 Hello and welcome. My name is Shannon Kemper. I'm the Chief Digital Officer for Data diversity. We want to thank you for joining the latest in the monthly webinar series data architecture strategies with Donna Burbank. Today Donna will discuss master data management aligning data process and governance, a couple of points to get a start in due to the large number of people that attend these sessions he will be muted during the webinar for questions we will be collecting them by the q amp a panel. And if you like to chat with us or with each other we certainly encourage you to do so. And just to note the chat defaults ascended just the panelists but you may absolutely change that to network with everyone to open the chat and the q amp a panels you'll find those icons in the bottom of your screen to activate those features. 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 any digital information requested throughout the webinar. Now let me introduce to the speaker of the monthly series Donna Burbank. Donna is a recognized industry expert in information management with over 20 years experience helping organize organizations enrich their business opportunities through data and information. She is currently the managing director of global data strategy limited where she assists organizations around the globe and driving value from their data. And with that, let me turn it over to Donna to get her presentation started hello and welcome. Hello Shannon always a pleasure to do these and hello to everyone on the call nice to see some familiar names and I know a lot of you joined these regular leg regularly so appreciate that. Good to see you all for those of you who don't join us regularly. This is a series each month we pick a different topic in data architecture or data management. But one of the nice things about diversity is that all of these are recorded for as long as the diversity exists so that you can go back to any of these you might have missed. And then hopefully, if any of these other topics are interest you throughout the year. Be great if you joined us on the ones coming up so without further ado, today's topic is master data management. And as Shannon mentioned I run company called global data strategy where we, we do consulting and do this for a living and we are just seeing more and more interest in master data management, coming up this year. We've been doing this for years but it does seem like this is even a much more of a hot topic than even in the past, probably because it ties into a lot of the other hot topics we're seeing which is data governance and data quality and architecture and all of that so it's tons of opportunity and the value of managing your core data assets which are your master data is super valuable which is why everyone looks to do it it's also can be a challenge and so hopefully what we can offer what I always try to offer in this webinar series is just really practical real advice from folks who have done this for living and hopefully avoid some of the pains and scars we might have had in the past and share some of the good things that have worked so that's the goal. Obviously this is just a high level. I mean this could be a whole series in itself just master data management there's a lot of pieces to it which, again, all of them at their core are pretty simple it's just putting everything together in one piece which makes anything this cross function like that can tend to be complicated, but we'll go through that and hopefully kind of demystify what master data is all about. If you've if you've joined any of my webinars in the past you've probably seen this framework tend to use it a lot that this is the framework we use in our practice aligns. In many ways with things like the Dama, the embargo data management body of knowledge. And we use this in our own practice of time so master data management is a key part of any data management effort and any data strategy we always try to make it make anything we do look strategic. I can talk today strategically across the organization, particularly with master data management it is the linchpin that makes your company run right so you need to align it with your business strategy to make sure that we're getting that correct single view of the customer and product and why we're doing it. There's a lot of other touch points with master data in this framework data governance is a big one right master data, why it's master or core a lot of companies are using that term now is because it's touched by a lot of groups and what what happens when you have data that shared by a lot of different groups you need something like governance right so that naturally fits in a lot of these other things like data quality that's a core part of master data management data architecture and the modeling and the hierarchies and also almost all of these. Tuts in some way or form master data which which makes master data super powerful, but also takes a little bit of planning and coordination across these areas because when you do get it right. That's what really drives your business strategy and really, you know those are the companies that are working really efficiently and really well across the organization. So what is master data management, I'm an data management person so I love my definitions with that in the glossary right. So, and I'd like to use Gartner they tend to have some really good definitions I know there's others out there but you know master data is the consistent set of identifiers and attributes that defy the core entries of the enterprises according to customers prospects citizens suppliers sites hierarchies chart of accounts things like that. And obviously master data management is the management of master data right so, but, but really I like their definition and that it does talk about it's a technology enabled discipline, but it allows business and it work together for all of the data stewardship semantic consistency accountability right it really touches all of it. There are master data management tools out there. I recommend you use one, you can build them yourselves to but you know there's a reason that folks have built tools around these, but it's not a technology solution only right and that's what makes it sort of complicated you have to get the right people in place. I like the fact that they mentioned can can semantic consistency will get will get into that and this you know what is a customer. What is a product, what is a citizen right all of these things that drive your business that seem like it would be so simple, you know the more that it drives your company there's probably a lot of nuance to answer that even that semantic consistency or the, the basic data definitions and data model and hierarchies around what is the thing is really important because everyone again the more people that touch it they all have their own viewpoint around that so that's what we're talking about. Another way to look at that and this is an actual picture from an early cave dwelling in Paris I think it was me when they were excavating excavating some of the tunnels obviously I'm kidding, but they. When you look back, we are visual creatures right and we often you know when you think of, you know, the cave people way back they drew on the walls, things that were important to them pictures of themselves the animals, you know, that was probably their early master data what did they do for a living they hunted right and you could just tell that I know I'm a big geek that's probably what I see when I see higher glyphs was early master data, but it really is right and I think I'm also if you've been on any of my other webinars a big fan of data modeling, you know, and data modeling visualization because I think it just makes it really clear, you can see right there cave man, you know, sometimes they've been kind of washed off you don't see the full you know cardinality rules on the cave dwelling. But they fear they had them right that a caveman could hunt more than one animal right. So anyway, but I do think that the beauty of that is that it is visual and you can get see the relationships because that is the core of master data is that it has a lot of relationships with other areas. Sorry for my bad jokes today. So, moving on what is master data, and this is another odd Donna slide