 Hello and welcome. My name is Shannon Kemp and I'm the Chief Digital Manager for Data Diversity. We'd like to thank you for joining today's Data Diversity Webinar, Data Management Maturity, Achieving Best Practices Using DMM. It is the latest installment in a monthly series called DataEd Online with Dr. Peter Akin, brought to you in partnership with Data Blueprint. Just a couple of points to get us started. Due to the large number of people that attend these sessions, you will be muted during the webinar. If you'd like to chat with us or with each other, we certainly encourage you to do so. Just click the chat icon in the bottom for that feature. And for questions, we will be collecting them via the Q&A in the bottom right hand corner of your screen. Or if you'd like to tweet, we encourage you to share highlights or questions via Twitter using hashtag DataEd. And to answer the most commonly asked questions, as always, we will send a follow-up email to all registrants within two business days containing links to the slides. And yes, we are recording and will likewise send a link to the recording of this session as well as any additional information requested throughout the webinar. And joining Peter today is special guest, Melanie Mecca. Melanie is the CEO of DataWise Incorporated and was formerly CMMI Institutes Director of Data Management. As a managing author of the Data Management Maturity SM model, she created an assessment method for evaluating data management capabilities and has delivered incisive actionable results for 25-plus organizations in many industries. Melanie led development of courses in computer-based training for the CMMI's Enterprise Data Management Expert certification and has produced CBT for data stewards across all agencies in the state of Arizona. DataWise is a certified CMMI institute partner helping organizations to accelerate their data management strategies, programs, and staff education. Melanie offers EDM consulting, DMM assessments, and CDO advisory services, teaches certification and custom courses, presents at conferences and webinars, and authors, white papers, and industry articles. DataWise partners with industry-leading practitioners and companies including Data Blueprint. She also contributed to the DMBoc 2.0 and presents seminars and workshops for data chapters. Now let me introduce to our main speaker for today, Peter Akin. Peter is an internationally recognized data management thought leader. Many of you already know him or have seen him at conferences worldwide. He has more than 30 years of experience and has received many awards for his outstanding contributions to the profession. Peter is also the founding director of Data Blueprint. He has written dozens of articles and 11 books. The most recent is your data strategy. Peter is experienced with more than 500 data management practices in 20 countries and consistently named as a top data management expert. Some of the most important and largest organizations in the world have sought out his and Data Blueprint's expertise. Peter has spent multi-year immersions with groups as diverse as the U.S. Department of Defense, Deutsche Bank, Nokia, Wells Fargo, the Commonwealth of Virginia, and Walmart. And with that, let me turn everything over to Peter to get today's webinar started. Hello and welcome. Welcome to you, Shannon. And welcome, Melanie. It's always a pleasure to be working with you. We actually call this, if you guys don't know, Melanie and I dance together on these things. So literally, you could wake either of us up at midnight and we'd be able to redo these things. That's what a great partner she is to work with. So, Melanie, welcome. And Shannon, thanks for hosting us as always. Thank you so much. Absolutely. So we're going to take you through a little bit of history here. And most importantly, address a question that many of you are facing, which is that, okay, if we want to do more with our data, is there a framework in existence that does it? Wouldn't it be nice if somebody had gone to the trouble, taken the time and effort to put together a plan that you could follow so that you could look at this plan and figure out where you are currently, what are the good things about what you're doing that you ought to preserve, and what are the things that need improvement that you could then turn around to address? And that's the overall process we're going to look at today. We're going to talk about motivation because we're assuming if you're doing all your data perfectly, you're probably not on this call. We'll talk a little bit about the research background here because it is an interesting little bit of side notes. But we'll dive in and spend a fair amount of time working with something, while you may not have heard of this before this seminar, your bosses have. And this is one of the most important aspects about the CMMI framework is that they teach it in MBA school and almost everybody at our level and up in organizations has heard of this. So if you get nothing else out of this seminar, when your boss says what are you doing, I don't understand, just say we're doing CMM for data and they'll go, oh, I get that. That's probably a good thing. Now let's talk about how to get here. So again, dive into the motivation here. Melanie, I'm just going to turn it right over to you and let you get started on this. Okay, so the Data Management Maturity Model was published in August 2014 and it uses the same architecture and approach that the CMMI for Development Model used. And that model has been used by thousands of organizations over the past 30 years to improve software development and engineering practices. So the little lights on the path in the woods there are showing you lights on the way to the path for improvement for your data assets. So the model is architecture and technology neutral. It doesn't matter if you have mainframes or a data lake or a huge warehouse layer, whatever you're running, it's okay because the data is the data. So it's also technology agnostic and independent of what industry your company happens to be in. So the use of this model and we're going to get into the guts of it a little bit. Why is it useful? How is it constructed? But it really shows you what you're doing in terms of fundamental data management practices and also what you should do next because inherent in the reference model is a built-in path to capability and maturity for all organizations. So you use this model to get a baseline for how you're doing to help you manage your data as a critical asset. It helps you greatly with doing what Peter's new book talks about building a data strategy that makes sense and is well supported by people and management practices. It helps you accelerate your enterprise data management program significantly engage stakeholders to create enthusiasm for governance and also to get the low-hanging fruit for what you need to fix and do better as well as the strengths you can leverage and the projects that are best for your organization for rapid progress. Next slide. Yes. That silent, silent pause. So I have to look at the picture first because I just love this picture. If you're familiar at all with Buddhism, this is the journey from Samsara ignorance to Nirvana bliss. And this is very similar to an organization's journey with its data. So you start out the data is complex. It's not documented. No one understands it. They don't know where to get their data. They have problems with data quality. They have multiple places where let's say client data is created a mess. It creates fear and confusion. So at the bottom of the picture there's this little person running from a mad charging elephant. And then as you improve your practices and more people get engaged and there's more collaboration and harmony and sensible disciplines that you're applying eventually you become an elephant, becomes tame and you see at the top the Buddha riding on the elephant perfectly enlightened and everything working well. So that is the underlying purpose of the DMM to get your organization to where you are using the data very effectively and it's great data. So you want trusted data that your internal customers and if you provide data to external customers everyone is confident in their data. You want to judge your risks more competently and make quicker and more incisive and more accurate decisions through analytics and predictive modeling. You also would like to get cost reduction and operational efficiencies and because this is a process improvement approach it excels at cost reduction and operational efficiencies over time. For example many organizations choose the DMM after a few years they decide that they're going to move data quality improvement initiatives from systems or even programs with multiple systems and create things like data quality centers of excellence which is much more efficient and gets the job done much more quickly than leaving it to project by project by project. And finally if you're in an industry that's regulated or you have extensive responsibilities to outside organizations this proves to your external customer base like your regulators that you are doing a good job that you're a good steward of the data and that it meets their requirements. Next slide please. And really key for that Melanie too is to understand that organizations are accomplishing this on their own right at the moment. And that's a very inefficient process as you mentioned the idea that every group is having to learn how to improve their own data management practices individually is a tremendous source of inefficiency within many organizations so this gives you the ability to pull it together get them all marching in the same direction. I have to ask you that Melanie I think you blame this one time but why is the elephant turned from blue to white in the pictures too? Because it's base nature becomes purified so you could look at that as enterprise data architecture and really high data quality. There we go so white elephants are our data quality that's probably not the matter for any of you. Anyway it's a beautiful picture and it does absolutely satisfy this. So the