitions of the data. One of the questions I get here at Data Blueprint often enough is, how do I make my agile processes go faster? And the only way to make agile speed up is by taking data out of it. That's another whole topic we could talk about that some other time. But again, what we have is the truth analysis process is told incorrectly to everybody who's taken it. So we've trained smart people that their job should be coordinating with the top information technology jobs through a governance organization and that will allow them now to start getting better at the process. Again, our focus here on the CDO is that they should be dedicated 100% to data set leveraging. They should be unconstrained by an IT project mindset and report directly to the business. Peter has a book out there. Thank you very much. It's the case for the chief data manager. So we're going to be talking about the different really components that make up a data strategy. Peter has talked about people. So data, determine what is important. This is a really important concept and sometimes not understood. Not all your data is important. Using the business needs and success criteria we've talked about will help to identify the scope of the data needed as part of your overall strategy solution. Peter listed the metatypes of data as we think of them. I'm sure someone else could have a different take, have a lot more, maybe slightly differently, but our collective years of designing, developing systems under different circumstances and context. These are the ones that make sense to us. It always comes back looking at data in these buckets, whether it's architecture, governance, or quality control. We have a strategy that is really fully focused on the consumption side of these metatypes. A lot of times it's looking at reporting analytical data and metadata and maybe some of the master data. We're proposing that the data strategy really needs to consider all the metatypes, data, event data, as well as reporting analytics. Master data impacts transactions and events. Much does reporting analytics. The point here really is that all the metatypes are really interrelated to creating a holistic ecosystem and the data stride needs to support that concept. Peter, I thought you invented Rod, apparently not, but if you could talk about that. Rod stands for data that is redundant, obsolete, or trivial. What that means is that it turns out when we measure it, 80% of your data out there is redundant, obsolete, or trivial. It just gets in the way of your decision making process. One of the things we can help to do with the data strategy is figuring out what data is important, why it's important, and what we have to do to make it more useful around this. Again, if you don't have that handled, it's probably five X times the size that you'd normally like to do it. While that's fine for consultants, we tend not to get paid by the hour, we get paid by the project. That's right. That picture is on the slide. It's from Rod. Now we've talked about people and we've talked about data, management practices. We're going to walk you through all the components of the technical foundational practices. You can see there the wheel on the right hand side. That's right. The important point is to understand which of those components are needed and to what degree they needed to support the data strategy. We'll talk about those. These are integrated concepts and they need to be thought of like that. Today we're going to walk through each of them and introduce each one. We'll be diving into these a lot more at the workshop at EDW and showing how they work in concert together. Just a quick note, there are other data management practices covered here, data security, business continuity or two clearly not mentioned. These are components and this view could be expanded to include those, to support your business and to prevent something that's been on the forefront of newspapers lately. The key for these foundational practices is to find when we go in and help groups either untangle the mess that they have or plan something to focus too narrowly on our at least three pieces. People get fixated on the data warehouse and they think the data warehouse is going to solve the problem. We've seen lots of very low-function data warehouse failures because of the functions but they've forgotten the other pillars that go into this. If you're not incorporating at least three of these practices around, you're probably not seeing enough of the picture on this. Healthcare data warehouse at this point and it took about 18 months and $13 to build this and we moved into it because we were asked to take a look at the value of it. There were almost two million members but almost one and a half million providers so it was almost a provider per member. That was a great state to go live in. On the other hand in this data warehouse 800,000 of the 1.4 million providers had no primary key which meant we couldn't access them and we were still this entire data warehouse ended up having one user. The volume and forcefulness with which the CEOs said to me I could have taken a room full of MBAs and accomplished this analysis much more quickly. Just to keep tying it back to the theme of the data strategy and tying this in with the business, we find that if it's a solution for ever to implement or if it's overly complicated it's just because you don't understand why you're doing it. You really don't understand the drawbacks of the solution. I'm almost always if I walk into a new environment and someone is struggling to implement some system for two years and it's taking forever to do it it's because they don't know why they're doing it and what the requirements are. It would be a great solution that would have been wonderful two years ago. Okay. The foundation practice is data strategy is one of the foundational practices. Again, we've already talked about aligning with the business strategy and the operational model and this is going to become even more important as we do this. Gosh, again another call that we get here at Data Blueprint a lot is people want us to develop a big data strategy for them. We'd be able to do that what's your data strategy? I know and