but one of the reasons I love my job on a good day is that, as a consultant you go into so many different kinds of companies right and you really understand how their business ticks, and then the data behind that and I just forget that you know I'm actually seeing a lot of stuff about a company because data runs the company right I'm, I'm not actually thinking about that I'm thinking about the data itself right but it really does drive and we think of examples of master data. I find it fun to kind of think of all the different companies we work for especially now that everybody's doing data, the idea of being a data driven organization is hot right so it's not only anyone who's been in the industry. You know, for many, many years, we've all worked in finance we've all worked in some of the big you know government and things like that. And the idea of you know customers and product is sort of your classic master data, but now that so many different kind of companies are coming into the less be data driven I find it kind of fun to see the different different types of things that are considered master data, and then also the business impact of that. So, you might wonder about some of these so the other thing is when we go, we often do kind of data strategy today that assessments, and we try to understand either opportunities or pain points around the data and I find it interesting that we go through the interviews there's generally that one story, or and I know we want to be data driven but there's, you know, the, you know, examples that kind of drive things the anecdotes right. And in this one company, it was cheese, right. It's sort of funny I felt like it was in this bizarre horror movie we're talking to people and if you brought up cheese they had this look of not the cheese. I still bring that I'm like what am I missing how can this cheese life be so scary, but they actually lost over a million US dollars. And basically when you think of their master data, it's not even the menu items right it's the components or the, you know, when you think of products and product components and materials that they're making a hamburger or something. All of those pieces of it right the, the meat the cheese the bread all of that goes into it needs to be costed and managed across the whole product lifecycle from when it's developed in the kitchen to one supply chain costs it to marketing sets the price to when it's on the point of sale system and you order it into in the restaurant right and that is master data and that's a master data process that aligns with the business process and that was broken. So what what it was cost should have been costed at from supply chain didn't match what the price was for marketing that ended up on the menu when people bought it so it was a good add things to this menu and people added this new kind of funky cheese, and it wasn't priced accordingly so this great new menu item that they had that was very popular actually became the bane of their existence because they were losing money every time someone ordered it, you know, worst case scenario right you've basically priced your product wrong. Well how that how that came about was everybody is now freaked out about cheese every time you mentioned the word cheese people look at you in fear, but that again that's a matter of thought you know I think of Swiss cheese as master data it really is right that was their core components. Now we had another one $2 million lost over baby bottles, we work with a company that manufactured and sold baby bottles and a lot of it. They sold through Amazon. And if you've worked with Amazon they have very good master data and they make you match it so if I want to sell product on Amazon I have to match their master data standards. And if you don't you get a fine in this company had such bad master data for their own products, they could not easily match, you know, Amazon's format, and they kept getting fines, and they had over 2 million it was worth selling it because they sold much more than 2 million it was a very large company, but just just because this literally was one to one a master data issue they were literally getting a fine because the format of their data. And one of my favorite ones is he might wonder about the dead fish up in the upper right we actually did a webinar on diversity with this company is or not a company organization is the environment agency of England. And this was a bunch of scientists and what what the environment agency does very cool organization, you know they track. And they also track how many cows there are and how many organisms there are in the water and we had a bunch of scientists get together. And what is their master today what are they counting what are they tracking it was living organisms. And then someone brought up. Well, what about a dead fish is that still living organism and I just came back with us a living organism with a status of dead. It was a conversation but it was very funny, but what they came out to be was that an organism, whether it's in, you know, I'm going to show that I'm not a scientist the thing in the water the small organism in the water versus a cow versus a fish could all be classed by this an organism, and the master data could have been the same and that was a huge aha moment for their scientific discovery, and they're publishing out metrics to the public they had a big open data set. So again, you probably don't think of dead fish or cows when you think of master data, but it was that's that is their business over there or you know was it was to really track organisms across the, across the country. When the lower left, you know, we've worked with hospitals and trying to get credentials for their, their doctors right so if someone's doing surgery can they get into the wing of that hospital they credential to do heart surgery. And you certainly hope your hospital is doing the doctors that's, you know, operating on you as the right credentials but that was a big part of it so doctors are master data customer is sort of that middle one that is classic we work with a big insurance company and they had done a lot of really cool advanced analytics. This was an insurance company that only only insured people like you and me you know high net worth individuals that have multiple mansions across the road globe and own several companies and have their Renoir paintings insured with them and things like that right so it was very beneficial for them to understand their, their high net worth customers and they did a lot of research and did a lot of web scraping and things like that to understand a lot of information about how many companies to go on and things like that. The problem was they did all that cool analytics but when they tried to look at their own policies, they couldn't tell you know is john Smith, the multi billionaire who owns 17 companies, or john Smith the courier who delivered our pizza today. It has an auto policy with us but isn't, you know, the john Smith that's a billionaire right so they did all this cool analytics it was a very important business, you know, decision for them but they didn't have good master data so they kind of had to stop all of that. Another one on the right, which was master data, I'll bring it back to the environment agency again, and they did great webinar and their data modeling efforts a couple of years ago now on Dataversities library. And another piece of their data, their master data was regions or they call them catch passments right of, if I'm, if I'm dividing up the country of England. What are the different areas that these organisms live in and they they couple of maps and things like that so actually location might might be something for a retail company whatever different locations around the globe. So you could argue that maybe that's reference data is in location reference data right and this is all subjective what what is one person's reference data might be someone else's master data right it really depends what your organization is doing for a living location is one that often is master data, the other thing and I won't go through all of my Donna rants but I'm also amazed that the importance of this information, multi billion dollar companies or organizations or governments we've worked with. We have this in somebody spreadsheet