key of course is that most organizations and if you've heard me speak before on these issues you know that the primary place I blame is the college and university system that doesn't really provide a lot of directions on this. So Melanie and I both are friends of Lewis Carroll and the treasure cat at the fork in the road is a perfect metaphor for this. So Alice comes one day to the district cat and she says cat I want to move our data management practices to the next level and the cat says well what level are you at now? And of course if you don't know you can't measure it how are you going to manage it effectively how do you know where to put the time the money the effort into these programs without some guidance around this and this is really the beauty of what Melanie has created in this process to do this. I had a job in the late 80s through the late 90s where I had the title US Department of Defense reverse engineering program manager and we sponsored research at the software engineering institute at Carnegie Mellon University. This group up there were very good at doing certain things and we asked them specifically to come up with a process that would help organizations understand how well people worked with them. So the real question was how do we measure the performance of DoD and our various partners and I was told to go look at the Navy because they were doing some really interesting work there and I went down to the Navy and it turned out Clifinkelstein and John Zachman were working down there and that was my introduction to my own personal revelation on all of these things. But the SEI came back with an integrated process and data improvement approach and data was not supposed to be part of this because they were the software engineering institute so they literally told the SEI to remove the data portion of the approach and what it grew into was the CMMI eventually becoming part of the DIMBOK and things like that and I ended up picking the pieces up with my friend Bert Parker who was working for MITRE on an internal research and development process. He's passed on since but it was a truly great collaboration that we had there working on this, basing these pieces that they had put together and pulling some key process areas and what you're seeing here is a reference to an article that we published many many moons ago that described the original piece which was that we did a survey of about 500 companies and how well they were doing and the answer was not quite as well as they should have. Again if you have trouble sleeping at night I'll be glad to send you a copy of this report but it's not really relevant at this point because what we were doing there was a bunch of good people who were trying to do some good work but it certainly took Melanie to come along and make this into a professional category here and Melanie I'll let you take over. So the CMMI institute spun out from the Software Engineering Institute in December 2012 and took over a lot of the models and development of IP and services delivery in the partner network which is over two partners and the DMM was part of that. It was just in its post content model period and we all moved over to the CMMI institute to bring it to fruition and put it to peer review and publish it. And CMMI institute is now owned by ISACA and offers joint product offerings with ISACA which concentrates on IT audit. Next slide please. Big stats on this now. Yeah this is the one about the Big Brother model to the DMM. Showing the same approach and basic architecture that was adopted for the DMM is a monster. It is the most successful process improvement model in the world and if you look at the bottom bullet there they had 1,950 plus appraisals against this model in 2018. Now that's a cumulative to start though right? That's how many went in 2018. So that's among 500 partners in many, many countries. It's excellent growth right now the development model is in South and Central America and China. And India of course used the CMMI. They were early adopters in order to help convince companies that they could reliably outsource software development because their practices were very mature. So it's been very useful. Absolutely and I want you guys to keep these numbers in mind for just a minute. I'm going to flash another screen up here which just shows one of the many pieces of research that shows that when organizations adopt this general framework what you're looking at here is a survey done by the conference board and they ask questions. Now if you look you'll see it says CMMI, ITEL, RUP, COVID and PMI on there. So RUP stands for the object oriented world and RUP, COVID and PMI when they add those capabilities to organizations that are practicing it actually slows them down. The percentage of projects on budget falls and that's not a good thing. It rises under ITEL and it rises further under the CMMI. It also gives us better on time performance as well and I've given you the reference to this particular study. There's a number of them. In fact the CMMI as Melanie said is the world's leading way of doing a process improvement. Now the reason that's important is because you will get approached by a lot of different companies. Melanie and I have both seen and worked for a number of them and they're fine companies but they say well we have our own method and I just want you to turn around and ask yourself why would you use anybody's method that didn't have a 20 year absolute grounded in literature with proven results process of delivering better improvements. And I just I've never found anybody have a good response to that answer. We've also found people and this is something perhaps to keep an eye out for who say they're doing CMMI but aren't really doing CMMI and we'll get to that a little bit later on. So Melanie back over to you. A little question here on DEMBOK and CMMI and maybe I'll take this one. We just are working at both levels to try and make sure that this collaborates and continues to push both of these organizations forward on all this. You want to add anything to that? No, it's just that whenever I get this question which is frequently because I do present to a lot of DEMA chapters I always tell them that the DEMBOK is a wonderful reference encyclopedia of proven best practices aimed at the data management professional. And so the content overlaps significantly but not entirely because there are things like data design of a physical database design, physical data security and other things in the DEMBOK that are not necessarily very accessible to the lines of business. And that's where the DMM its whole purpose is to unify the lines of business with data management and IT. So it's aimed at the lines of business representatives and their leaders primarily. And if there is a reasonable critique that we can make of the DEMBOK is that we really focus more on the allegiance to the pie wedges necessarily than putting them directly into the business context. Yes, and also the DMM was built specifically to be a measurement instrument with a gradated path that's abstracted representing a typical organization's development in data management capabilities. So it's definitely sequential in its overall scope and for that it was developed specifically for that. So you can use the DEMBOK for that but you have to kind of make up the path from the wonderful information that's in the DEMBOK. So they're very highly complimentary and I have the DEMBOK on my desk and participated in the DEMBOK 2.0 and you should use them both. Why should you limit your toolbox? Absolutely. So anyway the DMM is big. It has 414 specific practice statements that are kind of like requirement statements. Do you do them? Do you not do them? Or do you partially do them or they work in progress? And this is kind of exploration that you can do against the DMM. If you use it as a measurement instrument you're also going to find 596 functional example work products that represent the types of documentation or products like a business glossary, metadata repository that are typically created as organizations grow in capabilities. So it's very useful for that too, for core work products that you need. And if you use it you will find that the gaps come out very clearly as well as the strengths and accomplishments that you can build on. And Melanie, if somebody did want to just run onto the internet and buy a copy of it, it's available? It is and I put it on the last slide, the link. It just literally can go to Amazon, right? I mean it's... Yes, I don't think it's on Amazon. I think it's all you have to do is I have the link at the end on one of the slides but if you want to just search for it and be lazy just type in CMMI space DMM and you'll get right there. Bingo. So also this model is behaviorally based. Just like CMMI for development it emphasizes human beings behavior about activities surrounding data. So the model that's trying to help you and your organization bring about positive behavioral changes in what you do with the data. And of course the objective is to develop enterprise-wide effective and repeatable processes so that each project in various...supporting various business lines doesn't have to go it alone anymore. And it saves tremendous amount of time. I spent quite a lot of time as a federal consultant and I've worked with hundreds and hundreds of project teams over my time doing that. And I will say that not only in deciding what interface to use but also in naming data, defining data, the projects in most cases were left on their own. And what does that bring you? Over time bit by bit your data layer becomes less understandable, less clear, less pure, harder to navigate, right? So this is what we don't want. So we want effective repeatable processes and then of course also policies and standards against which you can align what you do. And then the activities of course result in these work products, guidelines, templates, training, etc. When you do have a good set of policies, processes, standards, guidelines that are well thought out and approved by governance, meaning that you don't want to surprise people with these things, you want