we don't care and I know that's the wrong answer. Again, we want to be very careful in terms of working through this. You don't want a big data strategy if you don't have a data strategy in the first place. Another foundational practice then is architecture. Again, I'm probably preaching to the choir here but when you explain what data architecture does to the organization if you talk in terms of developing a common vocabulary so that when they talk about certain concepts they know what they mean and the purpose of the data architecture is to help you get on that same page. Our observation is that most organizations have data assets that are not supportive of the strategy so the question becomes how can they more effectively use their architectures to support the strategy implementation? I'll repeat to all of this 10 years ago we might have said well, we have data governance covered in data administration but it's been a lot more successful since we brought it out explicitly. The key there is to avoid data governance and I've got a fellow in my class one of those credit card companies as well who says, oh gosh, you're giving me more reasons to say no to people. Well, that's not what governance should be about. It's not about saying no. It may be a sense that sometimes you have to say not yet but it's really not where you want to be. Governance is the idea that we're going to be treating data as an asset. After all, you have fiscal governance in your organization. You have the disciplines that you put in place that we need to make sure these are the right ones. Again, there are two components to the data management practices, the foundational and the technical practices. I'll introduce some of the technical practices here. They're all about building and stuff and how to go about it in a holistic way. We keep talking about thinking like an engineer using architecture and you've done all the other part of the strategy well. This starts to become self-evident. It needs to happen. Again, if you're struggling to figure out what the technical solution should be, I will go back and look. It doesn't mean that there won't be technical challenges. It just means if you're having a challenge figuring out what you need to do, then that should be a flag. As we talk through this, we'll talk about some examples as well. Data management is one of the technical practices. This is where knowing what data is important is crucial to data quality because we're talking about at the very beginning, huge amounts of data in our organizations. It's coming in and it can't all be standardized all the time. If you want to put controls and governance in place, you need to do it for your most important data element. This is that kind of way that we've ever seen being really successful with data quality. But at the same time, you'd be really surprised about how much your organization would run. There are two key data elements with great controls and governance around them. You could take something like a customer ID. If a customer can be uniquely identified, consistent across a whole organization, that organization could probably just change in so many ways that they hadn't even dreamed of. It's not like you have to fix everything maybe about the customer, but if you could just identify the customer. I've worked with a particular company where the customer type, as well. Someone who managed customer data didn't report up through the executive level. If you wanted to add a new data type for customer, you had to go through about a three month adjudication process and it was a big deal. But because they could do that, they could share customer data, they could roll up customer data and it really was a beautiful thing. The other key point on this slide is, and I don't know how many people get into the situation, but if you do data quality problems and you figure out what data to manage, the find and fix is just while it may seem like you can make progress in the short term, over the long term, it isn't sustainable and you can never really get ahead of it. So it's worth the effort while it will work. But it is a good way really to deal with data quality issues you want to fix. There's a requirement that is not typically well done. So when we talk about a data strategy, we don't tend to express the explicit integration requirements that are there. We kind of draw some pictures up and say, this stuff should all talk to each other. But talking to it is not really the same thing as integration and data turns out to have it perfect at the most atomic level for it won't integrate. Again, a half a character or a number off is going to result in a different object being pulled. So the idea to gain insight is to gain integration and the broader we can focus our integration efforts, the easier it becomes to achieve our data strategy. Data platforms, so decoupling the functionality using these engineering architectural concepts, we keep talking about. And we have two examples. Peter will talk about one in the cloud that is related to migrating from legacy platforms to newly engineered data platforms. For example, I want to talk about its own winding and a mainframe in a high volume transactional environment. The business strategy for that was that they wanted to efficiently process those transactions, high transaction volume. So the more people touched the transactions, the more people had to figure out if they wanted to touch the transactions, the less efficient they were. So the business strategy there was for operational efficiency, and that translated into straight through processing concepts in terms of the overall design and architecture. So processing of transactions minimized the amount of human intervention that's required to process these transactions. And so what you have to be able to do then is you have to then enable those transactional really a set of data to figure out what data components, whether it be exception management or some type of life cycle management or master data, they would be able to do that straight through processing. So those data platforms really driven from your business strategy become very evident. One of the examples I remember you told me earlier on too was you