right then and often their name though it's the Mary spreadsheet or the Joe, the Joe spreadsheet that has all of our locations of all our retail, you know, site across the globe and I'm, I'm still flabbergasted by that I'm not anymore because I see it a lot but that that's kind of scary that is those those are your golden nuggets of your organization, probably the worst one I've seen, or best, if you want to tell the story was with a big water company, and they did a lot of the merger and acquisition is always a great use case for master data because you're trying, you know, the core assets of one company to be aligned with another. So they, this water company had acquired a smaller water company and wanted to get all of the customers or the subscribers. It was a small local water company and they literally had it in a paper notebook where it was written down on paper and with ink. It was a little hard to scan in they were trying to automate that process but you couldn't it literally was on paper you don't see that a whole lot. Usually it's a spreadsheet but you know even even in this day and age that was only a few years ago. It was literally written down on a notebook so who knows that hopefully that gives you some different examples of master data just to stretch the story a bit longer and I just started to go through you know again this is why my job can be fun. Just to keep going on the cows right so the environment agency was actually counting cows that was one of their master data items of living organisms. And then we worked with one retail company or it was a, and they actually sold adult cow suits and master data was you know cow suit adult size, large, you know, but is large L or is it the word large you know so we had a good laugh over that of. Anyway, that was one of their master data items or you know when you look across a trademark of the upper right that was a master data for one of the government organizations we worked with. In the lower middle you know wells or drilling if you work with oil and gas. I mean that again a lot of the master data when you drill. No, no pun intended there when you drill into it. Seems so obvious until you know when it is the core of your company you're trying to tell me that an oil and gas company doesn't know what a well is, but it took years for it have industry standards they actually have some master data industry standards across multiple companies now. And then the DPM is a data modeling effort. And that took years to really decide what wouldn't mean by a well because because it's so common, a lot of people have a different view on that. That's right next to the broccoli because of course right so broccoli if you're a restaurant could be a one of your core, you know, material master to the left of the well is a school. So what is a classroom super relevant even to what day diversity is a classroom a class where people sitting it could be a virtual classroom with people sitting in it for them it was just the concept of a class that had a curriculum. You know, a lot of types of things like that now what's a hospital is location what's a part and you can kind of see all of these and kind of understand that, you know, we had we work with a truck company or auto motive companies not only the automotive what kind of parts go into that auto and how do we track all of that so again, I won't, I won't kill this one to death. Did you kill something to death sorry I'm going to move today. But, but hopefully it gives you some examples of the breadth of master data and if it's new to you what kind of might be some of your master data items so the classic ones I hopefully gave some interesting examples, but customer still does. You know, it really is one of the more popular ones that people need to understand do we know our customer, and that could be citizens or students or patient, you know, it's not too much of a stretch. But here's a fictitious company, sporting this company, and these folks want to understand their customer and I will talk more about this but they want to get that ubiquitous classic need for I want to 360 view of my customer. Why, why is this even important you might ask that those are fun stories Donna but why do I care right. So this is a sporting goods company. This is Stefan Kraus. I'm also a skier so this will come out on big story I use right. He's a he's a ski instructor. I'm a sporting goods company I sell ski boots. This should be our classic company right he's 31 he's fit he's a ski instructor in St. Merritt's he lives right there in the St. Merritt's Valley. He's he's he's actually a Nordic skier to he won the end of the ski marathon he's been in our loyalty program. We know how he likes to consume he likes text messages and we send him ads and he buys everything online, but when you look at how much he actually spent. He only spent about 500 euro for that particular year which isn't a lot right because he gets all his gear free he's good looking he's bad things and a ski instructor he gets you know people want him to be a spokesperson and you know he's actually you think he's our best customer but he's not because when you look at the data and really understand everything about him maybe we need to either market him differently or maybe we don't market to him he's not he could be a spokesperson but he's not going to actually be our top seller. So, when you look at, you know, the data again there's another Stefan Kraus, and he's 62, and he's a banker in Zurich and you wouldn't think he's going to be our classic outdoor sporting goods customer in fact he likes sports but as football and he wants to go see his soccer players football team, he doesn't actually do the sports so much except for when he does go on vacation once a year. And when he does he wants the best equipment he can possibly afford because he works really hard. And he makes a lot of money and I friends like this they're like I only do it once a year but I'm going to have the best equipment out there right and he spends it all to go on holiday and he likes to go in the store, and he wants physical mail he's kind of more old right. And he's actually your best customer. I will clarify that all of that analytics we just did around Stefan Kraus is not master data is the opportunity you can get from having good master data right the master data is who Stefan Kraus is at the banker, who lives in Zurich and spends a lot of money with us or is the ski instructor who lives near St. Merritt's who is not our best customer and different age and how do you identify who is Stefan Kraus might also be interested as one father and one son and maybe the father's buying all the stuff for the son right that's all the stuff you can do with analytics, but master data and I always joke this is the story of my life I've never been the sexy, you know front end we're always on the back end getting the stuff done is the master data that enables all this cool analytics how do we know which Stefan Kraus is it do we go by his name, his age his address all of that. So that gets into kind of I just want to kind of specify that a graphic like this really helped me earlier my career understand master data reference data transactional data and how that fits with warehousing and data management and maybe you won't have the big light bulb I did by this one slide but help me hopefully it'll help you. So, again, this is sample transactional data again maybe this is the sale system or track everything so Stefan Kraus. I don't know which one. But on a certain date that he, I assume purchased a product which is a telemark ski boot with a certain code 250 Euro, and the location was St. Merritt I assume. That's the store again I have to get Matt metadata in there right is that the store he bought it is that where he lives. I'm not sure, but right and then Donna Burbank also bought the same ski boot in Boulder, Colorado. Again, is that where she lives is that where she bought it but whatever. And then, you know, Stefan, but assume it's the son here who maybe it's a father, but he's probably a dad because in Zurich right he bought a North Face down jacket. This much he bought Wendy who bought some yoga pants at the same store in New York, etc. That's your transactions, right the fact that Stefan bought all this stuff. That's probably what you start summarizing