 Hello and welcome, my name is Shannon Kemp and I'm the Chief Digital Manager of Data Diversity. We would like to thank you for joining the latest installment of the Monthly Data Diversity Webinar Series Advanced Analytics with William McKnight. Today William will be discussing increasing artificial intelligence success with master data management. Since 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. 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 of questions via Twitter using hashtag A-D-V-Analytics. And 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 right-hand corner for that feature. And as always, we will send a follow-up email within two business days containing links to the slides, the recording of the session, and additional information requested throughout the webinar. Now, let me introduce to you our speaker for the series, William McKnight. William has advised many of the world's best known organizations to strategy and use the information management plan for leading companies in numerous industries. He is a prolific author and popular keynote speaker and trainer. He has performed dozens of benchmarks on leading database, data lake, streaming, and data integration products. William is the number one global influencer in data warehousing and master data management. And he leads the McKnight Consulting Group, which has twice placed on the Incorporated 5,000 list. And with that, I will give the floor to William to get today's webinar started. Hello and welcome. Hello and thank you, Shannon. Good morning, good afternoon, everyone. Want to welcome you to this webinar. And William McKnight, as I was just introduced, and on this series, which is a monthly series, I utilize my experience to present in-depth topics with strategies and organizations at the top of their game or deploying. And I've got a couple for you here today. Artificial Intelligence and Master Data Management, a couple of my favorite topics. And I'm bringing them together today for us because they belong together. And I've had some experiences that show me or prove to me that Master Data Management is a great fundamental underpinning for artificial intelligence success. And I want to share with you some of that information today. And I hope it improves your artificial intelligence success because that's going to be very important to you in the next decade. Artificial intelligence is really going to set apart all industries. And those industries that can conform to it, that can actually get on the train early and often and do some very progressive things with artificial intelligence are going to come out the winners in this new economy. But Master Data Management is part of that. And so I wanted to get into that. First of all, I was recognized a couple of weeks ago, maybe, in this nice list. Since we're talking about MDM today, I'll mention it. It's great to find myself in this list. And if I can help you with your MDM needs beyond this, let me know. Our planet is becoming a very different place. Yes, I said planet. I didn't start with our company or anything like that. Companies influence the planet. And our planet is where we live. It's becoming a very different place with artificial intelligence. Artificial intelligence will be everywhere. AI-driven stores will allow you, which already exists, by the way, will allow you to shop without cashiers or waiting lines. AI will diagnose and treat patients. Will manage transportation and analyze massive quantities of data in real time to do it all. AI chatbots and voice recognition systems will sound like people increasingly. AI will be implemented in robots, robot dogs and other pets that will serve as companion robots for the infirmed, for people who need that and who want that. And so we are seeing all kinds of AI implementations out there in the market. And anytime I talk about AI, I like to share with you some of the interesting things that I come across and think about. And here we have this gentleman's craze with neural networks, art that generates realistic faces of subjects from iconic paintings. So the Mona Lisa is actually of a person, and this is more or less what that person would look like based upon reverse engineering the painting with a neural network. And I've seen so many of these that he's done and others have done. It really helps bring history to life, sort of a hobby of mine. So I like to do that. In the upper right, you see a painting. It looks like any of a number of paintings, right? Well, this painting actually sold for $432,500, so I guess paintings do sell for that. But the thing about this painting is it was created using artificial intelligence. That's right. Artificial intelligence was just fed a bunch of other paintings of people of the time and said to do a great one, asked to do a great one. And apparently it did, at least according to the $432,000. And artificial intelligence is making its way into so many other things, like, for example, the whiskey process takes years, and there's so many variables that go into it, you better get it right. And so by reverse engineering all the recipes out there, we are seeing that artificial intelligence recipes manifest in that market. And let me play you a few bars of this song. And I jump ahead a little bit. We don't have lyrics, but nonetheless, a very calming Susan Song, I suppose, created by artificial intelligence. And it's creating poems and poetry. And it's doing a lot of writing to a lot of the journalism that we see out there is actually from artificial intelligence today. So I just want to make you aware of kind of where things are today, case you're not deep fakes, of course. And let me see, jump ahead. I don't know who that person is, but OK, that's Barack Obama. And yeah, but it's not really Barack Obama. It's a deep fake of him. So if you're not familiar with deep fakes and how realistic they are, probably that's something that you should be just so that you're not caught unaware, because I think that that can be used obviously for good and bad. I'll stop there. And then on the upper right, we have Sophia. She is a robot, of course, or I wouldn't be showing her, but she is a citizen of Saudi Arabia. And she's famous for saying, I have feelings too. And she kind of does have feelings. Well, she has bits that get flipped based upon the sense I that she is experiencing at the moment. And so she is getting her feelings on. And then there's so many ways to uniquely identify people that goes into these deep fakes and these robots. And this issue of the MIT Technology