to do it in a collaborative way, once those are in place you find that everyone is more productive and you save time in development projects, you save time in data integration and everybody is essentially happier and more relaxed, which is very important because that way they can do more creative things. Real key for all of this of course is that I sort of mentioned a little bit before but your organization has learned how to do data management without this guidance. What do you think the chances are that they all happen to learn how to do it correctly? It's not a simple task as Melanie shows here in this picture where we just have to row and make sure we're rowing in the direction. A defining characteristic of a work group is a group of people who exchange data. It's one of the things we do when we're identifying work groups is to identify where the data flows are and the chances that they just happen to all learn this the same way in the same time is mill. So it's a major source of inefficiency and as Melanie said it's not just inefficiency, it also gets into morale as well. Definitely and you know it makes me very happy when people have used the DMM to good effect and then they tell me that things are so much better. That thrills me right because what are we doing this for? We're not just doing it for a paycheck, we really care about our data. We ourselves, like everyone on the phone, we're passionate about data and we want it to be done better and we want people to enjoy their jobs around the data more. So the levels that this takes you through is a very standard five level process. Now I get a lot of organizations that say we're not even a one. That doesn't exist actually. Everybody starts at a one. If you have a pulse you have one. And those processes are labeled as performed. Then we move our way up the tree and if that staircase I said a tree earlier that's not right. But the key is of course that you once start figuring out how to do it and maybe what your process is is give the data to Peter and Peter will figure out what to do with it. Or hand it to Melanie because Melanie's an expert in these characteristics. That is a good process and it is defined it's not really good because either Peter or Melanie could hit the lottery tomorrow and say take this job and we'll see you later but as Melanie's already tipped you off neither of us are about to do that because this is our passion, our rest for existence if you will. Third level then you get to is are you doing things in a way that others can follow. That's the standardization process. So defined is level three you get three points for it. Level four now we start to measure things. How long did it take you to do your standard defined process? Ah, because if I do it I can then see who's doing it faster or slower. And more importantly why and what results are they producing. And finally the fifth point in the scale is we get up to the optimized scale. Which is we can now look back over the metadata that we've accumulated while we've been learning how our program functions. And in doing so it gives us the ability to say hey we could recommend some improvements. For example a little more documentation around drawing the model other than just draw a picture of your data. We might want to specify standards and levels of inclusion and prescribe logical as well as conceptual and physical data models in the process. This whole process is grounded in and is the same type of thing that stands for ISO 9000 in the sense that if ISO 9000 all they really want you to do is follow your processes correctly. So it's only a level three process. Some of you may be familiar with Larry. English's work that he did early in the last decade is on total quality data management, total data quality management. Again this is the fundamental underpinnings of each of these things here. Let me just insert that Larry English, I worship Larry English's work and I met Larry English and worked with him at NASDAQ for a little while and the data quality process areas in the DMM are just enhancements and distillations of all of the key principles, concepts, methods and approaches that he so well articulated. So he's a real giant in our industry and let me talk about capability and maturity difference on this slide. So capability is we can do this and we are doing it. We have these practices in place, we're doing them well and we've documented them. So that is the measurement of capability and that's levels one through five. Maturity in the CMI world is not the same as capability or doing things well. It's more about the stability of what you've put in place and the resilience of it. So if you're at a bank and we're having 2008, the big bank crash, are these processes still working? Do you have them supported so that they're going to be repeatable, easily accessible? And it's just sensible supporting things like policy, training, quality assurance for the process. So we have mostly, since the model came out, we've mostly been applying capability measurement because capability is what everybody wants. They want to grow and improve their enterprise data management program or they want to get ready to put in a data lake and they know they don't have the practices to support it or they want to kick off governance and they know they don't know what governance should do first. So for all those reasons, they often turn to the DMM. The first organization that will be applying maturity practices because you really need to be about level three before it's worth your while to do that establish the practices, then put in the supports or make sure that they're there and that is Wells Fargo who this fall will do its second assessment against the DMM and they have done enormous and phenomenal work with their data management program especially in the last two years. So we will be applying maturity practices and the maturity practices are either level two or level three. So even if you're a level four or a level five in a particular practice area you will be a maturity two or three. Three is the highest you can get and it just means that everything is instantiated, well supported, embedded, adopted across the organization. So given that kind of a framework in here, Mellie is now going to talk about the structure of the actual document itself and how it works out. So for each there are, we have another slide I guess that tells what the processes are and I'll go through that briefly, but for each process area in there are 25 of them. The items in green here are the only thing that is required for scoring. So the functional practices that we in levels one through five in each area are scored. And they are scored I'll tell you how in a few minutes. And then the infrastructure support practices are the maturity practices that are also scored if you're ready for a maturity assessment. Normally you have to be relatively advanced in the breadth and scope of your program before that would be advised. So capability first then maturity. And then everything else is informative. So we give you the purpose of the process and that purpose is what's the business value in doing this process well. Introductory remarks explaining the content of the process and a lot of tips like outlines for data management strategy, outlines for data quality strategy, tips for instituting data profiling, tips for metadata management. So there's a whole lot of good information. We give you the goals of doing the process as an organization. Questions you can ask yourself to see if you are highly capable related processes that are highly linked to it. And then of course the work products. So there's a lot in the model. It's about 270 pages. It's not as long as the Denbach, but the print is not as small. It's a little easier to read. Go ahead. So absolutely and as Melanie was saying the real key to this is understanding that there are a bunch of different areas of the divide and conquer process. Again if you're not moving your data in the right direction, if everybody's not concentrating on the same thing, then you do end up with an exercise in frustration. So data management strategy is about approaching it with a unified process in here. But there is a class of professionals and Melanie you mentioned you were going to be going to a conference right after this. I know you're in Arizona now. You're probably going to go home before you go back, but you're headed for San Diego, right? Yes. Data governance and information quality conference we're going to have out there in a bit, but we are developing a class of data governance professionals that understand what it is they're doing. But we understand now that data quality is not about making all the data perfect, but we are what data we are trying to get. We are trying to make it fit for purpose in here. Using the correct infrastructure and the correct processes around all of those. And those areas are really critical to understanding what happens. In addition however, you need some supporting processes in order to get you to the next level on this. You can't do this without good levels of management support. So Melanie's now going to talk about the specifics of each of those areas. So the DMM is really focused on data management strategy. Why? Because if you have an enterprise wide or agency wide data management program, you need to know what you're doing and why. So what are you going to do with your data? What is the rationale? What is the business vision you're trying to achieve over time with the data? So if you want to plan forward to manage the data assets to have them be managed as important in themselves, or as we say here, a critical component of infrastructure, you do need an organization wide approach. You do need the approval of all your key stakeholders. And you need to build a strategy and then keep working to implement it. So we have a big emphasis on that. And the line of business participants in data management assessments and training understand this very well. Communication is as important because the data is forever. Governance is forever. Data management is forever. And you need to keep people informed to keep up their