the data warehouse. The master data is who is the customer is Stefan Kraus was the first one the son, who bought his boots and same words and the second one the father, who bought his down jacket and Zurich. Perhaps right that's what you need to understand which was Stefan Kraus also your product is going to be your master data. Now you'll see that that same telemark ski boot is a different product code. Is that a typo. Is that wrong. Is it because we have different product codes in the US and Europe and they should be different right that what is our core list of products that we sell. And is there a difference in Europe in the US is the same boot package differently other different components is a plastic different based on the region all of that is your product master data and all the components that go into that your price, all of that location. It could either be your master data. You know that could be these are our store locations right and we know that there's a Zurich Switzerland for a store and there's a boulder store we know that reference that could be referenced it depends how we we model it out. Clearly I would say some of your reference data is maybe the code right and right there I can say Boulder Colorado that's a state code co but CH is the country code for Switzerland so gosh we we need to even manage our reference data better because those are not apples and oranges right so anyway a lot of examples from one example but again your transactions are the fact all the stuff stuff on bot or Donna bond. The master data is who's Stefan and who's Donna and who's Wendy and who's Joe right or what a location what they mean right so that's kind of shows you how all these things hopefully fit together. I kind of talked about this before but again one person's master data is another person's reference data, you know it could be that, you know, regions or markets locations are second, you know, it's a secondary they just lists, or it could be again I'm a map company or I'm a in the environment agency and I track locations for limit right, or it could be the location of my store which is a first order thing. I'm not too techy or nerdy about that. Right, but but you know that I almost think of reference data as the little cousin of master data right there often your code lists, things that you know your country codes your state codes things like that that might kind of be managed more statically worth talking through people get themselves in and not about that a lot I almost mad you know so I do care but do we care what we call it's how do you manage it right. So how do you how do you understand all that process to understand how do I get a single view of who Stefan Krause is and it takes a lot one one is the architecture. How do we model out how do we know what the unique identifier is is it for just first name last name, first name last name date of birth, social insurance number whatever right. How do we match words and understand that, but also the governance who's accountable across these different systems right of, you know, point of sale and who enters the data and who, if we see that there's more than one step on cross who know. How do we, who decides that is it the system, we automate that sometimes. Sometimes you never want to we work with some medical companies and the risk of even if there's a 99.9% match that I know that this step on cross is the dad and the other one is the son. I'm going to amputate his leg I want to do heart surgery I want to be absolutely sure I have the right patient here. So often there's a kind of a human in the loop where somebody has to validate that before matches approved right, not always that extreme is something like medical but you know often the data is important or you might do that as you're getting the model and then maybe you trust the model more and more for the match rules but data governance and stewardship and who quote owns your master data can get really tricky and we'll talk a lot of more about that. Often how you understand who quote owns or updates the data is through business process right how is the data updated across the journey of you know maybe you know someone signed up and they're in the loyalty program and that's where we get their address but then when they buy the product they change their address. They told the sales person but does that get sent back in right there's a lot of different touch points to that data and that's why it's master data. Right that's why you do need to understand both business process or the customer journey. And I think often that's forgotten we think of it from our system perspective. Okay person was. You know it came in the store then it goes into the sales system and then we invoice and that comes out of the finance system, but what was the customer journey I often give a good example for master data when. When master data works well. That's a really good customer experience and the only place in the world I have any. I don't know leverage or feel important is with airlines because I'm a consultant and fly around a lot right, but I had an example. I was with I was trying to book a flight online from Denver to London there was a client I was going to a lot. And I couldn't book it online for some reason and they said, could you call so I call. And they recognized my number clearly and they say miss Burbank is that you calling about the flight you're trying to book from Denver to London. Would you like to take the same flight you took last month and we booked that for you. It wasn't creepy, but it wasn't creepy to me because I'm a data person I knew they should know all that data they should know my phone number I gave it to them. They should know that I'm booking online they should know because I logged in with my user ID and they should know all of these different things because I've seen all my previous transactions. So, because they were had good master data they knew how to identify who Donna Burbank is. They had a really good customer experience right. Compare that to something like my insurance company where I've been trying to update my address with them online for about three years now and they still send throwing place and I call, and they update one place and it doesn't you know, almost every negative customer I've ruined this for all my friends and family they said you know I had a problem with the bank and I updated my whatever and they didn't I told them they had a master data problem. I actually have a friend that does it all the time, because once you now see that you almost can't see any customer interaction that's, you know, bad with your data or they, I don't know I checked into a hotel and they didn't have my number and they so they didn't know my the room preference I had right that's a master data problem, or I have my bank sends me both the credit card statement and my checking account statement and my savings account statement is separate letters they still send me letters even though I've asked please I want it all online right that's. How do they not know that Donna Burbank one person is Donna Burbank has three different accounts and still one Donna Burbank right that's master data right so thinking your own life how many, you know, either positive, which is when master data works well, or negative when you don't have good master data. Because you just thinking your own life and you know we're in your own company right you probably have a whole lot more in your own company but sometimes it's fun to just get any experience you have how could master data have helped with that right. So there's different ways to implement bastard data. And sometimes it's an evolution. And then there's different modes of master data so the example. So for reporting trying to get that 360 view of customer that that's either analytics or just basic reporting. We call that almost analytic master data where I'm getting that single view of customer. And why because I want to report on that I want a data warehouse that shows total customer spend by product, and you'd have good product. I have good data and I need to have good customer master data so often and you know if you've ever done a bus matrix for your, your warehouse, you know, kind of the, your conformed dimensions basically in your warehouse are going to be your master data right. It's