Review on the cover here, you can see all the various data points that go into our unique faces, which they are unique, believe it or not. And there's so many other examples. Another one I'd like to share about artificial intelligence is there was a movie that de-aged some of the actors, like Robert De Niro, the movie called The Irishman. It de-aged them by 40 years so that they could be the actors in this movie, but it took years and years to do that. And it only took, well, this was years ago. Last year, it took a YouTube beatfaker who did a better job with it in just seven days of work. So that's how rapidly we're able to exploit artificial intelligence today. Robots can read better than humans, putting millions of jobs at risk. You'll see this manifest in some of the examples that I share as we go along here. Yeah, so if you thought that was sort of the last bastion of humanity, let's keep thinking. There are, they do exist, of course, but reading clearly is not one of them. We get tired, we certainly don't process information at the same rate and so on. And that's all getting better. So in legal research, medical research, and industries like that, robots are taking over, sensors are taking over, factories, AI will dominate factories soon and begin to serve as teachers, cooks, pharmacists, law enforcement officers, athletes, and other professionals. Universal translators will remove all language barriers. Hundreds of sensors will be installed in our clothes, homes, and overall environment to monitor us and share information back and forth with us. Medicine, of course, is an area that is undergoing extreme transformation with artificial intelligence. Computing is changing itself. And we may be running on quantum computers based upon the properties of quantum physics. People are controlling computers and prosthetics with no physical interaction. And we may soon be controlling things with our minds and communicating from brain signals. So the nature of reality is changing. And by the way, if you thought, well, I'll at least be designing that AI. Google is now teaching AI to play the game of chip design, which obviously pretty involved. So pretty forward there. I'm a beta tester of this GPT-3, and I am impressed. Let me tell you, just about whatever you want to do with text, it can do, it can abbreviate, it can extend from its library, it can help you out with song lyrics and all sorts of things filling in the blank. It's pretty smart about that. It takes context into consideration. And it has all these 175, GPT-3 has 175 billion machine learning parameters. So pretty in-depth. This is an Elon Musk property or invention. So that's, I think, the basis for a lot of the AI innovation that's happening today. Now, let's get to our enterprises. Oh, maybe I already did. Some of you are in those fields that I just touched on there, health care, distribution, pharma, and so on, retail. But here's a few more examples of AI in the enterprise today. I want to impress upon you that it's here now and it's here to stay. So maybe you find yourself in one of these examples. I won't read them all, but I'll mention financial fraud detection and reducing costly false positives. That is really getting dialed in. And it's not because a human anymore sitting there looking at information that's a day old. It's real time. And it's incorporating vast amounts of information, including when it's available, information from master data management. Because master data management is highly summarized information. That is when we do it in an analytical fashion. It's highly summarized. And that's going to speed up. That's going to speed up AI right there at the point of sale, at the point of the person sitting there on the website at the real point where this is critical. And automation is also in my list here. My clients are very innovative in terms of what they can automate with AI. I don't have to think about it. I just mentioned the word automation and we start to go. It is really something else. So here's some more, marketing, cybersecurity, smart cities. Oh, that's a huge use of AI. And think of all the data that it needs. And that's why I want you to start thinking about here is we segue to master data management for AI, which is all the data that these apps need in industry. Oil and gas, the data needed to determine drilling patterns, ensuring maximum utilization of the assets, manage operational expenses, insurance safety, predictive maintenance. Any of us that have equipment have those things more or less. And artificial intelligence, acting upon great data can be a big boon to that. And a lot of people will ask me and others, well, how do I get started with AI? Well, we'll start talking about your data. You have the data in order to do artificial intelligence well and a foundation element of that I'm going to suggest here as we go along is master data management. We're to look for AI opportunities in the products that you make, in your supply chain, in your business operations, how you design your product and your service set, your design. And your marketing, your approval funnel. These are all places where companies out there today are exploiting artificial intelligence to do better at these things. AI affects the entire organization. It should be affecting organizations from the strategic level, from the very top. From the very top, we should be thinking about artificial intelligence today. We should be thinking about how it can impact all the things I just mentioned. And making that happen within the organization. Of course, technically, we need a framework. We need a stack or two. We need the tools and technologies in place. We need the operational processes in place. Talent is hard to find, and it's very important. And we need data, and that's our focus here today. We definitely need data. Any of these really is a knockout factor for overall artificial intelligence success. Yet I would suggest that data is probably the most important one of all. Now, some of you who have been joining this series on a monthly basis, you may recall, I believe it was September, maybe August, that I gave the presentation on data maturity. And I'm not going to review that whole presentation here today, of course, but my level three, which is right in the middle, which is the goal for everybody right now at the least. And that slide looks like this, and I'll just highlight a few things. My data strategy for maturity level three, which is, again, where organizations need to be today, all in on AI. And I defined that with a