engagement and their motivation. So we devote a lot of attention to that. The data management function, or organization, if it is an organization centralized, which we recommend, is all about the people who are the backbones of the data. These are the people who data is in their job title. They are the ones that help maintain the business glossary. They help maintain governance, action items for governance, governance tasks, keep track of that. If there's a metadata repository, the data management function owns that. If there's an enterprise data model, the data management function owns that and probably helps build it out. Persistent products and policies, processes, and standards are the purview of the data management function. So we emphasize that a lot. And then business case and funding model, do your business cases align with your data management strategy? And do you have, among other things, a way to track data management costs? And do you have funding for the data management program? That is non-discretionary. You need to fund it permanently, like HR, or Finance, or Facilities Management. And Melanie's heard me say this several times. The key is to convince your management that you'll no longer need a data management function when your organization no longer needs an HR function. After all, nobody walks around your organization and says, hey, I think we've done enough HR. We've kind of got this under control. People know what they're doing. We should be okay, right? And of course the answer is no. And then you have to ask the same question about your data. And say, what on Earth are we thinking when we think that we can be done with the data, the pieces? I'm going to add one other little point before we move on to the data governance area. And that is that as we've been working with organizations collectively over the past 20 years, one of the things we do find is that organizations are concentrating so hard on writing the first data management strategy. And it's the best that they ever do. And they're trying to write it perfectly the first time. It doesn't work. It's much more important to say, look, let's start off and crawl, walk, and run our way towards the data nirvana that we're headed towards trying to turn the blue elephant into the white elephant, the pure elephant that's there. And part of it is to realize that your first data strategy should be thin. It should be very accessible and understandable by the business. And it shouldn't be too ambitious. Nobody ever writes a data management strategy that's going to take them down the road three years. It doesn't work that way. There's so many things that need to be addressed in organizations that you've got to look at it in a much smaller bite-sized series of chunks. I like that very much. I'd also like to say not only thin, but you have to be able to do the first one quickly. Because for the first version of the data management strategy, you need input from governance, input from the business lines, input from IT, but mostly you need agreement by the key executives. And if it takes you a year to do it, you've lost momentum and you've failed. So you really need to make it no more than a three-month effort with a big trumpet at the end. Here is our strategy. Yeah, celebrate. Absolutely. So governance is the next area, Melanie. Okay. So data governance is critical to everything. I want to say that 25% of the practice statements in the data management maturity model are expressing or implying governance. So 25% of all the model contents surrounds governance activities and decisions. However, the function of your governance structures and activities can be evaluated by themselves. That's a point I didn't make earlier. Everything in the model is meant to be evaluated if necessary completely separately from the rest of data management. We call that orthogonality. So governance as such has its own process area. And the way I like to look at governance is I look at it as these functions, the building function, the nurturing, that means always thinking about the data and how it can be better, sustaining function and the controlling function. A lot of people emphasize more of the controlling power of governance and I think in our case we emphasize more the building, nurturing and sustaining power of governance because the business owns the data. Period. End of story. And they will always own the data. Even though in many organizations, especially those without very developed capabilities, IT has borne the brunt of this. And I've seen this so many times that I know Peter has too. Where because IT is building the systems and the features and buying the COPS products and so on, sometimes the expectation is they should also have total responsibility for the data but nothing could be further from the truth. So that's what governance is about. The next one is business glossary. Business glossary is absolutely critical because it is the heart of data management from the business perspective. This is the agreement about key concepts that everyone in the business needs. For example, if you're a very large organization and you have clients and products, it would be very helpful if you had the word client defined in the same basic way across all of the business lines. That is marketing, sales, shipping, billing, customer service. And in many, many organizations that have external clients as a big part of their business, this is not done very well. So that's just an example. So you need the terminology defined. There was an insurance company that we worked at a couple years ago and they said that their overall reporting and their risk assessment was affected by the fact that some business areas used the word product versus program interchangeably. And that caused tremendous confusion. Having worked at Fannie Mae and Freddie Mac with the DMM in the past, I can tell you that both of them said to me that it took them quite some time to come up with a basic atomic definition for a vastly key concept for them, loan-to-value ratio. And that is critical in the loan business. You have to know what you mean. So they ended up, of course, with many, many flavors of loan-to-value ratio representing the event life cycle of the loan-to-value ratio and what they meant by it. So that's the importance of business glossary. You have to develop a glossary of business terms for shared data, also known as enterprise data. Metadata management is all of the other information that you collect about the data assets. Who's responsible for it? Who's the custodian? Where does it come in? What is the assortative source for it? And on and on and on. Technical operational and process metadata. So that, all together, has a huge business engagement component. And that is our data governance category. So to give a quick summary of the thing, data governance is about managing data with guidance. And the reason that's a wonderful definition is because you can easily turn the question around to management and say, would you want data managed without guidance? And they kind of think about it and go, no, I guess we better not. And then you say, yes, because that's what you've been doing up to this point. And that's where you don't understand terms like metadata management, et cetera, et cetera. Let's move on to the data quality area then. Quality. Okay, so Larry English, if you're familiar with Larry English and his work, then you would get this right away. But I'll just say there's four parts to it. Any process area that you do in your organization, any of these, you can apply them anywhere and have success and improve data quality. They're sort of in a rough sequential order. So we think that logically speaking, the first thing you should do is have a data quality strategy. And this would be something you would do right after the data management strategy because it builds out the focus on one of the three critical parts of data management. So you have governance, you have architecture, that is what you build to house and provision your data. And then there is quality, the condition of the data. And those are kind of the big three pillars of an enterprise data management program. So have a strategy, know what you're doing. You can say, for example, a large bank said, this year we're going to focus on client accounts. Next year we're going to focus on financial products. So you need to have an idea of the scope of where you're going to try to improve quality. And then you need to determine the disciplines and the tool sets you're going to deploy against that scope to make it better. So the strategy comes up with that. And if you look in the DMM you'll see a very good outline for the main content of what the data quality strategy should contain. Profiling discovery of anomalies and defects in the data. The cliche you don't know what you don't know definitely applies here. And the DMM talks a lot about when profiling should be applied and best practices. Data quality assessment is the business driven determination of fitness for purpose as Peter mentioned earlier on another slide. And this involves the business deciding how good is good enough? Do I have thresholds below? I can't think and still do my job. Do I really want 99.99% client identification uniqueness? So it's setting those targets and thresholds and writing quality rules in conjunction with the data management function and IT. So how are we going to keep the data clean? What rules are we going to apply to the scripts when we're ingesting data etc. And data cleansing is about root cause analysis approaching data cleansing in an efficient way and also improving the data through enhancements. So altogether it's a 360 degree approach to quality and organizations have usually achieved a great deal of value here. I've written a couple of articles in TDAN recently and I did a presentation at EDW on how to do a data quality strategy. So if you went to EDW you have those slides. But there are very many discrete steps that you can do not only to do the data quality strategy, but to come up with a complete data profiling set of processes, policies and standards. And those are available as articles. So I