SAPPHIRE '06 PARIS
SAP NetWeaver MDM: Empowering Business with a Single Version of theTruth
Sunil Gupta, SAP
Rod Hall, Senior Project Manager, Ericsson
Gary Biggar, Data Quality Manager, Diageo
Sunil Gupta, SAP
Quite an audience here. Ladies and Gentlemen,Welcome. My name is Sunil Gupta. I drive solution marketing with SAP. With here, are my partners and colleagues in crime, Rod Hall from Ericsson, and Gary Biggar from Diageo. You may also know his company as a maker of some fine beverages. What we are going to take you through real quickly is first of all, exposure and understanding of a few key concepts around what is Master Data and understanding the pain? Getting to a little bit of the solution followed by some real life customer case deployments around MDM. Simplifying your business, let us look really quickly at the problem, the state of your business today. Here is a simple diagram, the state of your business. Different departments, you have Master Data today trapped in different silos.
Different processes generate Master Data every time. You have an incomplete view today in your business within the four walls that you have. Now does the problem stop there or does it get bigger?
Well, if you expand the picture of your ecosystem, you are doing business with your distributors, your suppliers, your end customers.
A lot of faith relies on sharing accurate information about your products, about the vendors that you are dealing with and so forth. The problem really is not limited to your business alone. It extends out into your ecosystem. And as you can see here, you are sharing for example a different view of the same product. Maybe this is a boot for example or the same end customer. And as the result of doing that, what is the pain that you are feeling? Lack of unification really impacts your bottom line. Here are some sample cases of some customers of ours today. And you will hear more today, on what their business pains are. Whether it’s lost sales for GE, whether it is compliance, financial reporting, and Sarbanes- Oxley compliance for example with Nortel. To a large B2B distributor who is looking at $400 million dollars being lost every year, a direct 20% margin hit, because they did not have an accurate view. In this case, of the products they were selling. The problem is really large. The question is what is the
impact on your business today?
And we can almost guarantee it in every line of business. You will see this problem. Now let us look at some of the principles behind Master Data and understanding the data quality challenge that exists today. On the left hand side, you see a chart that shows on the vertical axis, data quality versus time as events are taking place on the horizontal axis. What is happening? Over time, your business is dynamic.
Data gets generated. Events happen. You go out and do an acquisition. Low and behold, you are now acquiring other systems that you need to integrate. So as an example of other events, you may have a new product introduction cycle; again the same dilemma. As you go through with these event cycles, you are constantly trying to make sure your data stays accurate. The challenge is as soon as you start fixing it, something else comes along and now you have the same problem. You never get to a point where you have optimum data quality. So what should the ideal solution be, because the events will never stop happening? Which you want to put a solution in place, which will hopefully minimize the time it takes, if you look at the graph, to recover from the impact of these events. If you do an acquisition, how quickly can you integrate the data about the new products, about the new customers, about the new suppliers within your landscape and continue your business? How can you have a foundation for creating business processes that you can creatively design, make them flexible, to drive value in your business?
That we believe in short should be the ideal solution. This is what brings us to data unification with NetWeaver. It is designed such that you not only consolidate and have one view about your products, your customers, your vendors, and any other data elements you designate as Master Data. But you facilitate its consumption within processes. If you look at the world of Enterprise SOA, you are trying to get to a point where these business processes can automatically consume the right data at the right time at the right place of what is needed within your environment. And that is where we are going with MDM. That is what it brings today as data unification within NetWeaver. Now some more data points about MDM itself. What are some of the things about this? This is a fifth generation technology in finitely configurable SCMAs, which means you can quickly and accurately create SCMAs to model your own data. It is a forward cell phases of consolidating, of harmonizing or syndicating central management, some elements about prepackaged IT and business scenarios. If you are for example in
the consumer products or retail business, you may be familiar with things like GDS, Global Data
Synchronization.