worked with a trading system and the link was not so much the number of trades they could do but the number of errors they kicked off that they couldn't process so well. So by reducing the errors you were able to increase the volume of that trading system by a very significant amount. The errors were being created essentially because master data was wrong and they didn't know really whether it was an error or not. Just another bit about platforms here. You'll see this a lot where people are trying to get their data into the cloud and the typical word that they use to describe that process is forklift. There are a couple of other words that you use to describe it too but it's just taking your data and jamming it into the cloud. It's great for the cloud vendor so we say if you're going to put your data in the cloud it should have three attributes, data in the cloud that it doesn't have outside the cloud. One there should be less of it in volume just based on the rock that we've talked to you about. I hope you all agree that the data in the cloud should be cleaner than the data outside of the cloud if it's not what's the point. And finally it should be engineered to be more shareable and those three attributes make a huge difference in terms of platform. So here on the technical practices business intelligence so we really think that this is reporting in analytics. This is a spectrum of really that consumption. It's at the end of the food chain really of the data and this is clearly highly dependent on the other aspects right. This is all about integration. This is all about platforms, quality analytics so it's obvious when this is not done well right because spread marks everywhere and there are shadow IT solutions built by your operational staff using Microsoft Access. So you can pretty much walk into an environment and determine very quickly how the business intelligence solution has been developed if people are having to create a better beginning to end the system every time they need a new report you know that it's not working but clearly this is exploratory nature. Business intelligence tends to be when you want to develop an agile method you put things in front of the users having them react because typically once they give answers to one set of questions they begin to have a new set of questions right behind that. Again, this is something highly dependent on governance, quality metadata in order for it to work really well. To all this is that you've got to have an idea of how your data strategy is going to interact specifically with your business and the goal here is that we want to tie it to specific business processes. A tool that some of you are familiar with is a CRUD matrix but let's talk about how the CRUD matrix actually works. The CRUD matrix talks about whether the data in the center of the matrix is created read, updated or deleted by the various business processes that are going in this case across the top of the matrix. So when we do this we understand scope, we understand user data, we understand life cycle pieces all these things are important to them and your strategy can get down and say we're going to start in these processes in one form or another. Here's an example on this as well which is that when you're trying to decide what technology selection to get again a platform piece typically what happens to most of the data issues in there is that when you discover the software doesn't match your organizational practices you then have four choices you can change the software, you can change your business practice, you can do some combination of the both or you can ignore the problem but once you get guidance to do this data strategy would have revealed the problem in advance of selecting that particular approach. So the example I'm showing here is one that we've used a lot. A person can be an employee a person can be multiple employees. In this case we're explicitly building a system that permits moonlighting but also in that it's going to give you simpler reporting again we can do a lot more on that we won't. The key is that as you understand these business practices around this you can take these components of data management and I've got them listed here program coordination integration stewardship development and operations and start to match them up because if you don't match them you end up with some numbers like this and if you're just looking from one particular group here where we have lots and lots of things that are happening, if they're happening a lot and not optimized then your organization is going to suffer death by a thousand cuts. So when we summarize this particular piece what we're really trying to do is to get people to think differently about the solution not how we've taught them. It's a holistic process focusing on people management technology and processes that addresses the gaps to give sustainable solutions and most importantly it's matched to your organization's ability to deliver cross walking and running which is our final section we're getting into. So develop the roadmap and plan and again we're really introducing the topic here with the expectation we're going to get into it in depth at this point in time. So really when we're talking about the implementation plan we're not saying that a project plan with the start date and the late finish date we're really talking about the approach to implementation which is a long term vision for this will be executed what are the key milestones again the important part here is it needs to be achievable and just to get the organization's capabilities into this time and time again go into organizations where they're just trying to do too much their level of capabilities. So Peter his crawl walk one concept and how that's applied I think is absolutely perfect and we have found people have been burned by trying to do too much they come back to this and it makes complete sense. In terms of crawling and this is the message that you want to have in your strategy too is that we haven't done this in the past we need to start but we're not going to go from 0 to 60 miles an hour in two seconds so crawling is identifying the opportunities and determining a just larger scale