often why people do master data it's often the first step right because I think the ultimate is going to help with your customer experience is operational master data where you actually sync the data across systems right that Donna Burbank has updated her address in one system that has engaged to all the other systems so we know we have the single clean record which is why master data and data quality go hand in hand right I am passionate that you always want to clean data quality at its source, not in the not in the warehouse not in your data, you know, reporting system, because you're just cleaning it up over and over and over and so master data can help enable that let's have a common set of rules and push back to the systems right. Sometimes you have an all one system sometimes on the left you might just register it and still lives in the system. I will read through all of these rights but you know the registry is basically just you have the IDs you're providing across reference right coexistence is where you have in the different source systems and you kind of harmonize back across centralized which which often people sort of have a if there's a bad rap about master data is sort of what people think it is that means I have to force everyone to enter the data into MDM system and I would say you very rarely want to do that maybe for reference data right. But you want to make it as minimally is improved. It's not as effective as possible. Before I get there you know consolidation would be selling one place but it's coming kind of sourced from the source systems and it's different flavors of isn't always even an all or one. What I find interesting about master data you can have an academic view of it, but really, you just need to look holistically at the real world because some systems you can't push back to right it's ideally I think and I'll show my bias. So when people enter the data in the source system that's what they're doing as part of their day job I enter it into the CRM I enter it into the finance system. And you do have some sort of central hub that I almost call that like your liver, the liver of the system it cleans it and validates it creates a golden record, and then it does and perfect world cascade back to the CRM and back into this, you know, the finance system and the marketing system and all these other things, because it when it's done well, just fit into people's day job and when they look for a drop down with to the customers is to clean visit customers or the different locations it's just there. So I think a perfect world. You do have people enter into the systems in their day job. It then cascade into the MDM that does the proper, you know, govern match and merge and push us back and then go into your cell system. And the real world it's, it's just more complicated than that and it might be done in a phased approach. Often it's the sources that aren't very friendly, right, it could just be, again, theoretically they'll be great and it's all API enabled or, you know, event driven when the event happens it cascades to everything, but you're working with a mainframe system, right that can only accept certain, you know, files that can be uploaded once a night right so that's not ideal. But that's just is the reality of the system I have right and when I get frustrated. I love to hate the vendors but you know you'll try to do a master data system and they'll say but I'm the system of record no I'm the system of record I'm the golden yeah I'm the golden record, even something like the one of the biggest misconceptions I think is CRM right your customer, we have a CRM that means we have master data. No, you might enter your leads into this year I mean maybe that really is the only place where things are entered but is that then cleansed and is it cascades across the other systems and is it you know you really need to look holistically and and very rarely if ever is your system of record the master data only source that sits by itself in a vacuum right. Moving ahead, one way to think of it and this is sort of amalgamation of those different styles I feel that previous lives of an academic it is very also common. I like to look at it, I always draw one of these pictures in the real world what systems do we have. How do we want to manage it and what's the purpose right so this could be a standard and this does kind of show that more centralized approach I will show my bias you kind of have to have something that's that liver and being too passive and just doing a kind of a linking everything together generally is is not enough to really get the hope that you need from MDM. So, but again you want you might have a CRM system your in store sales your finance marketing online supply chain all of these different systems that either, you know can be kind of mastered into this kind of showing it like a database and it generally is right. You can kind of do your data quality or match merge. If you have to have a data steward look at it, and they can kind of create these golden records, I kind of have your reference data sets as the cousin, they kind of live there that they are related. And then you can hopefully publish back to those systems of the end user applications can kind of do a look up so they only have the right source, and that often can feed your data warehouse for reporting right. And that kind of complicated but also super valuable that each of these systems that often they all think they I'm the golden record, and the people managing them think they own the system often right that's that's what makes MDM hard you have to get a lot of people and a lot of systems talking together and people are complicated and tech is complicated and at its core MDM is completely simple. I just want to get the same first name and last name for my customer how hard is that. Right, but it really is course did not none of this is brain surgery, it's just getting a lot of pieces working together and that that's why you have to have these approaches I'll kind of show you through modeling it and mapping it out pictures worth a thousand words just you know often just laying this out can can solve a lot of problems. But the CRM system might track person last name address, you know, email spouse of that person so you can send them a birthday card you know that has a certain system purposes. Maybe when you go to the store, you know when you buy something to stress me crazy and I always say no you know what's your postal code or what's your name or what's your phone number. I've actually appreciate it. I won't go through all my ranks but I've actually refused to buy something at a store you do not need my email address I'm just trying to buy something. And when I walked out they were shocked like no you don't need my email I'm trying to buy a thing from you. Here's my money. But anyway, they might get your first name in the postal code where you live. I won't read through each one right, but there's some that's an overlap and that some are new maybe only marketing has the Twitter ID of the person because who else cares right but maybe they do. So what MDM and there's an art and a science to this of what is I put kind of super slash subset is a super set in the sense that it is the best set of records from all of the different systems, but as a subset in that it isn't every attribute that's your identity that might be in your analytics so work with a customer in London right now. And that's their biggest challenges, you know in the analytics system they have there may be a thousand attributes about that customer. They want to manage. But there's a misconception that that's your master data your mastery it might be 10 or 20. How do I uniquely identify that customer, and what are those core and again this some methodology to define I often love to go back to common sense what are those core things to identify and manage that customer that is Don Burbank right my name, my family name my address state email that that kind of stuff right and then maybe the reference data is the country codes and state coach sort of around that. So what what these tools or what you write yourself the master data ecosystem can help with is creating that what we call the golden record right so what's in gray and top or all those systems that do have master data. What wasn't bold in