It is something that is built and delivered right on top of MDM. Or you may be looking at a specific vertical application. And we will see that, for example customer data integration. Today we have over 230 customers in many vertical industries. And obviously, we will have some customers who are speaking with us today. Now let us look very quickly through some of these scenarios. Consolidating is a first step in really laying the foundation for quickly and accurately consolidating the data coming from various sources. Whether it is SAP or non-SAP sources. This is probably the single most important step for anything you do. If you do not consolidate, cleanse, and normalize your data, you really do not have Disclaimer: Transcripts are the exact conversation for the presentation and may contain grammatical errors and other mistakes.
SAP is providing these transcripts, so you can read the presentation in its original form.
a foundation for moving forward. Various elements around products or customers, etcetera, can be quickly harmonized once you have this foundation.
Once you have that foundation, now you are at the next step of actually being able to syndicate this data as needed. Now different companies are at different stages of the business. If I did an acquisition today, it may take me one year or two years to get to a point where I have one central view of all of my systems. During that time, I need a strategy, which lets me share data from one system to the other. That is what harmonization or syndication in this case lets you do. And then Central Management where you are centrally defining the information elements about your Master Data. And then synchronizing that, syndicating that to any system or application, which needs it.
Now let us look at how you can apply these core building blocks to various business problems. You will hear today something around vendor consolidation but let me walk you through some examples. One of them is around the ability to manage information about your products. Products demand not only the information about the part number or the material. But for example, images, part specification documents, and any other pieces of information you designate as Master Data. What are the value and elements we bring here? The ability to manage taxonomies, hierarchies, built-in units of measures and conversions. The ability to search dynamically on part numbers or other attributes where you only know the partial strength.
For example, you know part of the product. That it is the color red. You do not have the ability to know anything else. You want to search for all red hoses for example. You can quickly and easily find this information in the system. ot only that, if you are in the business of requiring printed versions of these informations, whether it’s printed catalogs, or circulars you can do the same thing. Global data synchronization allows you to syndicate trade item data in one language to data pools like Transora or UCC Net. And we are working on supporting some of the other data pools. For example, that is shared here in the European and other areas. Now what this lets you do is build on top of MDM. It comes with its own predefined data model UI validation rules. Which allows you to bring data from your ERP systems, your R/3 systems. Augment that data as required for the data pool and then validate that you are publishing this either by trade item number or by the supplier. And then publish this data into the data pool.
Again, this is built on top of MDM. It utilizes all of the various NetWeaver components, such as XI for syndicating the data, and that is really, where the beauty of the architecture comes in. All of this information, by the way, you can also analyze data by bringing the data from MDM into analytics, into the world of BI. And that is where the predefined, pre-built integration makes that happen. One last example, something new that we are introducing with customer data integration; gives you the ability to quickly have one view of your customers. Bring together various elements. Define matching and consolidating strategies. I would encourage you once we are done here if you have time to go to our demo pods here. And take a look at some of the new matching functionality that is coming out in the Q3 time frame. I think you will find that to be of interest. The road ahead, new matching immersion capabilities, that we are defining. The ability to again locate identical duplicate records, to expose the matching process. Again vastly enhanced from what we had earlier. And this will again simplify the way you consolidate information.
So to summarize quickly the benefits of data unification. It accelerates process change. You cannot have processes without having an accurate foundation of data. Neither is it useful to just have data without using it in processes. You need both of them working together. And that is what we are delivering with data unification and with Enterprise SOA. It helps you improve productivity. And companies like GE that not only see top line benefits but bottom line benefits as well, so both cost reductions and revenue generation. Which is an important point. In cases like Ebuild, we have seen SKU maintenance, which has gone from $19 a part to $6 a part for example. But enough of that, why don’t we hear something more real and tangible from our guests. And I will introduce Rod.

Rod Hall, Senior Project Manager, Ericsson
My name is Rod Hall. I work for Ericsson in Sweden. I am a part of the IT management function there. And I want to share with you this afternoon some of the experiences that we have had working with MDM in some of the areas that Sunil has mentioned so far.
I am going to take you through a series of pictures here. Going to talk about the plan, we have for MDM. Our time plan that we have worked to so far and how we have actually gone on with it. I am going to tell you a little bit about the landscape that we work with in Ericsson and the prices that we are trying to support. And then think about the approach to match the criteria. And then talk a little bit about the experiences so far. Because I understand, there’s a certain amount of skepticism about the ability of MDM to meet up. But I think that’s not our experience so far at least. And we will talk a little bit about that anyway.