few hundred words, but I think you get the gist of that. All in on AI, not questioning it. Maybe it's okay to be looking for the opportunities today, but you have to at least be doing that. And be all in, got your stack ready, and so on. And also, by the way, since I'm here, I'm already in here on this slide, saying these few words about maturity level three, MVM is in there. I've got all functions for all major subject area. That's your goal. All functions for all major subject areas. And I'll help you with that a little bit here today by defining more of what I mean. Starting with this slide, the data to manage. The data to manage for AI comes from all over the place. It comes from our e-commerce systems, our ERP and CRM systems, our IoT systems, of course. That's a huge point of entry for AI data. Publicly available data, reaching out beyond. Our four walls, looking at that data. Third party data, syndicated data, if you will. So all of this data needs to come together and be ready for artificial intelligence. And some of that, we're doing a pretty good job of today, I would say, like ERP and CRM data in its respective databases and also in, of course, the data warehouse. So there's a lot of data to manage and here are some more detailed examples. Call center recordings and chat logs. Customer account data and purchase history, user website behavior. Some of this data that you see on here may be data that your company is currently treating as sort of temporal. It's here today, gone tomorrow, we process it, we move on. Well, I think the future dictates that we still not only just keep that information around just in case, of course, we do that, but we exploit it on an ongoing basis. And I am suggesting here today that the outlet for that data is master data management. In other words, when you come up with the analytics for your customer, where does it belong? It belongs in MDM. Operationally it belongs in MDM, analytically it belongs in your warehouse and perhaps other places as well, maybe your data lake. Maybe the lake is fed from MDM as well. I'm not gonna get too deep into architecture. I suggest it is actually in most of my architectures, but anyway, all of this data in their respective places, but a lot of this is summary data that we want AI to act with in real time. And we want AI to act with, I'm going to suggest master data management, not just old school master data management but core data like what you may see here on my left, but also in terms of what I call empowering attributes and some call these the analytical attributes. This is obviously of a customer. And you can see that the ones on the right, like the propensity to churn, that's not exactly something that the customer would enter in a form for you. That's not exactly something that you can pick up from one transaction necessarily either. That's a pattern of transactions over the course of time that go into a propensity to churn. And so that sort of transacting, I'm suggesting is great to do in master data management and store and make available from master data management. All of these analytical attributes. So if you're just doing core basic MDM, you've solved a problem, great. Pat, on the back, you're distributing data across the enterprise, you're distributing one copy of the data. That's a big problem for so many companies. So congrats on that, but there's more to come. There's more to do with MDM. There's more value to be had if you get into these empowering attributes. So we want our data to be ready. I say this repeatedly, get your data ready, get your data act together. So what do I mean by that? You might say, well, we got it in a data warehouse, okay? But are you hitting all five of these bullets with that? All right, well, we got it in MDM. But what about these bullets? Is it in a leverageable platform? Now I just said data warehouse and MDM. Those are the two big, maybe you throw a data lake in their leverageable platforms. This is where you need to focus your energy in leverageable platforms, not one-off, not application-specific platforms, but in leverageable platforms. That's what I'm talking about here today. In an appropriate platform for its profile and usage. So is it, are you shoving graph data, massive amounts of graph data into a relational database, for example? Are you putting a lot of unstructured data into your data warehouse? Are you filling that up with petabytes of unstructured data? Are you putting that in a data lake? Or are you trying to do your financial reporting off of Hadoop, you know, or some such thing? Something that's inappropriate. And I've talked extensively about what's appropriate and not. But for however you define it, make sure it's in an appropriate platform. And if not, I suggest that there's still value to be had if you were to do some re-engineering to appropriate platforms. With high non-functional, availability, performance, scalability, stability, durability, and it's secure. Do you have those high functions in place? Now we're getting close to data that is pretty ready. Capture it at the most granular level, not at summary level, at the most granular level. And does it meet a data quality standard? Not as defined by you and I, the data and analytic professional, unless you're in data governance out there, but as defined by data governance, which does imply you have data governance and you have that business input to data. So data is ready when you're meeting those five bullets. And master data management is a great platform for helping you meet these bullets. Projects are a series of subject area mastery. This is very important. When I teach MDM, I want you in that subject area mindset. So maybe we're talking about customer, maybe we're talking about product site, whatever the case may be. You have to be an enterprise that is oriented to what they are and what that is and how they relate and when each one of them is going to get mastered and how it's going to be mastered and so on. And if not whole cloth, we're done, okay, we're done with customer, it's ongoing of course, but there's that huge hump of work that is what I'm talking about. And you might have to break that up, but when do we get to the end of that? And when is it in master data management available, performing, scalable, et cetera, et cetera? So this is what I mean by data that is ready. Now I'm going to suggest that robust MDM, one that meets all the criteria that I just laid out by the way, is half of the effort for AI success. Yeah, AI success. So I'm talking in terms of, you might call it