the grad would be those are the good golden records right so I have john Smith I've jack Smith I've j Smith I have john spell drawn Smith. How do I know you know that could be Stefan or Stefan to ends right how do I know. Is that really the same person is it john Smith that has the father and son or this is a very common name. And once I do know that is john Smith. Do I have all the best records even things like do I spell out the word street, or do I abbreviate street. And I know that's where we data folks seem really nerdy and weird, you know, because we get obsessed about that stuff but it does matter because you can't match the word street with the word, you know, the abbreviation ST. Well, you can that's what a lot of these, you know, data quality tools that are embedded in the end to master data management tools can help with right but the more you can standardize. I know that this is one one main street or maybe one system has the main street but not all have the apartment right I mean I won't talk forever about this but that that is kind of the magic that that simple at its core but can be very complex to implement. So, this is how do I a define what those fields are do we all agree. What what is good look like how do I identify what's good, are there standards that we need to apply this data. And then, once you have that then you can publish so the john that is spelled wrong can be updated correctly right the people that don't have the email can be up, you know, updated of the email so that's the beauty of it. If john changes address online, everybody now knows about it immediately would be the perfect world right sometimes. And a lot of these tools and this is where you know I get nervous using the word AI machine learning because it's overused, but there is a lot of automation that you can do with this now that humans don't have to manually go through this right that you know that there is. There are other algorithms you can use to say, hey this looks like a social security number, or you know and some of its human enabled cannot can I kind of make a list that john could also be Jack, or these are common nicknames for the same name maybe we can you know help automate some of that right there's a lot of tools that's why these tools can be helpful. You also as I mentioned before, maybe want a human the loop to say, hey we think this is the same john Smith is this right. And because it got the more, you know, we were just talking I was talking to my team this morning all the funny or strange stories we had, you know, one company had 75 different versions of the word AT&T. Right. And it's only what for four characters how can that be but you know AT in order and is spelled out American telegraph, you know, all of that stuff to be complicated and then just people are complicated names and family names are complicated. We have one school we work with, and I want to have the parents be, you know, have a data literacy course because they know there's a high number of twins for whatever reason and there's folks that have twin girls named Janie and Joanie. And I apart from whether those are fine names or not I felt like you are creating a data quality nightmare for your children right how, how would a system know that. You know, Janie and Joanie same parents, same birthdays, same gender, same, same address, you really only know it from the teacher who looks at me goes oh yeah Janie and Joanie are in my class. Right. Some stuff you just can't automate. I want to teach parents not to do that or Elon Musk that has, you know, you can you have whatever opinion you have of him but he has you know, non standard characters and his children's names right so does a dash or the numbers exist in a, in a name right so he's probably messing up the algorithm himself, probably why he did it right. Anyway, so names are historically complicated and family relationships and people have same name for the family and all of that. You can automate a lot of that but sometimes you do need a human, especially as a massive it's a medical thing or, you know, if you're just doing, I would say just because I know marketing is important but if it's just a marketing campaign and the worst thing that happens is someone gets two emails. Not great, but it's not the same as someone doing surgery and the wrong person right so anyway hopefully that helped to help me when I was first learning this stuff. A lot of complexity in this when you're actually doing it, but as course not that complicated in the business value was amazing right that I know when I'm trying to sell to this person or market to this person or help this person or educate this person. I have the single view of this core master data. So, I've kind of already talked through this but what drives me crazy and the vendors are just as bad at this as anyone else is you know we're going to buy MDM because we're going to give you the 360 view of the customer. And then I do my Donna rant of that's not the 360 view of the customer it's a single view of the customer it's, you need to get the single view of the customer to enable analytics right so your MDM golden record might feed your data warehouse. I want to say customer by region customer and region or both either master or reference data right, or maybe I want to do a cool graph database to understand all my social networks across. Again buying patterns of customers but if you have duplicate customers, you can't really your graph, you've got to have good data to enable your analytics right, or even across your data like data like isn't just garbage dump there and magic comes out and even, even data scientists who are maybe doing, you know, cross broad swaths of data would really like to have that data be clean right so even if social media sentiment analysis do I know that which Stefan crowds tweeted something about my company. I'd like to know which one is it right so again, MDM enables reporting analytics but it is not reporting analytics and I just, I want to stress that because more and more I've seen that and it can be an initial first step, but I've got several customers kind of done it can work in many ways as a first step. We don't have clean master data so we're going to do some graph type analytics and get these patterns across that we have 16 different customers and we think this is the best customer, but until you actually enable that as a golden record and push it back to the source systems, it really is just analytics you have some ideas and you can mainly manually update but it's just a first step and a lot of the analytics tech tools are the CDP platform you have customer data platforms and things sort of. I get a little nervous they market themselves as being close to golden record but they don't go all the way that right they're doing it kind of through analytics you're going to learn a lot from that, they can be admitted in MDM but it is not. MDM or master data management. So, the other part of it and that what I just talked about was a bit in a not super techie but a little bit more in the architecture implementation side. I like this quote from Gartner's a few years old now but it's, it's still holds true that master data management as well is super critical and beneficial, they often fail like anything that's been around for a long time doesn't mean that MDM is impossible just, you know, a lot of folks don't finish a running race doesn't mean you stop doing running races right but but what makes things fails often that alignment with business process or not having the governance and I would 120% agree with that. But what makes it valuable and also makes it complicated right you can't just buy a tool and slap the tool on or do some even fuzzy matching across the stuff and expect magic to happen right because why, or even buy a tool that says we already have customer master you can just buy our thing. What makes your company unique is your master data and you run your company unique way and there will be exceptions if it were that easy that every restaurant or retail company or hospital or school ran the same way. Then, that wouldn't be your differentiator this is not, this is not, you know, that's not how the world works, and you need to think through these things and folks will often say, Well, do we have to do the data model do we have to do the business