A little bit about Ericsson first of all as you can simply see. Ericsson is a large international corporation. We operate in virtually all countries around the world. We have a 180 operating companies. These companies do a certain amount of trading with each other as well as with the outside world, both at the supplier and customer level. We employ 55,000 people, just over the current state of play. We have focused on already MDM pilot on the purchasing area. In particular trying to consolidate the data that we have actually got for the various purchasing systems in use. To summarize in the picture we have got. We have basically two large global systems supporting purchasing activities within Ericsson. One, which we call our Market Unit Solution, which is supporting over 95 companies at the present moment in time. Growing almost by the month. And we have a second large system which basically supporting the major supply functions. Which tends to be mainly in Sweden and China from our point of view. And in addition to that, like most large corporations, we are not 100% pure SAP. We actually have about forty other local systems that are also involved in purchasing activities. We had to basically look for a solution that would cover that full spectrum of systems. Therefore, what we needed was a tool that would support our Global Master Data Group. We have established a Global Master Data Group to try and to manage this monster of trying to contain different views of Master Data within the whole corporation.
The group has been functioning for several years now. And largely it has worked with manual tools up until now. And we needed to give them a tool that would enable, to actually automate some of their work. Take away some of what we call the grunt work from what they are doing. And put the uplink to the brainwork. Use their brains rather than the time they have to spend sort of crunching through this sort of manual task that they have had at the present moment in time. We started to look last year for suitable tools. And as an SAP shop, I guess it is fair to say that we looked first and foremost for something that would actually work with our major SAP systems. But let me say, it is not really enough to do that. We have to have a tool as well that would actually work with the whole of our landscape including the non-SAP systems that we have got as well.
We had an evening session, where we invited our Master Data Group colleagues in and we spoke with SAP. And I think we all left the room fairly infused that MDM looked like a good solution. But of course, as we say in English, the proof of the pudding is in the eating. And although it looked very good on paper, let us say, we really did not know what we were getting into. But I think our experience has been pretty good. And I will come back to that a bit later on at least. We were looking really for a tool that would enable us to work with SAP and non-SAP solutions. And most importantly as well, we are fairly committed to the idea of actually extending our use of service-oriented architecture within Ericsson. And using NetWeaver as the sort of driving platform for that. We wanted a tool that would actually be compatible with that. We were fairly convinced from talking to SAP that in fact that MDM is very much in that line of work.
We set up a pilot and we did not want to sort of go forward on too broad a front to begin with. Because like everything else, it is a new product. And it is new for us. We wanted to sort of test something first. Unfortunately, with a company as complex as Ericsson is, even a pilot can be a fairly large and extensive. We focused very much on the supplier, Master Data area. And in Ericsson, we have something like a 130,000 vendors. And within that of course, you’ve got a fairly certain number of employees so it’s normally the case that we have actually two vendors for every employee. But we have sort of a large number of vendors, of course to deal with. Roughly about a 130,000 in the present moment in time including duplicates. Now that of course is what we are trying to find out. So we do not know how many of them are unique vendors yet. That is what we are going to find out now over the next few weeks. We are talking here of everything from sort of one-person consultancies, to multinational corporations that we are dealing with. It is really the whole mix of different vendors that we are discussing here. The application we have chosen to target, first and foremost was to try to support a BI solution. Which its task is to basically report on our global span across the full range of suppliers. Currently, up until now we have been trying to do this manually by aligning the different
suppliers. A fairly gusty sort of task that somebody needs to do and to make it more automated. That is what we are trying to get towards. That is our target system. To get there of course we need to clean the data. And that is where we are at, at the present moment. And I will come back to that.