encourage you to look at them and at the DMM. Right from there in operations then. Yes. So data requirements, people do a great job on requirements for features. They don't do as good a job on getting the data requirements for either a given data store or for data integration components of the architecture. So we have a lot of best practices for gathering and documenting and maintaining data requirements at all levels, high level and also down into the project level. Data life cycle management is the designation of a sort of data sources or building them if they do not exist or buying them if it's appropriate. And also it is about tracking the data that's important for your organization from acquisition or ingestion or creation all the way through to sunsetting or archiving. And this is particularly important for any highly regulated industry. But there's a lot of very good practices in that process area. Data provider management is about data quality, SLAs, contracts and selection coming up with a more standardized way of selecting the data vendors that you're going to use based on criteria and requirements. Wow. Okay. It all links together, right? Yes it does. And platform in architecture is about how you approach designing the target data architecture. There's DS9 at the top there, so hopefully your data architecture is not as complicated as that space station. But how you go about designing it, what data standards you need for representation, interface, movement, security and those we highly recommend that you establish and then work towards them over time. What platforms you choose to host the data? And we're talking about one or more large systems or data integration platforms that you're going to select. What are the best practices for selecting and managing those? Best practices for integrating data, whether it's just consolidating two systems or developing a big enterprise data warehouse, managing historical data, archiving data and data retention rules. So this hits the business interest areas of the data architecture. And every single business person who has worked with us to evaluate how their organization is doing has been able to follow every practice in here. They find that they do have a stake in what is built to support their data needs and it has been very helpful to the lines of business going through these disciplines. And last but not least, we need the supporting practices in order to actually get all of the resources to do all of the things we just talked about. Yes, so these are borrowed and adapted from the CMMI for development. So these particular process areas have a lot of very good detailed explanatory information and steps and sub steps and so on. So it's basically managing data from metrics and also managing process asset library. So when you develop these policies, processes, standards, how does team A out of your 300 project teams know how to get these artifacts and how to apply them? Process quality assurance, are you doing these processes well? Managing schedule and budget risk and of course managing configurations. And that doesn't just apply to software, it applies to databases, it applies to the business glossary, it applies to the data quality dashboards, et cetera. So given all of that, that's a fairly rich overview of the DMM and what we've described to you so far are the five areas as well as the evaluation criteria that goes into them. And when you take the two of these pieces together, this is really the beauty of what's been put together, we can do something like take a look at the insurance industry. So in the first sample that we had done of these, in that article that I mentioned before, the one that can put you to sleep, we were able to come back and say for the average insurance industry that they were not performing terribly well. Now this is 10-year-old data at the moment. We've updated this since then but this is still where we started out. And the idea is, okay so we can put some effort into these areas and then that will start to improve things. Well in this case you can see platform and architecture underrepresented in this particular process but there's one more key to the model that's real important that you all understand and that is that the model is only as strong as its weakest link. So if I have a model built of concrete bricks and I turn around and make some of those concrete bricks out of marshmallow, I'm not going to end up with a very useful piece. So to dig a bit more specific, this is an airline industry that we were working with at one point and they had ones and ones in those first couple of areas. And then the second lines here that we're putting up are the components against them. These are their industry counterparts. So the executives may not have cared looking at this but they were ones and twos. They're kind of like I don't know about that. But then they look here and say that wow our competitors are ahead of us and in fact when we look against all respondents in the model this is a problem. We need to put effort into those areas that we're only scoring ones on. In fact the only one company that we're going to talk specifically about their results are, Mellie's mentioned a couple of companies that she's working with but she's not talking about their results, was the World Bank, one of the first groups to go through this process. And the reason they had such a wonderful story at the time was because they asked us to do three groups at once. Their treasury function which didn't turn out to be so good and their information systems group that also wasn't terrific but gosh their business was world class practices. And so the idea was that we could look here and tell these guys you don't need to go in and look at everybody else's. You need to walk down the hall and ask your colleagues what they're doing. And one of the wonderful things that happens here is that this process will allow you to look not only what is competition within your industry and competition across all industries but you can also find out what's going on in your organization that you may or may not know about. Now we call this next chart the era of big data because this is when big data was really popular from 07 to 12 and the point was that data increased an enormous value, enormous volume in order to do this but the practice level didn't so we really considered this a missed opportunity a lost generation if you will on this and Melanie's dinged IT a couple times on it and I'm going to ding them too. IT is what has gotten us to where we are. It's not that IT is bad it's that we in the college and university have not taught IT that this is part of what they need to do and consequently they didn't do it. They focus on other things in there. We're addressing that in Virginia here with something called the Governor's Data Interns Program. I haven't got time to talk to you about that but I do get called out to a lot of the other states to talk specifically about this because it's a way of involving students and getting the university sort of back on track. We're going to spend some time now talking about some pieces that Melanie's been doing and I have to tell you Melanie is literally going to finish this piece here and she's going to jump on a plane and go out to Phoenix. By the time I get to Phoenix it'll be midnight. She's a trooper on all of this stuff but she does it because she loves it and most importantly the folks out there in the state of Arizona are doing some very very innovative things that the other states are paying a lot of attention to so Melanie I'll turn it over to you. Yes, not just to start this little section by saying that Jeff Walcove, the enterprise state data architect had a vision for improving data management across the entire state of Arizona. All 100 agencies some years ago and there is an IDIQ that allows him to purchase data management services. Data Blueprint is on it, the CMMI Institute was on it and as a result of this there's been a lot of activity in the state of Arizona but he had an embracing vision of where he wanted to go and he started with governance and he decided of course that since he was in the Department of Administration Strategic Technology area he decided that policy was the way to go. So one of the things he's done over the past few years is issued a governance policy that he had representatives some key data stewards if you will from over 20 agencies work with him on and agencies are now starting to implement that. Let's go to the next slide so that's really good and everybody's setting up governance and they didn't have it before. So they adopted the DMM as the best way to allow agencies to gauge where they were and make rapid progress to build enterprise data management programs because most of the agencies in the state of Arizona did not have centralized data management organizations. A lot of them get funding from various federal agencies for example the Medicare agency, transportation, the Department of Corrections and so on. So they have disparate programs with disparate data requirements and disparate funding sources. So overall in some cases there's not a whole lot of impetus for pulling things together at the enterprise level. However when you talk to these agencies and work with them you find that they have a whole lot of data sharing needs you know like other organizations quality issues and so on. So they adopted the DMM as their standard for measurement and standard for implementation guidelines. And then they're also doing lean process improvement. That's okay they're also doing lean process improvement across the state on the business process area. And that is really, really good. It's TQM in its latest, latest form. So lean analysis is getting them to major processes without regard to organizations, sub processes and key metrics. So let's move to the next slide. So we've done a lot of training there. I'm going out tonight to do a three level, three day training for building enterprise data management capabilities for 24 students. And we did one a year ago for 22 students. And a lot of those students took the work that we did in class and the exercises and