scope allows learning to occur again remember back to the four quadrant model and focus on developing the ability to take data management and derive specific tangible amounts of money that people in the corner offices will pay attention to walking then maybe doing that particular process optimizing it and then running is the fun part where we get into all the analytics and stuff like that and by the way if you've saved in a 20 in the quadrant 2 the quadrant 3 pays for itself college upon these miles in the last point being the foundational component to be developed like concurrently executing tactical solutions so I think for the content for the top of the hour now it's time for your questions here and again this will be exceedingly more detail at our session we hope to see everybody in Austin because it's a wonderful fun town to have a lot of fun and interact with hopefully about 800,000 800,000 data management seems like 800,000 some days. We'll get in there on Sunday afternoon we don't leave until Thursday and it's a full week so Megan we'll get over to you for the questions Thank you Peter and Louis that was an awesome presentation now it's time for everyone's questions it's time for the Q&A so just click on the Q&A window feature at the top of your screen you should be able to submit your questions through that Q&A window and we've had a few questions already on them so we can go ahead and jump right into it the first question is what are your comments on an IT shop which is completely cutting data specialists out of the picture because the only thing valued is speed of application delivery what are the consequences of this approach can it succeed we agree that this is a shooting yourself in the foot sort of strategy one of the things we talk about is putting the eye back in IT and when you throw your data specialists out you're really focusing in on T and T can do things really really well but it can mess things up really really fast as well so we definitely advocate an approach that says make sure you've got qualified individuals in your organization to do this so we're about to move to go back to the slides this is my first time you know what number it is now I don't but it's a great question and see it all the time it gets back to these slides here so this is really what I want is that I think the question is around this this here where I can't get to the right one it's where there you go the challenge to me that just screams that the organization is struggling to see the role the data plays in the organization and so they're looking at it as being something that comes after the systems application development easy to that I think I would take these two slides and see if you could even have some conversations around that about getting data to be the thing that you focus on because as soon as you buy SAP application then all the other conversations from there on are about what SAP does and doesn't do and how it does it and what does my organization need regardless of what we end up purchasing who ever asked that question if you'll identify yourself I'll send you a pre-production copy of the CDO book there's a lot of good arguments in there see if that doesn't help turn things around recommend that data governance ownership sit in the IT organization our ideal approach is that data governance is a shared function between business and IT but to correct for past errors we advocate that the leadership of that organization report into the business it's not just that the IT people don't know what they're doing but it's that if we can help to educate the business and make them understand that they have a stake in this and that it doesn't help them then definitely we want to try an error in this case to correct for some of the past errors where we've had done a question from the person before and they're indicative of this they clearly didn't see the value that these individuals were producing and they're de-emphasizing it where they can measure the lines of code that are produced by their agile teams and they like that one thing and this isn't directly related to the question but kind of and I wanted to mention the presentation a lot of times we deal with data governance we apply the CROC one concept and the CROC so what we end up doing is because people we find that people will go away and create an 80-page document about how the governance should work and who's going to play what role and it just becomes it's just a very large task without really having any practice at governance at all so we think it's important to and crawl as an example would be to pick an element that the business and IT and the data team can work on together maybe it's from our ID or maybe it's some other key data element and what are the governance rules and what are the business rules around that and sort of with some little g and build into the larger g over time we can extend that a few more further we had a hospital group that we worked with recently that decided admission date was the place that they wanted to start and they found about seven different definitions of admission which was kind of low for an organization but they also found that by using multiple definitions of admission date the hospital was leaving literally millions of dollars on the floor because they weren't able to correctly bill for the services that they were putting out again I think going back to even though we just have a hard time when they look at their whole inventory of data assets of trying to to figure out how to tackle the whole thing once you start to look at it you'll find that five percent or less of your data can make a huge difference in your organization it's all about leverage we actually had somebody ask the question for the unable to attend the conference how can we get the presentation slides for the conference and pretty much if you want to come back to or go to the data blueprint website check back and have the slides posted closer to the actual conference so end of April so just keep an eye on that and so another question for you guys is what is the best way to develop business relevant analytics that can be put to good use so that should be one of the goals of strategy is to say