We also wanted to introduce the whole concept of MDM to our global Master Data Group. Because we see this not finishing of course with suppliers. This is the first step we will take. The intention is that they should then become the sort of standard tool. This will be used for managing Master Data within the whole Ericsson Corporation. And it hopefully is going to be enough to be a major labor saver for the Master Data Group. And enable them to do much more clever things with their time than just crunching through doing both realignment that they do today. And I think last but not least, we had to set up the operation infrastructure. We had to actually place this within our fairly complex SAP landscape. And that we have successfully done. The project timeline looks something like this. And we started around about August last year after the summer, just the sort of preparation stage. Went through the classic of missing a blueprint of how we are going to hang it all together. And then we went into realization phase. And it is fair to say that our realization phase is taking a little bit longer than you might expect to take because we basically had to wait for SAP to release the various service packs. We needed service pack three but we saw some value in starting the service pack two, to get
some experience. That is what we did. We did service pack two, then we upgraded to service pack three. And then did an in-house realization after that. We’re also fairly cautious, it’s fair to say. We were expecting there to be some problems because it was a Ramp up product. And those of you who have been with it, I think it has been my fourth Ramp up. And first experience has helped me; it is always worth best not to assume anything. I have to say that this Ramp up has been an actual revelation. It has been really very smooth. We have had very few problems in the actual product itself. So I think for those of you thinking of doing this, we are out of Ramp up. You should not expect any problems. But we did not get any in the Ramp up either. We probably could have been more aggressive in our timelines than we actually were, as it happens. We had our few challenges in just getting the platform delivered to us but that is nothing to do with the product. It has to do with our hardware supplier.
Where are we today? We are into cleansing mode today. We went live with the product about three weeks ago, loaded all the data across as a one-time load. And we are into cleansing of the data now.
An exact how long it’s going to take, not absolutely clear. But reckon on probably the next four to six weeks at least of more work.
It is going to take us up to about summer break in Sweden doing the initial cleansing. And then we will get into a sort of Go Live with a BI application off the back end of that after the summer’s actually completed. That is where we are at today. The actual data flow itself, we are using, as far as possible, the various NetWeaver components we have got. We use XI to drag data in from the SAP solutions that we’ve got linked to. That has worked extremely smoothly, been our first major use of XI within our landscape. We have had it there for a while. Have not really used it very much up until now. This has been our first big use of it. And it has gone extremely well. We also use XI to pump the data out of the MDM tool, back out to the business bench ware system at the end of the day. Once again, that has gone fairly smoothly. For the non-activators since we’ve chosen to use one of our existing pieces of architecture, Purple CBeyond, which is now owned by Sun Microsystems. And in fact, I do not think it is going to be called CBeyond very much longer but nevertheless it is a product that we have some experience and confidence with within our organization. And we chose to use that to pull in the non- SAP data into the MDM. We have actually used a mixture of data for different data sources and pumped
it into MDM. That too has gone fairly well and smoothly. We have had no real issues regarding the Sea Beyond stuff either.
And you can see the sort of data flow that is described there. I mean I won’t take you through it point by point. But it’s very classical of the hierarchy we use for consolidation mode. And that’s what we’re into at the moment, by the way. We chose to do consolidation. The consolidating data from different sources, leading it the same way in those different data sources and consolidate you a single view within our business warehouse system. That’s what we’re trying to do as a first step. I’ll come out of it later to what we’re planning to do with it going forwards. Because we see it very much as a first step in this area.
This is the detail scenario. Again, pretty much classic out of the book stuff, for those of you that are familiar with MDM at the moment. We haven’t really tried to do anything particularly clever or different here. We’ve tried to stick as far as possible to absolutely full standard SAP approaches. And I think it has paid dividends, because we’ve actually had good support and I say very few problems. We certainly haven’t made the mistake that we’ve made in previous years. We tried to customize SOP to exactly meet our requirements. We’ve taken it as a good enough solution. And our experience has very much been good enough. We haven’t had a very much resistance to it either. That’s been a positive experience all around. And down at the bottom of the thing you can see the range of systems we’re connecting. We’ve got the two major inputs being the two global SAP systems covering all those companies. We’ve even got some local R/3 solutions out there. Because we have, a number of those set in place, supporting various industry standards, like cable solutions and things like that. And then
we’ve got majority of them, scholar systems, those in Sweden will now scholar. It’s a small ERP system, which will be used for our very tiny companies typically of 10-50 employees. And we have various other systems as well in the mix as well. And that’s going to be extended later on.