immediately started putting it in place in their agencies. So the director of data management, that's her new title for the Department of Water Resources put together a data management strategy, data quality strategy and a business glossary and metadata process for their enterprise data warehouse that they were constructing. And she did a case study at DGIQ last year about that. So let's move to the next slide. So basically now the students who have the training before are very interested in more advanced training. They've done three annual data management conferences in Arizona and there have been five agencies that have conducted assessments against them. So the DMM has helped them and the training based on the DMM has helped them. And there's more. Next slide. So these are the assessments that I've led in the state of Arizona. So water resources, corrections, Medicare, that's the healthcare cost containment system, Department of Economic Services which is a large agency with multiple programs like child support and welfare and so on. And the Department of Transportation, which I just delivered all the deliverables on Sunday night, Mother's Day. Mother of Data, right? There you go. So anyway, these assessments concentrate on gap strengths, organizational themes that we noticed the fixes and then we gave business cases and a suggested sequence plan for all these agencies. So the shared themes among these five agencies so far have been the disparate funding sources, as I've mentioned. There is a big need to share data among agencies as well as with the public and other federal and state agencies. None of them had a centralized data management organization so you wouldn't be surprised that our number one recommendation was established this today. They all were going to work with the policy that Jeff Walcove's group had come up with. They were all in the process of just the beginning of implementing data governance across the agency and now they can turn their attention to the agency-wide data management strategy and many of them have made very good efforts towards a business glossary of key data and metadata. So this is all wonderful progress in the state of Arizona and it's getting a lot of visibility at the governor's office as well, at the governor's level. Emily, I'm going to transition to a couple of sort of overview slides here to let everybody kind of see the results of one of these assessments. These are samples of course. So this basically at the end of the workshops that we do this basically shows you exactly where you scored. This agency obviously, this is a notional organization but they've obviously paid a lot of attention to certain things and not much attention to other things, which is quite typical because data management is broad and it touches so many people in the organization that you can't necessarily do it all at once. If only you could go back in time and have written this 20 years before you did, we'd be a lot further, right? Right. And this one I have, I did not, I confess I did not update this in the last six months, although I've had a lot of information to update it, but this is about six months ago. This is kind of how it looked across approximately 35 organizations who had used the DMM and done a comprehensive assessment. And this is either assessments that I've led or assessments that are trained enterprise data management expert partners have led, but it shows you how up and down capability building is. Notice data life cycle management in the middle there is pretty sad, right? Not a lot of people have done this. And then there are some organizations who have made it their top priority. For example, banks. The banks have needed to satisfy their regulators, so in general they're at the top of that blue line rather than at the bottom. But depending on the business strategy and your needs, we recommend that all organizations try to target level three, because that's the 85-15 rule. You're going to get a tremendous value from that if you work towards that. Okay. So the training courses, I'm going to lead this three-day class tomorrow, so CMMI offers, and also as a partner, I offer as a certified partner, these three-day classes that are very, very intense, and they are based on everything in the DMM, but they're also based on giving you tools, handouts, and techniques that you can immediately apply to get going at your organization. There is also an e-learning bundle that now, if you look at the last bullet on this slide, now offers the enterprise data management associate certification. So if you don't have time to bring an instructor into your organization for multiple people, you can do e-learning individually. It's about eight to ten hours without the cool exercises and the discussion and learning from your peers, of course, but it's really intense education. And then you also get with that little learning bundle a copy of the model for yourself and an exam fee for getting our certification. Then occasionally, like once a year or twice a year, we offer our advanced and enterprise data management expert classes. And depending on the size of the class, that's either two five-day classes, or if it's a small number of people, we compress it to five intense days. And that prepares you to work with the DMM officially with any organization in any industry. Quickly, we've talked about specifically how to look at your organization not just as I'd like to get better with data, but I'd like to get better with data in a way that moves us forward in the same way that everybody else is moving forward. And this is a wonderful model where the last five years we've made so much progress in this compared to where we were when we were all trying to do it with our individual best efforts. And while that was good, this is much, much better. So we've got to stop here at the top of the hour, but we do have time for some questions for Melanie and myself, and I don't think you'll make sure... Peter, can we return to one slide really quickly? Let's go back to the one you skipped, which is about the data stewards course. Yes, okay. So the slide before that, I'll make it very quick. Oh, we can certainly get everybody too, yeah. Yeah, I just need to speak to it very quickly. This course is one of the pinnacles of Jeff Walcov's vision for Arizona. And this is essentially, I need to train thousands of data stewards across the entire state. So we partnered with Arizona and KIK Consulting to come up with a two and a half hour intense course on everything you need to know to be a data steward. Next slide. Five seconds, Shannon. So this is essentially it talks about how to define business terms, how to gather metadata, how to read a data model, how to write business data requirements, how to do work groups, how to improve data quality. And it's essentially everything that we would have wanted other people on the business side to know to do a great job with their data. So this has now been rolled out to hundreds of people, and it's going to continue to be rolled out in Arizona. So this is great. I mean, they're really taking it seriously and I just wanted to shout out Arizona. Absolutely. Let's point out one other thing about Jeff, too, which is that of Jeff's tenure as the lead data person in the state of Arizona, it gives him more experience in this than most CIOs have on their job. So again, we're picking on Jeff here. He's a great guy, but he's really got a wealth of experience because he's been dedicated and working on it for so long. So we're pretty sure that the other states are going to be knocking on Arizona's door and saying, hey, how can we get some of that stuff? Because what they're doing in Arizona is absolutely tremendous. All right, he here. Well, let's dive into that questions coming in. If you have a question, feel free to submit it in the bottom right-hand corner of your screen. And to answer the most commonly asked questions, just a reminder, I will send a follow-up email for this webinar by end of Thursday with links to the slides and links to the recording. So diving in here, what advice do you have for someone charged with implementing data management in a specific data domain at a faster rate than our Enterprise Data Governance Group will be moving in this space? I heard the question from the words charged with implementing data management at a faster rate than the governance group, but I didn't hear the first few words. Sure. And to both of you, what advice do you have for someone charged with implementing data management in a specific data domain at a faster rate than our Enterprise Data Governance Group will be moving in this space? Yes. So I wanted to say that the DMM can apply to anything. We have applied the DMM in one case. One of my students applied this to one large database at the US Geological Survey called Science Base. And they were wanting to institute data management practices because they're giving data to everybody about geological features. And so that was about the smallest scope that we'd ever applied it to. But you can definitely use the DMM for one domain and increase the practices and your capabilities quickly in one domain. An example of that is a well known insurance company and the trading and markets area of that insurance company used the DMM and made wonderful, wonderful progress and got a whole lot of things implemented quickly. So yes, you could apply it to a business domain. Let me expand on Melanie's point for just a touch. What people don't get from this model is that it looks again like we'd like to do this for all areas of the organization to make the entire organization more productive. But Melanie's mentioned 8515, 8020 regardless of how you characterize it. If you do better with your important data around these issues, then that's also a good way of scoping the application of the DMM technique. Yes, your organization will benefit a lot if you do this for everywhere, but not everybody's got the time. So we know we change the dimensions of time and schedule and functionality that other things have to give in response to that. So if we want to go faster in some area, we have to do that by changing