a budget or a project or an app. Okay, they all kind of mean the same thing in context of this presentation. So on the left you see different, I'm going to just say apps in the organization, fraud detection, a call center, chat bot, et cetera. All these are heavy hitters in terms of artificial intelligence. It would be hard to deny that any of these you would undertake today without heavy artificial intelligence within them. All right, and on the right, I have some enterprise data domain. Not all, of course, not all for any company because they go on and on and everyone's unique. And you'll look at this and go, well, that doesn't apply to me, et cetera, that's fine. These are just an example, data domains or subject areas. And these are what I want you to be oriented to within your enterprise. This will help you a lot achieving MDM success. So let's go on and talk about MDM here. It's not an option for AI apps. You might be thinking, well, I can do it. I'm already doing AI without MDM. So thank you very much. But you'll do, you're doing master data management, but you may not be doing it with a discrete focus on it. You may not be doing it with a tool. You may not be doing it with data professionals. You may not be doing it in a, using all the things we've learned about master data management. You may have cobbled together some master data. Okay, so at some loose level, that's master data management. That's what I'm talking about here. So you either will have an application focus to your MDM or an enterprise focus. Now I like the enterprise focus, but I frequently do an application focus and walk my way to the enterprise focus. And you may have to do that as well depends on where you are in your maturity. With an application focus, your focus, it's on an application. Massivity first, usually it's a work effort to get to the second, third, fourth, et cetera application. Be sure you build what you build so that it can scale so that it can get there without starting over again. Or without massive effort to go from product to let's say a site or product to customer. I'll use the big two, right? So you could also take an enterprise focus which is a focus on the subject area first. Yes, of course you will have. You will have application customers for this immediately, but you don't build it just to serve their needs. And I suggest, you may not believe me, but I suggest that if you take an enterprise focus, you'll be so much better off when you have that first application up and running. You'll be so much further down the path if you're having the second, the third, the fourth, et cetera up and running on that MBM hub as opposed to if you just take a strict application focus. But hey, I'm not in your shoes, you do you, but you wanna get to an enterprise focus. Take a higher chance. You have a higher chance of creating new organizational possibilities too if you take an enterprise focus. If you take all the needs of all the apps into consideration when you build out that customer, all the needs of all the apps into consideration when you build out that product, et cetera. Of course, the danger with an enterprise focus is you get all on your ivory tower there on that hill and you build it and hope they will come. Can't do that either. So I've left a lot of latitude in there, haven't I, for your judgment, from making sure that you make it a success on both counts. On one count is the business count. You have internal app customers. You must. The second is that you're doing the right thing by the enterprise long-term in terms of scalability that you can move beyond this without it necessitating an entire redo because we do not do it scalably. Either initial focus needs a secondary focus on the other. It's the MBM leadership challenge. That's right. I'm leaving that out there for you, the leadership challenge. Either way, you'll do MBM, but without a discrete focus on it, you may not do it well. So I suggest you do it well and you do it with data specialists. You do data modeling, you do data integration according to the science of data integration, not just winging it. You have data quality inherent in there. You use a tool. You're not just hand-coding your MBM. The tools have stronger value propositions now than doing that and they have for some time. It's just that hump of, it's just that hump that you want to get over of incorporating yet another tool into the enterprise. I know who needs another tool in the enterprise. Well, we are going through a time period here in the next, I'd say probably five years when it's kind of like the accordion has been opening up and it will continue to open up. It's going to be more complicated before it gets less complicated. Before we get to that truth, one side fits all. One architecture, one dare I say tool that works for the enterprise. This is still going to be a challenge. It's operational in real time. Let the hub create the analytical and the empowering elements. Let the hub do that processing. Let the hub incorporate your transactions not for storing and distributing. That's for your data warehouse, your data lake, et cetera. Let the hub absorb that information and glean the intelligence out of it and save it and make that data available. Just like that propensity to chair and I was talking about a little bit ago. Make it a discrete project with high touch points with an application. There's the conundrum. It has to be a discrete project done with these data specialists and so on, but there should be high touch points with all the application owners. And for this, you need to organize. You don't just want to up and do MBM. You need to organize and you need to bring business applications into it. Those are your customers if you're building MBM. Focus on the total cost of ownership first for justification. You want to justify MBM? I've tried it different ways. Let me save you some time. It's a TCO play. It's going to save you time in the long haul because you're going to do it anyway for all of these apps, for all these AI apps and other apps. You're going to do it, but can you do it once and build it to scale and have that be the one and done or are you going to redo the, how shall I say the analysis? The ocean. We build the ocean, we boil the ocean every single time. So build it to scale. It doesn't take longer to consider all known requirements. It really doesn't. We're making some efforts early and often like they voted in Chicago early and often. All right. The real decision point, you're road mapping around sponsorship. Road mapping means planning