and do we want it to work right because that's going to actually find the issue. Yes, this, you're not the first person to ever modeled an address. Right. So yeah you don't have to reinvent the wheel every time but often, you know, is that the mailing address is that the ship to address who updates the address right that all that can get very very complicated you have a PO box and address yes if it's the bill to but not it's the ship to write all of that nerdy boring complicated stuff kind of has to be thought through, you can't automate everything. So, I'm thinking of data governance and stewardship. Not go too much into this but there's different ways to think of it can do we, how do we create that owner of the data so I have patient data or product data right would I have the owner of the manufacturing process owns that product data right or customer data is the owner of the CRM system the owner of my customer data that should make you nervous right I don't think so they might own that CRM and know how to manage the CRM, or this one is a hold on a rant webinar and I'll hold back but it's more and more common to say, I've seen will just have a single owner for product or customer and that seems really clean you I would think maybe a single data architect for product or customer. But who owns a, who owns a patient, right I go into the hospital, and I go and I see the receptionist and they take my billing information and my insurance information and then I go to the doctor and they give me a diagnosis, who owns my patient information. Well, the person who gave took my insurance to understand that I hope for that person doesn't know my diagnosis. I hope the doctor doesn't own my billing information they would they wouldn't want to do that either right so there's all because your master data is touched by many many processes you really need to look at it holistically it's too simplistic. And I will die on my hill on that one is too complex simplistic to just have a single owner it would be nice. You might have a single person kind of coordinating teams of people who need to collectively have input. You might have an exception with reference data right I do own the list of these codes or I own the locations or something, but not so much with master data because the point of it is that it's cross function. You might have your, your, your governance by org, you know I have sales but I have North Mexico sales and North America sales and I have European sales and they all have a different viewpoint right or I have a finance team, which is a little different from capability, and I kind of prefer capability or, or, or exchange capabilities don't I still have finance I still have marketing I still supply chain, whether marketing is now reporting up to sales or not marketing still happens, right the name of that department might change but anyway different ways to look at that why am I bringing that up in a master data conversation. We'll talk more about that so there's also different roles right I have an owner that creates some of these rules, what are the fields around a customer and who manages customer. Your word might get into some of these data quality rules, and you might have a technical data student who runs your CRM system and things right. You need to look across all of that. You also need to think of the data model. You know, often when we come in do work with the company we have a high level view of all their data and how it fits together, and we start to color code it might be a staff and customers and products. Your stuff is kind of your, your master data the reference data might be their departments and locations the department of staff works in. And it really almost creates a roadmap and shows the interconnections across your master data find this super helpful and hard to not do that as we go through. Why I had the comment of you can't just have a single owner is really that this is why you need to link the governance and the process and the data together right so maybe our. Our master is things like customer and vendor and material but customer is used across order to all of these are used across your different business processes and there's different touch points and different stewards across each one of those so your business processes are managing all of that data right so the same flow can be, you know, the data is touched across the way so really it's those touch points so customer is going to be used in order to cast and source to pay and for, you know, or vendor is going to be in all where those dots are across right. And each person is really looking at a different area we had a successor we often when we do these we do a workshop and we whiteboard it out and we, we draw out the business process we work with a big aerospace company, and they were having problems with their billing payment terms for their vendors. They always seem to be wrong and they couldn't do financial forecast and we drew it out. Actually, say say say vendor right there has three dots someone had updated the payment terms than someone else along later in the process. Updated the payment terms and a third person update and they just didn't see it because they only saw that it was a very easy thing to then quote, do we have a single steward, or they're an approval process when someone does want to change it. That hard to figure out but if we only had one owner for a vendor, we wouldn't have seen that right we had to actually look through the process and see who touches it and who owns it across the way across that journey. And that's what makes it complicated there's data management processes which design those rules need implementation there's governance of then who does own that or steward it and generally it's a group of stewards maybe there's the different add again the patient. The front office owns my my insurance information the doctor owns the, the, what's wrong with me my diagnosis right. And then it should fit into your business process operational workflow back to when I was saying the MDM when it works well should be minimally invasive I put the right data for my job. I have the right rules in place so that whatever I enter those rules. That's where you need those data owners and stewards and we collectively build those rules as part of governance and then going day to day. It should be really easy, but you do need that little bit of that conflict to get people to agree on those rules and that's needed. One of the things I we like to do is a good old fashioned cred matrix right where is customer data created read updated and deleted and often it's those multiple updates or multiple create or we've got into companies that I kid you not three different master data management systems in the same company for customer. No one saw that right because no one had the time or the effort of the government for the architecture to step back and everyone was doing no one's trying to do the wrong thing they're all trying to do the right thing, but they didn't do this holistically. And that's why master data right and right can be super easy and beneficial when you don't take a time to map out the architecture and the governance and the processes that's where it gets complicated. Again, getting all of the often there's a committee right because who owns patient customer student. So it's, you know, customers involved in sales and marketing and product development and legal and it, you know, all of that finance custom, you know, all of you really need a cross functional view to set up those rules that you're not breaking something for somebody else. One quick success story. This, this brings the cheese incident back to full circle that I brought up in the beginning that it did work out in the end and again, we came in as consultants, they didn't know necessarily what was wrong or why, or even the words master data management. But when we sold it up to the CEO, you basically told that story of, you know, when they created the product in the design kitchen, and then we went to supply chain and the marketing. By the time you got to the point of sale system, there's certain data across the across all of your business processes across several different owners slash stewards. We did all the things I just talked about we did data models, we did a process we did credit matrices