the scope. And maybe the goal is to move, as Melanie said, more things up to become a three in some key areas and let things languish at a two or one in some other areas that may not make as much of a difference immediately in order to do this. In fact, all of this should be governed with a good ROI model in all contexts here so that you're not just really trying to become perfect for the sake of becoming perfect, but you're becoming perfect because it's a valid business case. Yes, and at EDW, someone asked me in my presentation why do you recommend level three? Shouldn't everybody try to be level five? And our answer has always been if your business need dictates that you become like a superstar in a particular process area, then yes. But if your business strategy doesn't require that, then no. And I'll give you an example. The Federal Reserve Statistics function had to be a five in all of the data quality processes because the stability of the US and the financial system depends on the data that they provide to all the different functions that are gauging interest rates and so on. So if it's really life or death or it's the space shuttle type of priority for you, then yes, you need to be a five. And you can find that out based on your business strategy and your data management strategy would reflect that. I was just going to offer the comment. Don't you feel good that the Federal Reserve is in fact following this as part of the guidance to make sure that things stay on track from a financial perspective? And so yes, absolutely. So what is the good resource for gathering data requirements? What are your favorite requirements books, Melanie? Yes, so resources basically, if you're redoing a system that already exists, let's say that you have an old legacy system you're replacing, then of course you would want to mine the DDL and see what kind of data is being used there. You would want to decide whether it's still being used. You would want to check with the key business stakeholders to determine if the data in the way that it's structured and so on is adequate. And then you would want to redocument that in a way that you could easily update it over time. So how you get data requirements is going to depend. If you're building something new, I'm a huge proponent of logical data models that are reviewed and approved by all of the producers and consumers of that data, the stakeholders, as the best way to make sure that you're getting the data requirements well specified. Even if you're not going to implement in a relational database, the relational data model is the best data organization and thinking technique to make sure that all the data requirements are captured. And then of course there are documented requirements where you can include let's say technical requirements for the data like availability, security requirements for the data. You can include that by expanding the structure of your functional requirements document. So those are just a couple of tips. I'll add a couple more. I have a long explanation and it would differ slightly depending on the nature of the endeavor. Absolutely. And again this is an area that we just do a very poor job on in the university community because we teach people that use cases can cover all of this. The major issue with this is that most people only approach it from a two level model perspective. So data is a combination of facts with meaning and information is then the data that is provided in response to requests. But of course we know that the most important request for data is just give me all of it. I'll take care from here. And we know that doesn't work. It also throws into question a lot of the research that we've done in the academic community over years because we've done it based on what people say. It turns out that actually watching how they use it is much more important than listening to what they tell you about it. And if you're able to get in and learn some of these more advanced techniques going into this, you can find out that going and living with your users is a very, very valuable exercise that will complement the things that Melanie has already described to you. Again our users are generally very good about telling you what they think they do or what they think we want to hear. But when you watch what they do it's considerably different and that can affect your requirements in a major, major way. Well and also because in my case much of my professional background was enterprise data architect. So I'm always thinking about as it were making the world safe for democracy. In other words protecting your data layer from future endless complexification. So I'm always thinking about streamlining and simplifying and good simple clean organization of the data. So in terms of data requirements if you have a business area data model or an enterprise data model or anything of the sort or even if you don't have it documented if you have an enterprise data warehouse that people are happy with you would definitely want to start from that centralized approved vision of the data. Again shared data means at least two people have agreed on what the meaning of the data is and how it should be used in there. It's a non-trivial problem really good question. So the DMM applies to a more mature organization, a startup or all? Yeah and especially one of the more interesting questions that the Mellonette left the answer is that when do you get to build one of these? And the answer is only when you're a startup. So if you're building a data architecture from the very beginning the only time you're going to have that is when you have nothing else that exists. In 99% of the cases you're going to be reworking an existing data model transforming from one legacy system to another. And this is where the maturity process is critically important because it also speaks to things like data integration. There are good techniques for data integration and there are bad techniques and we've seen lots of organizations get burnt by paying people to do the process badly in a way that benefits the consultants and not necessarily benefits the organization. So that's the importance of governance too. My latest article on TDAN, it's based because it solicits my columns and I love doing them. My latest one is the importance of business engagement in enterprise data architecture. And what I do is go through all of the disciplines and thinking and perspectives and consideration avenues that the business needs to apply because since the business owns the data you can't just let IT do it because IT tends to be influenced by shiny objects and you don't want that. You want to think about the what, the substance of the data that will be managed by the technologies and platforms. So you really need to think about the organization along with IT to have a successful architecture plan and implementation. IT by itself is insufficient. And then if you make them do it they get blamed so it's bad for them too. Everyone needs business engagement both the business and the IT side of the house. It's data protection classification retention and disclosure should be part of data management responsibilities and if yes is this going to change the focus from data quality to data security. Is data protection part of the DMM? Data protection as such is not but we have mapped the DMM to the general data protection regulation and it turns out that applying DMM processes already without any additions already matches 85% of what an organization would need to do to meet the GDPR, the letter in spirit of the GDPR regulation. What we would like to do is enhance the DMM with a data security category which would have in it certain key process areas and only the ones by the way in which there is a significant business engagement. The business doesn't really care unless they're in the intelligence community and they're analysts themselves they don't really care what encryption algorithm you're going to use. They'll say give me encryption. So we're not getting into the technical side of it at all. It's going to be things like identity and entitlement specific considerations for classification of PII data etc. So we have planned to expand the DMM but it already applies in great part. Absolutely. This gets us into a broader category of data leadership which is to say that the Millings already mentioned the data governance organization is one of the ways in which you enhance data leadership but I want you to just try to imagine that we're taking a data leader we'll call them a chief data officer because that's the current term of art for it. I've got other categories I like to call them enterprise data executives because it's a little bit safer category in this case to label them as. Chief data officer is being told both first of all make sure that everybody who gets the data gets it but don't give it to anybody who shouldn't and while that's an interesting thing it's not really a compatible mission. Our data leadership should be about using data more effectively as a tool and that security is a separate function. We don't ask our chief medical officers in our hospital our hospital systems to also do surgery. They're responsible for medical practices around the whole process and they have a risk officer that addresses some of the other issues. So there's very definitely a case to be made to say that yes while the permissions are important and the defining role based security is absolutely part of the remit that goes into this DMM process that we also do not want to hand these individuals in this function the CISO type activities as well. It's just a no win situation. How does an individual get started? Do you recommend just buying the book going to a class or do you have to hire a company? Yes buying the book is very very good and coming to one of the trading classes or doing the online bundle for the Enterprise Data Management Associate Certification which includes the model and the exam fee is a very good way to start for an individual. For an organization there are basically two paths to get started. One is sometimes organizations decide they would like to train a cadre of their key