not just we're doing data for data sake but in fact this data strategy is designed to achieve these business values and those objectives are tangible and meaningful to people in CCCLE and I would suggest that even the concept of we're going to be exploratory and we're supposed to be discovering new things that we didn't know before this has some directional component to it direction from a business strategy perspective so just to do big data for big data sake where people built a data warehouse and they don't know why they built the data warehouse they wanted it for reporting and they wanted to be able to you know have better reporting basically and it ended up they didn't know why they built it and therefore it wasn't adding value so I think I just can't overemphasize the crawl walk run approach to business analytics especially business analytics because that is very much an exploratory put something together, get some feedback and so that means a lot of foundational work needs to be done so that you can get into that sort of rhythm with your business users of presenting them stuff seeing what questions that that answers and find out what new questions they have but they're going to want to be able to trust the data they're going to want to be able to see an integrated view so doing some of those foundational things will put you better I think to get into business analytics and to be able to work with your users more easily and if guided by a strategy another important component can be the reason we're doing the reason what we're doing today is different from what we're doing in the past is because in the past we used to follow an old SDLC and that produced a certain level of results but if we instead for example put into our strategy that we are going to fail three times in the time that most people would take to build it once we will be better at that particular process because we have the opportunity to iterate where other organizations try for a big bang approach and again miss it entirely which then people say well what the heck was the point so and Meg if I could just continue to make this a long answer and I think that we're beating the whole tier of data strategy but you know I think the important thing is especially with business analytics is you know there is something important to the business and one of the things that's important is the business if the business is measuring you know it's important so that's the thing that you're looking for that's the thing you want to go after when you're looking at business analytics what does the business measure how do they measure their success and they may even be doing a spreadsheet but that's the thing you want to go after I would refine that a touch though because the things that are measuring are things that are important to them what we need to find out is what they value after those measurements though and so again get up beyond that first layer of management where you're collecting these aggregate statistics the example I use many times is that if we paid by the amount of orders that they had every order would contain one shirt because people are very fair to maximizing their particular measures that may not be the measure that you want to use although the things that are measurable should be the candidates from which we take the constructs and put them together okay great the next question is you spoke about a common language in semantics is a dirty word in the current IT environment so I think the answer to that is generally yes and it shouldn't be canonical sounds like it's sort of biblical and you know people don't want to get into that because it implies that there's sort of a war about arguing about definitions and that's not what we're talking about what we are talking about is practicality saying if we're not speaking the same language we will run into some problems in there I agree Peter with what we're saying again what we tend to see in this industry is that people will get fixated on an idea five years ago we might have said that master data management was falling into that same category everybody's got to have master data management I can remember explaining the concept to a wizened friend of mine who says oh you're just implementing the concept that some data is more important than others yeah absolutely we should do that we have a master transaction file that would go all the way back in the 1950s to use on this sort of thing so by the way master data management didn't ruin anybody but I've seen a lot of failed master data management implementations we would never try to take you down in an analytics path unless you had some concept particularly the straight-through processing that Lewis was talking about that's impossible to do without master data management absolutely okay you can't make the conference it's actually a video only that you can do if the funding isn't an option so there's yeah that's pretty good who is the capital for metadata management that's a good question and the short answer is everybody but of course that doesn't help that's why we got into this situation in the first place I like to say that the language of data governance is metadata we're not expressing your data governance goals and objectives going all the way back into your strategy and metadata management in terms that are specifically identified as pieces of metadata then you risk that disconnection the gaps occur things start to drift if again I can't say it any simpler than that the language of data governance is metadata perspective and for everyone in the things I've built metadata when you're doing this intelligence reporting at some point you're so much touching the business that your business has such a big role to play but basically in all your data that really is part of that common language and semantics and figuring out what the data really means to the business is in that layer and it's when the data is being consumed I understand there are other types of metadata that can help drive processing and help you manage the overall environment from a technical perspective but really where the business value is when the people are able to consume the data is that that point is where the