we we designed data governance to be minimally intrusive they already had a product launch process, and we added data that conversation did we check the data before we launched this product and make sure data had a first order conversation and do we have the right quote steward that are talking to each other across the supply chain and marketing and are we agreeing on the right price. It's not a small enough answer but it took a bit and we and we sold master data management up to the CEO because you got that she understood the risk and and the reward is part of that so it can work it does work. It takes a little bit of cross functional planning and design but but none of this is rocket science and a lot of it can be done and workshops and things like that. So, MDM is more important and then ever it's more popular and ever because it's so valuable, but you really need a little cross functionally across the world to do that. Again, I do want to open it up for questions, but actually next month if you're able to join us data architecture and governance or will go a little more detailed into that if governance and architecture and new to you that will help with some of the MDM. We do this for a living if you need help there's my blatant plug and with that I will open it up to questions with Shannon to over to you. Thank you so much as always appreciate another great presentation and been feeling the love from the community so just I'll reiterate that I always break that we have the best community in the world. Yeah, or so engaged, right. And so, and just diving in here to answer the most commonly asked questions. Just a reminder I will send a follow up email by end of day Monday to everybody with links to the slides and links to the recording. But diving in here Donna so they're an implementation style which is most popular or preferred. Yeah, I wouldn't say I think the the full when you're pushing the ultimate is a centralized source where you're pushing back to the source system to publish and subscribe like that is actually what's making it usable. I think the most popular unfortunately and I don't think it's unfortunately because as long as it's a phased approach. A lot of people start with the analytical method, and I would agree with that right so even if I know I want to do the full publish and subscribe back to source systems do all of my common business rules is a bit of a safer bet to report on that first and make sure we're okay we're reporting out we trust it, and then start pushing back to source systems because pushing back to source is where you get the most value, but it's also the most risk if this is the wrong Joe Smith and now I've cascaded across the org, then that gives MDM a bad name so I think yeah the full publish and subscribe I do think is is the best is the most. But the most valuable, but I think a step along the way is to do your analytical version first get that trust get people buy in and then start doing the pushback. Do you differentiate mastering data from master data management. There's a subtle question. I'm not sure where you were headed with that but I think the mastering data would probably be that creating the golden record right there there's a whole methodology of how you do that match merge master data management, I think does bring in that more holistic view of what's the business process around that how is this master data being cascaded across the different systems that different do we have to change the different distance process for how that data is entered who enters it. So I think that holistic view including governance and process is master data management mastering data is probably the, the, the technical aspect of how we arrange that goal direct and they're related. But that's how I would answer that question. Okay. Is it essential to start MDM with taxonomy. I'm surprised to see too many MDM effort skipping that and tend to mimic the source systems which tend to fail. I think it depends what it what it is I think for example, product is a classic one with kind of a taxonomy. I think the common aspect of taxonomy and just final master data or your hierarchies is your data model because that's going to help you know is there even a hierarchy embedded in this, and then how you manage a taxonomy is more the values within it. I don't think you have to I think you have to start MDM with the design and I think taxonomy depending on the on the domain is a super important effort is as part of that because that's really, you know, and then who, who, who are the stewards that agree on that product taxonomy and that kind of thing is a big part of the governance and the process around that. So yeah, glad you mentioned that taxonomy is important as well. We just have a few minutes left and there's a reference question if you can go back to slide 21. Hold that up. I'm trying. If I can get up and get that. I can't because it doesn't want to move my slides. Yes, I can. Yes, okay. Well, I'm going to leave that up for the person to ask their question but in the meantime, if you have a known data quality issue and are headed toward an MDM solution when it's the most optimal time to clean the data. Oh, great question. I'm excited to ask that one. Yeah, I saw that one. I wouldn't. Oh yeah, I didn't let people hear it. When do you claim the data like before or after the improvement. I think it's a key part of it right but there's a bit of an art and a science to that right. If there's known data quality issues and everyone's around that and that's why people are asking for master and management, but all means you want to get that is the value of master is part of that data quality they're embedded I see them both the same you can't do without data quality jump right into that and then you're going to be the hero because we solve the billing issues right because we didn't have the right addresses, not always are people on that right. Understand that yet. So sometimes you have to do this carefully so don't tell everyone that Donna Burbank said build dashboards with poor data quality right but you sometimes do that right because often the business doesn't see it or doesn't get it right so sometimes. If everyone's aware, please clean it as soon as you're possible as we'd like to do in a perfect world but sometimes people have to see the bad. In order to understand right in the class and I had a customer this morning saying they've done that successfully right up, you could say hey we need to clean up address data okay sure I have stuff to do. They see the dashboard with all the different addresses you know we should clean up this address data like well thank you for bringing that to my attention we will right so it's a bit of a push and pull there. You don't want to do risky things by you know, doing things with known data quality issues but sometimes it is the bit of the Jedi mind trick of if people have to see it in order to understand it in order to clean it right so there's a bit of a push and pull which is why that analytical first step is often a really good one. So people can see it. And I'm just wondering if you have a 32nd version of the difference between data domain centric and capability centric on this slide. This data would be the data itself customer product right the data, those are the data, almost the things in your data model capability is like a capability diagram of the or like finances a capability of the organism customer isn't a capability it's a thing. So one's a thing one's a noun and one's a verb right in a way customer your things capabilities are the things you do to manage the things like marketing and finance and supply chain and sales and things like that. That might have been 20 seconds. That is perfect timing. People haven't get that wrong, like what these are all words that mean things in the organization but is it a noun is it a verb like what is it you're describing I think people need to think through because I see that not being sloppy a lot. All right. Sorry, go ahead. Thank you so much for another great presentation like I said and thanks all of our attendees again, just for being so engaged in everything we do just reminder I will send a follow up email by end of day Monday for this webinar with links to the slides and links to the recording. Thanks Donna. Thanks everyone for joining.