stakeholders first and that's a fine way to get started. So some organizations I've worked with want to do training first so they train. Others are eager to or they have a real pressing need to get a baseline evaluation of their data management capabilities so they do an assessment first and that is also a very good way to go because it's kind of like jumping off the deep end but the companies that have a need let's say they're going to buy a huge new data lake and they know they don't have processes and they really need to know where they are so they know how to support this massively expensive major commitment to this technology. They need to do an assessment. Organizations who don't have a clue what to do with governance once they stand it up they would be well advised to have an assessment. Organizations who are doing things like some advanced organizations who've been around for a long time and are technologically very capable are replacing their entire enterprise data warehouse layer. They should have an assessment so it depends. So the organization is going to have some driver or other why they need this or they have the regulator beating down their door. They should just do the assessment and then train later if they want to bring along individuals and internal experts and who can help gauge progress effectively over time and that's what a number of our clients have done. One of the wonderful things of course about working through data diversity is that there's a number of different platforms that you can go through and learn more about these various tools and techniques and so we've already mentioned the conference upcoming. We're also doing a specific data governance conference in Washington DC in December that will have a number of these topics addressed as well. I'm going to give you a quick piece of advice here from Gardner though one of the things Melanie said was you know replacing a data warehouse with a data lake and we have seen a lot of organizations that are attempting to do this. Gardner has now put out advice as of January this year that says surprise, surprise if you have both a data warehouse and a data lake you can meet more business requirements. I know that's got to be motherhood and apple pie to a lot of us but it is kind of interesting. So don't look at any of these as replacing things in your organization. It goes back to the governance part. Part of the process of maturation is to realize that more capabilities means your organization can do more things and if you do them using the DMM framework you will do the more things better. That's a terrible summary on that but I think it actually does work when you go back and parse the words there. Do we have a question Shannon? Absolutely so does my need to follow a top down EDM approach or is it fair to start in bottom up approach and have a retrofit to come out with the enterprise model? Are we talking about data models? Okay, so if you're going to start bottom up because you don't have the corporate will or budget or approval to work on an enterprise data model which by the way I'm going to have to say this it's become very unpopular to do enterprise data models because they take a while, they take a lot of input from project teams and governance participants and so on. However, every organization that I have worked for where I've either led the development of an enterprise data model or they have an enterprise data model, those organizations have always achieved efficiency gains in designing new systems, transitioning systems mapping custom to costs and so on. So I am in favor of it. We don't make it a requirement in the DMM but I personally would recommend that all organizations over time work towards developing an enterprise data model. Now if you're going to do it on a smaller basis then of course what's the best place to start? Easy peasy, master data management because every organization has core entity types where they need to have the latest, greatest, best and most high quality information at all times like we've mentioned client. If you're in an organization that deals with clients and you don't have a single hub that allows you to identify a client and have key descriptive information about that client then probably you're capturing clients in multiple operational systems and that's really costing you and I could go on and on with anecdotes but I won't because of our time constraints. So that would be the best place to start because the set of master data is typically small. There are 25, 30 data elements for client and product another popular one is typically no more than about 100 data elements and you can, to use my metaphor, make the world safe for democracy in a small scope much more easily. That then would be one of the top level prongs of an enterprise data model and it will get you a lot of benefits. So that would be my recommendation. Peter, anything you wanted to add to that? We covered it well. And I love this one. So if you're a vendor who needs to incorporate with client data what are your best political tips for how to influence a large client organization to set up governance so that the data is higher quality? I love the question that I asked sort of midway through which is that many organizations will come to us and say how do I convince management to put some investment into this and the answer is quite simple. You've got to ask them a simple question. Do you want your data managed without guidance? And usually the answer is no. And they say okay good, so you want your data managed with guidance then how can we start that process and what are the things that are going to be effective and that leads to another whole series of questions which gets back into the things that Melanie was talking about earlier. If somebody is willing to stand up as an enterprise leader and say you know I do not care how data is run in my organization because it's immaterial to my business I'll sell that company short on the stock market. I mean it's just real obvious that they're not going to be they're going to run into some problems and neither Melanie and I believe in you know fortune telling types of things but I think both of us would lay even money bets on that one real easily that that company is not going to succeed in the long run. Absolutely. So what is the process to govern these slices of the circle? So the governance activities in the DMM are built in to each process area where they occur. For example in the business glossary there are practices essentially that reference having a defined process for how business terms will be defined and modified and you cannot do that without governance because the place that's important to start in the business glossary is with shared data and even more specifically highly shared data to use our example of client for example. So governance is inherent in all of these process areas. In terms of a governance roadmap that is going to vary with the organization. So let me just use an example to Arizona state agencies. One of them was not able to answer the question for the governor of who is the served population, citizens and residents of Arizona and they had nine different programs and many many systems that captured the client data natively so as we say in the data architecture world multiple creates which is the big tar pit of data architecture. So they had lots of different and they had tried to resolve this in a number of ways. So they were going to take that client data and they were going to first set up a data management organization and among the people who are going to be starting in a data management organization would be a governance lead and the first task they would be leading would be an overall high level requirements for and then development of business glossary terms for what is a client across all these nine programs in the agency. For another agency they had a lot of big programs. This is actually the department of transportation but they had a number of priorities and many of the programs were actually in pretty good shape so they had as a strategic objective this year to stand up data governance and that came right from the director as well as from the CIO but it didn't bubble up right off the top of the head either from them from us or in collaboration together what would be the best initial project for them. So we decided that their first initial task should be gathering key stakeholders to look at all the key subject areas and programs and determine what the enterprise priorities were. So mostly it's mostly most organizations it's easier to see where like if you're putting in a data lake like Wells Fargo and they're putting a lot of effort into maximizing the quality and definition of the data for everything that goes into the data lake then that is sort of an anchor point for the other agency I mentioned client was an anchor point for department transportation they have a lot of key things as I said they're doing very well in general and they need to figure out how the best way is through what avenue to pull these things together through their data management organization. I hope that helps. What to do with governance is the question and I gave you a couple of examples of the answer. Peter anything you want to add to that one? Once again I think Melinda and I dancing just perfectly. Awesome I love it. Well just there's several questions out there about classes availability and books and we'll get all those resources to everybody in the follow-up email which will go out by end of day Thursday with links to the slides links to the recording of this session and the additional things here we're tested within. Peter thank you so much as always Meli so great to have you join us thank you so much and thanks to all of our attendees who are always so great and with all the great activities you guys come in and so how engaged you are I just love it. We hope to see you next month for our presentation with Peter and hope you all have a great week. Data Governance next month right? A hot one. Meli thanks so much. We'll catch up soon and Shannon thanks as always. Thanks y'all. Thank you very much. Have a lovely day.