over the course of time. When are these things going to happen over the course of the next year or two or three as the case may be. My crystal ball gets a little fuzzy there, but two or three, I think it's totally appropriate. Sub by subject area, publishers or workflow, don't get third party data. Who are my subscribers? So publishers or workflow, that's where the data is coming from for MDM. See, an MDM data comes in and data goes out. That's it. Just like any other database for that matter, except it's a specialist database. The data may come into a workflow that you build or it may come into a publisher. And then the data goes out. This is where it gets exciting to subscribers. It doesn't just sit there in the hub. You haven't really done MDM if your data is just sitting there in the hub. Come and get it. No, you need direct links to your customers, your AI applications. Don't forget common artifacts like the data warehouse of Dave Blake and various operational hubs that you may have. And what I mean by that is, don't forget to push the data to them. They need it too. That you should not be creating a different customer master in MDM than you have in the data warehouse. It should be one in the same. So if you created the data warehouse first, like most companies have, and you're along your merry way, and then, oh, you're doing MDM now, and you build a customer in there, that customer should feed the data warehouse, customer dimension. All right, how are you gonna communicate? That needs to be on your roadmap and bringing in sponsors. And governance, yeah. You see, MDM is data as a service. And a big question, and I'm getting in front of it here for you, a big question is how far does the build team go? So maybe you're thinking I'm on the build team, maybe you're thinking I'm not a builder, I'm a user, great, whatever. I'm an app owner, then you're a user of MDM, but how far does the build team go? This is so often left to, oh, whatever happens. We'll see when we get there. No, think about it now. Build a service level agreement for the master data. Everybody wants to know, what's in it for me? So if you have an application and you hear, oh, what are they doing over there? They're building master data. What are you thinking? You're thinking, what's in it for me? How do I get involved? How do I get that data in my app? Well, if you're on the build team, you need to be able to tell them that very succinctly. How does that work? And of course, the devil's in the details and the pudding and all that, it'll take some time to get it all detailed and so on, but you gotta have a way, all right? MDM communications and center of excellence. Okay, the build team might do that. Integration planning, this is the big one. This is the big one. I put a question mark here. Actually, I put a question mark on a few because there's no right answer. The right answer is the answer you all agreed upon. So is the build team, is it built to push the data into the AI apps or is the build team built to build it and they will come? I don't know, everyone's a little bit different on that, but be sure you know. So you don't get to the point where, great, we've done it. And then both teams are looking at each other going, okay, you're responsible for bringing the data into our app. I've seen that happen a few times. And so I just want you to be in front of that. Hub model and rule expansion, how does that work? Build team will probably do that. Mapping the elements, this is part of integration. Mapping the elements from hub to subscriber. I've had situations where, and this can work different ways, but the build team does the mapping, but the subscriber, the app, if you will, does the actual integration. Okay, that's a model that works. And I'm not here to say which model. I'm here to say, better make a decision. And I can help you with that or somebody can help you with that as well. Customization of the elements and data quality rules for the subscriber. So you build MDM, but it's not perfect for your subscriber. You know, honestly, it probably never is. Hopefully you're 90 plus percent of the way there. And hopefully the easter to that gap to 100 percent, most of the time, it's let me take MDM. Let me expand MDM, so it's right. But sometimes we know the subscriber's gonna do it at their end and they're going to exclude MDM and you're always kind of living in that. Oh, okay, I get that. But who does that work? Every new integration will have some. And hopefully you have solidified your MDM hub in the enterprise. Hopefully you have established your credibility in the enterprise so that the answer to where does it go is back in the hub, back in the leverageable place, the place with governance, the place with data experts. Let's pick on some applications. You got my data domains here on the right from before. Let's run through a few applications and think about this, fraud detection. A lot of companies are doing fraud detection. It's not just for financial anymore, but what do you need for fraud detection? Now I'm suggesting a few things here. You might look at that and go, well, we need our assets as well in that. Okay, okay, do your mapping. Do you, do your mapping. Put it on the roadmap. And where you see all of these applications needing customer and product and whatever else, that can prioritize your MDM roadmap, can it? It should. Anyway, fraud detection needs, I'm just gonna say customer, need your stores, need your product, need your contract, need your policies. I think you can see why for every single one of them. But let me ask you this. If you're stepping up, if you're an application owner and you're stepping up to do fraud detection or to take it to the next level, we're all doing it, but take it to the next level with AI in your organization. Wouldn't it be great if the customer was already mastered according to my rules from before? Wouldn't it be great if that were in a leverageable platform with an SLA and that was taken care of, it was true data as a service, wouldn't it be great? Well, most organizations unfortunately have to say, yeah, that would be great, but it's not true. And if they're not thinking about MDM, they're not gonna cut off a discreet effort out of the budget to do it, may I say the right way or the architected way with MDM. So they'll do a job with it for sure, but it'll probably only be good for fraud detection. This is where I'm giving you the warning. Please build these subject areas that are gonna be needed throughout the enterprise, throughout the course of time to build