definition needs to be managed and there's no way around that the business has to be so involved in that or there's no way that a data person or IT person can really define the metadata for someone else there's a little more piece on that Louisa and I keep meaning to make a slide on it but it turns out the business actually understands metadata a lot more than the executives in the organizations do and I'm not saying bad things about the executives but you'll find that when we go out and find these shadow IT shops in the business you know the guy with the people serve others desk who says you caught me but please don't tell IT that I've got this system running here that individual absolutely understands metadata they may not know the name for it when you start to talk to them about it so very definitely the business is largely responsible for pushing some of these topics and also for providing the actual words of the vocabulary that we're going to use in there. Question is how long would it take to go through the foundational stuff for a departmental group of a company to have that manufacturing work? So the engagement process should be something along the lines that Louisa has described before which is to say first of all what is the business strategy in there hopefully there's an IT strategy driving that and then we want to look at what the format of the data assets is what are the characteristics of the assets that they have taking that particular process learning from the strengths and weaknesses of their existing data pool and everybody has some strengths and weaknesses we can then come back to that IT strategy in conjunction with the business strategy and say how do we leverage data in there best for that organization. The process can be done in literally a number of weeks I don't mean full time weeks but it does take some time to assess and to assimilate this. Just to get the metrics around that Peter because this is our bread and butter in many ways so in a large organization that has the world operates in Silas so it will have multiple business units and even if there's other products or administrative Silas but in a business unit in a large organization we'll spend anywhere from 8 to 16 weeks developing the overall strategy. You're just looking at foundational pieces you know it could be potentially a subset of that but still that up front work of understanding business needs and measures it still has to be fully done it's just the solution that you actually develop could be smaller and I think it's a really good question because a lot of times we don't ask to do strategies we are you know how much time for me about time you know but it's very focused around a particular area and it will be clear to them why they need architecture or it can be very clear to them why it's mostly around all those things will have multiple components in the foundation of technical components but it becomes clear that there is kind of a focus in on what they're looking for what is needed but we still you know we just will lead up our customers until the point we understand what it is they're trying to solve from business perspective or we'll say to them we don't know how to do that. Now we do have a question it's what are characteristics to get an organization to begin to put business rules data standards data models and writing with transparency? Data is an organization's sole non-degradable non-degrading durable strategic asset and if you unpack that particular phrase you'll understand that data is the only asset that organizations have that doesn't degrade over time you can't use it up and it's a durable asset which means it needs to be treated differently from a financial perspective than a non-durable asset or in this case an office consumable type of a process if that argument doesn't get to them then I'll go back to the CDO where I ask a couple of really insightful questions like who's if nobody's doing this organization how is this function occurring how are these types of things going it's kind of a difficult process but there are a couple of key questions that you can ask that will really start to identify that. So I think you know and I guess we'll assume that's right data strategy is a top-down approach to improving um making data more manageable within an organization right and sometimes organizations just aren't ready for that and if you're looking for particular areas you know to me that almost is more of a bottom-up approach and it would certainly understand that you know and that is a valid approach as well um you know and so some of the things that we look at are when you talk about standards or quality and what's the carrot is that typically the low hanging fruit is operational efficiency so we call it non-value added work so if you look for those areas where you know in our example of the PhD researchers at the chemical company you know I'm not kidding 70% of their time was spent manipulating the data just so that they could go be big data scientists 7% of guys who god knows how much they make you know it's been manipulated I mean when I think manik well you saw the diagram so um the typical low hanging fruit is you can start to give them a carrot you know I can eliminate so many hours a day across the team of 20 people and that adds up to this much money those are typically the easier things to find um but that isn't you know um you know that's a bottom up approach as opposed to the top down strategy approach both can work the top down is more favorable yeah yeah questions but um some great questions came in from you guys it was a great presentation um thank you everyone for participating in today's event we hope you enjoyed it uh thanks again to data diversity and Shannon for hosting us uh once again you will receive today's materials within the next two business days our webinars will be emerging trends and data jobs hopefully you'll be able to join us for that as well and as always feel free to contact us if you have any questions have a great day thanks Megan for another fantastic presentation just let me love and thanks everyone for attending and for all the great interaction and questions and again as Megan said I hope everyone has a great day