it in master data management. So it would be great if all these things are there. It would be great to have customer, product, account, contracts and assets, and what have you when it comes to building your call center chat bot. Now this is a new app for many organizations, obviously organizations that have call centers, have had them for quite a while, but adding chatbots is something new. And when you step up to do that, you need a lot of information. And I'm suggesting what some of those pieces of information are now, do any of these apps like the call center chat bot, do they need any of these enterprise data domains to be 100% fully mastered and error-free forever like assets? No, they need what they need. And it's probably a subset of that, but it's hard to sit here from a data perspective and know exactly what that is. That's why we wanna build to the Mac, to the SuperSet as best as we can. And truly, while you're in there sourcing elements from wherever the data's coming from, that's a great time to source more, to over source and to get what is necessary for the future as well and have it ready. I want you to be ready. Wouldn't it be great? Now, unfortunately, not too many organizations can say this today, but more and more, it is becoming true that applications such as this new wave of AI application that's creeping up and becoming very important in organizations today, that they have this wind at their back by having MDN in place with customer ready to go with data as a service to that application. So the application can focus on, get ready for the application stuff and not the data stuff. And that is what I would really like to see. Self-driving transportation, if you're doing that, you need some different data domains. Customer, customer seems to come up all the time, doesn't it? That's probably why it's number one for a close 1D with product. What else do you need with self-driving transportation type of AIX? Policies, what your policies are. Geography, that's becoming really, really important. And this isn't, this is an old geography like what you might get from the post office. This is very in-depth, informed by GPS, lots of information on every segment. And here it is again, if you're predicting flight delays. I know I'm getting a little specific here. We're not all predicting flight delays, but we're predicting transportation delays and other things, but this is just an example. You need your assets, you need your policies, your geography, and I'm sure if we got a war room and sat there for a day, we could find needs for some others. The trick is drawing the line and providing value. Not providing 100% value, but drawing that line and providing enough value so that your use, your MDM hub is actually used and valued and also informed by the app, right? Some of the MDM owners think that if they thought ahead so much that the apps don't ask for more, that that's success, and I don't think it is because there's always more, there's always more and either the apps go to just do it and not tell you and just make it harder for the overall enterprise, including data, which is a domain of responsibility for an MDM owner, right? Or they'll get it from you and they'll inform you how you can go get it and help them out even more. Again, apps doing app things, what a concept. Okay, marketing, segmentation analysis, campaign effectiveness. I think you've kind of got the drift of this pattern here that I'm showing you, but you need customer, any product, you need equipment, you need media, probably other things. Smart cities, ooh, there's a lot here, facilities, financials, assets, equipment, geography of course, and now I threw in citizen, not exactly customer, but citizen. That's the customer of record for smart cities. Supply chain, yeah, you need a different set. I won't run through them, you need a different set. Have you done this exercise? Have you done this exercise? If so, or if not, if no wonder you're feeling like, well, I don't really know how much I'm contributing to the organization by doing bill of materials. I don't know. I know that they're screaming very loud from the supply chain group, so that's why we're doing it, but I don't really know if it's top priority for the enterprise, if you haven't done this exercise. So just do it, don't let it take very long, couple of days, and keep it active. You may want a council that meets on a regular basis just to keep this active. And this keeps those groups informed, those AI apps out there that are emerging, informed of how they're going to get their data when their time comes. Oral and gas exploration, yeah, I've done this app, supported this app, I should say, and yeah, quite a bit, quite a bit to be effective. This is one of the higher barriers to entry, I would say, from an MDM perspective, but it can and should be done. Doesn't matter if an app needs all of the data domains on here, it should be done by MDM, not in other way. So here is an idea of a roadmap broken down by quarter. So first quarter, and for some of you that starts today because you're not anywhere yet in your MDM journey, but most people are somewhere in here, but hopefully you've gotten past what I show a quarter one activity, all right? Tool procurement and installation, business prioritization, stakeholders and roles, MDM architecture, yeah, there's that, workflow and data flow, MDM life cycle planning, your MDM ops, if you will, and then you've got phase one customer. Now what I like to do is for an organization that's immature about MDM is start with one, get going, get ramped up, get the agile team, the scrum team going on that. And when it's evident, we know what we're doing, when it's evident that the end is in sight, the light is there at the end of the tunnel, delivery is going to happen, then let's bring on another team and ramp up product. And maybe the customer team will flip the supplier or some such thing, you can figure that all out, but you wanna put it on a roadmap. And by the way, data governance, data governance is on here, it's on here for the long haul, it starts on day one, it's not something that, oh, we'll fill that in a little bit later. After we've done customer, we'll do data governance. It really doesn't work well that way. You don't have to do full blown data governance, and I've talked about this extensively otherwise, but I will say, because they're so intertwined here, that in terms of data governance, what I hate to see are these isolated data governance groups where people in the organization don't really know what they do, they don't really affect the apps, but they're meeting some book standards somewhere, okay? And they're ticking and tying things apparently, maybe they're building out the data catalog, but who knows when it's gonna be ready for somebody to use something like that? I'm sorry, I don't mean to be like disparaging of data catalog efforts or anything like that, I'm all for it actually, but you've gotta stay connected to the app. And this is what I tell my MDM clients, this is what I tell my data governance clients, and we find ways to do that because that is very important. So in whatever you're doing out there, stay connected to the app, be a big part of it. So when the app goes to production, you're right there, you're right there, and they know it, they value what you've brought with MDM to those AI apps. So in summary, artificial intelligence applications in the enterprise are about putting AI to work on master data, and we can do a full stop right there. And what I didn't get into much today, but is also true is we're putting the AI to work on master data that's built with artificial intelligence increasingly over time. So MDM itself should have many elements of artificial intelligence built into it. So you wanna look for that in the tools that you buy and so on. And that completes the sentence here, right? Artificial intelligence apps in the enterprise today are about putting AI to work on master data built with artificial intelligence. That's what they're about. And half the work effort is the master data, and the way to master data is with MDM, as I talked about here. So Shannon, that brings me to the end of the presentation. Do we have any questions? We have a lot of questions coming in here, William. Already in the Q and A portion. I'm just gonna answer the most commonly asked questions. Just a reminder, I will send a follow-up email for this webinar by end of day Tuesday, actually. It's usually Monday for this, but Monday is the holiday. It's end of day Tuesday with links to the slides and links to the recording. So William, diving in here. Should data governance and the data strategy drive MDM? Should data governance and data strategy drive MDM? I'm gonna say yes. I'm gonna say yes, because as I define data strategy, it is that roadmap that cross-references applications that are probably already on the roadmap with data domains and those data domains. I wanna have them mastered as much as possible before the application needs them. But secondly, I wanna remind all us data professionals that we shouldn't be sitting here beholden to the application strategy. We should be influencing the application strategy. Many of the apps today, if you will, should be coming from a base of data expertise. So we need to be influencing the data strategy. But once we do have the data strategy in place and we do have data governance to some degree in place, yes, that would influence MDM heavily. Operation data start different from master data management. Hmm, I haven't heard that term very much lately. We've tended to not do them in our architectures anymore because we found that data warehouses can be real time. And so there's less of a need for them, but they exist, of course, and there are pockets when they make complete sense. MDM, thinking about it here right now, I think it's kind of a form of an ODS, but in MDM, hopefully I've made it clear, there's no transactional data that's stored and distributed from within MDM, whereas an ODS has usually a single or a couple different data sources for it and we're building it for a discrete report or two or three or 10, but it's not meant for widespread consumption and it's limited in the data inputs. And MDM is neither of those. There's a lot of consumption of MDM data. There's a lot of potential data that can end up in there so it's not limited at all in that way. AI be used to address data quality issues so this data can be leveraged for AI-centric projects. Yeah, yeah, I think I'll start that answer by starting with kind of where I finished this presentation talking about some of the ways that MDM can have artificial intelligence build into it. And I think that we wanna see that, first of all, the way we've been doing data quality for years and years and years, myself included, of course, is it's not scalable. It's not scalable to ask the data stewards for input on a hundred different data elements. What do you wanna see here and let's go make sure that it's true. That has to be automated and artificial intelligence is the way to do that. So what we see in the better MDM tools out there and of course, most of them have some semblance of data quality to them. So they all apply rules to data as it comes into the hub and before the data is in its chute or in the staging area, if you will, ready to be distributed, it is of sufficient quality as defined by the tool or the parameters of the tool. So I think that you wanna look for robustness in that area of any tool that you select. Some of you have data quality problems that are too big for an MDM data quality support. They really are. And for that, you'll need a specialist data quality tool and some MDM tools out there just to kind of round out this conversation. Some MDM tools out there don't really have data quality built in but they'll sell you a separate tool or there's a partner tool or whatever that's focused on data quality. So I mean, you wanna look at the overall value proposition and how much artificial intelligence is built in to data quality. And I'll give you just one final example here is a big part of MDM is those workflows, right? And that's where the data will originate by getting the workflow passed around from department to department. Everybody adds their good stuff and at the end of the cycle we have a master record. It's a beautiful thing when it comes together. But what I hate, and I didn't really get into it in the presentation, but what I hate to see is for a lot of that, the workflow to be consumed by and the sales manager will review this and approve it. I wanna know what the sales manager is going to look at and how they're gonna think about what they're seeing and how they're gonna approve it. And I can build that in or artificial intelligence can build that in. So we can heavily streamline our workflows. As a matter of fact, those of you that have MDM in production today, I challenge you, you are going to need to speed up your workflows with the modern tooling and modern possibilities probably by 2X in the course of the next two to three years. So be thinking that way when it comes to what you're going to do with MDM that's already in production. Low hanging fruit for AI be data governance and data quality. Low hanging fruit for AI. I think about automation as being the low hanging fruit for AI. I don't see data governance. I believe it was mentioned and was it data quality as well, Shannon? I think that's what you said, right? Okay, so I don't see them as necessarily apps for the enterprise. I see them as enablers of apps. And so I think when it comes to AI apps, they are things that have a little bit more direct impact on the bottom line of the organization. And most things you can't do well without data governance and data quality. So I think low hanging fruit for AI to me, it's automating tasks that are being done in obviously non-automated ways. And so we're looking for those things. It may be, I don't know, website analysis. It might be supply chain analysis. In some cases, if you're implementing a chat bot, that's all AI. So I went through many different examples in the presentation. I think those are more the low hanging fruit than supporting a discipline like governance or quality, which I just mentioned in the prior answer that they should be supported by AI as well and have further than that into real apps that support the enterprise. Any tips on how to get multiple functional stakeholders to agree to ownership of master data management? Let's just gloss over that and pretend it's not a problem now. That's a huge one. Thank you for bringing that up. It's outside the realm of technical architecture and things like that, but it really gets to the heart of MDM success. The question was about multiple functional stakeholders, getting them to work together and so on. I mentioned in the presentation, I like to have a council for MDM. I like that to be current and future subscribers to the data. Current subscribers, they're gonna have issues. They're gonna want things. They're gonna need to know what's coming out next week, what's changing, et cetera. So they can change rapidly with it, but future stakeholders, they need to see all this happening in real time too, so that they can prepare, so that they can know how they can best interact with the MDM hub, so they know we're here, what we're doing, that we are attending to data quality, we're attending to scalability, we're making this data as a service for you. You don't have to go do it yourself. And we've already done customer and product at least to this level, whatever it is. And we're working now on assets or whatever, whatever the case may be. We'll have that done in a month, at least to this level. Never complete, right? But you have to be able to articulate what you're gonna have done, when and how, and offer up the possibility of accelerating progress. Those of you that have some success, you may offer up to the enterprise with a straight face the option to accelerate progress by spinning up multiple scrumtees. Those of you that don't have that success that have nothing in place today, I don't suggest you do that yet. I think you're just creating too big of a risk situation. So therefore you're kind of saying that the dates are gonna be more out there, but I like to offer up that possibility. I don't like to say no. I like to say yes, if you, Mr. or Mrs. application owner can offer up some more budget, we can prioritize that subject area and you too can have all your subject areas ready to go, date is the service. So I like to get a council together, have them meet monthly, maybe, maybe, depending upon how you've done this, maybe you can pony this onto your data governance function because data governance and MDM are hand in hand and in some organizations, you're just going to upset people by making them come to an MDM meeting and then to a data governance meeting where you repeat half of the same topic. So anyway, get them together, get them talking, create service level agreements, create real documentation, create a portal page on the internet where information can be have, remember, they're thinking what's in it for me. So tell them up front, tell them up front. I've done other crazy things too. I remember once when, this will take this a minute, but I remember once when my team put customer into production for the first time and we had a couple apps teed up and they were already using it day one, great. But I knew that we had, we're sitting on something special here, goal, the goal of the organization, right? And so we created a banner, if you will, a poster and we put it up in the lobby, had our lovely faces on it, but more importantly, it had information about what it means to you. So if you were off our radar or maybe we knew about the supply chain that we hadn't had time to connect, here's what it means to you. Here's the data that you can have. Here's our roadmap. And here's all the great things that we're doing with that data. And we are building this with data as a service in mind, let's talk. And that was successful at making that MDM program go forward within the organization. So take it outside the box too. We have a lot of great questions coming in, but I think we've got time for one more here, William. So I love this question. Master data equals chicken, AI equals egg. Yeah, is there a question in there? There is more looking for a comment on it. Well, master data is the chicken, AI. I guess it could be argued the other way as well, which comes first. Well, I guess it depends what you believe came first. Let's say they go together and you can't. I made the point in the presentation a few times. You can't do AI without master data. You can do AI with poor master data, but I don't want that. I want real master data here in place that's scalable into the future. So it'd be great to have the chicken laying the egg of master data, if you will, so that I guess another chicken can consume the egg to just completely, fully exhaust and damage that I can now achieve for you. William, thank you so much as always for another great presentation. Really appreciate it. Lots of great questions. Thanks to the attendees for being so engaged in everything we do. But that is all the time we have. Look for this webinar today. Again, just a reminder, I will send a follow-up email by end of day Tuesday for this webinar as Monday is a US holiday with links to the slides and links to the recording for everybody. And that's it. So thank you so much, William. Thank you. Hope you all have a great day and stay safe out there. Thanks, William. Thanks all. Thanks, Shannon. Thanks all. Bye.