And it’s pretty much the standard sort of flow, as you’d expect of an MDM solution. We have matching; try to match something like about 90% automatically and pretty much achieving that at the present moment in time. The key plans we’ve got involved from the business side of things, largely the MDM group, they’re our Master Data Group. These are the central functions supporting our Master Data function within Ericsson. Supplementing them, we’ve got the sourcing group, the Central Sourcing team who are providing certain key attributes of data, that only they are actually to set in. It’s a collaborative effort between the two things. And then we pass the data when it’s finished over to the BI team to actually consolidate into the BI solution. I won’t go through this in detail but just to say one thing, you have to spend a good deal of time getting a matching strategy right. And for those of you looking into the sourcing area, you’re going get this as a bit of free consultancy from us at Ericsson in the material that’s sent out. It’s really giving a lot of what’s gone into getting this scoring table working as exactly as it should to. I recommend this to you as a very good starting point for your own work if you’re looking
into this area. Because it certainly did take us a bit of good time. Well, concise right away is the sort of intuitive way that MDM works. Was very easy for us to talk to our Master Data Group. They liked what they saw. They understood exactly what we were trying to explain to them very early on. And I think the tool is very professionally put together from that respect. This is a very good thing. But it does take time to fine-tune this to get the right results. Experience as a whole and we’ve had actually a very positive experience. I think MDM has been quite well received by our Master Data Group. We’ve had a small group of people from Global Master Data team involved so far. Now we’re spreading it out to the whole team. And because it is fairly intuitive, because it actually matches the way that it would naturally tend to work in any case, we’ve had very little resistance. Often when they see a new product, you think you get very enthusiastic to begin with, and that enthusiasm tends to sort of go down as you get into experience. That’s not been our experience here. It’s been a generally positive steady state of enthusiasm all the way through. This is not to say we haven’t had our moments of ups and downs and doubts and so forth along the way. But we’ve come through all of that. And I think we’re actually in very good shape now. And generally speaking, the enthusiasm is actually good to carry on with this
product. Not only in this area, but looking to extend it into other areas as well. I think that’s a very positive signal to as to take it to stage. And to stress, it’s still early days for us. I mean we still have been using it for a few weeks in real terms. We have us an extensive testing as well to base these experiences on. I think it’s fair to say that the enthusiasm is also taking hold. We want to see SAP moving faster to do, to release what we call standard business contact tool to us. I mean it’s got a very good tool set already of standard links within to the SAP products. We’d like to see even more of that. And I think it’s no secret that we’ve expressed to SAP that some of the standard financial elements we’d like to see. But much more assistance from SAP and how to manage those going forwards. Let’s say that our Master Data Group today has to separately manage consistent Master Data in 21 different SAP clients, you can see it’s pretty much what we call a horse job in Swedish. And so we’re not absolute that can grind away, maintaining the same data in all of those different clients. And it’s not a fun job to do. Nobody would actually wish on his or her worst enemy. So to get some assistance in that area would be extremely valuable to us.
So I think we’re very much pushing our SAP to up the pace, so to give us those things out there. We’ve had some challenges so to say. MDM was a product that SAP bought. And I think we rather naively expected it would slot very nicely into the sort of standard global security concept we’ve got for the rest of the landscape. And to be fair, it doesn’t actually slot in quite as nice as you might like. We’ve had a few challenges in terms of getting it working. SAP has listened to us in that and I think some other customers have said the same thing. And we expect to see some improvements coming. But we have managed to get an acceptable security solution working at least. I think that I can sum up by saying that overall MDM has really fulfilled the goals we set for it. We’re pretty pleased as a customer generally. It’s doing the job we expected and it’s doing it in a fairly undramatic way. And that’s what we really want to see in this sort of situation. It’s not very exciting or sexy if you like, but it does the job. And that’s really, what you want to see. General comments as I said earlier. The Ramp up experience
has been unnaturally positive. I’ve been waiting for the first big foul up to actually come up that we’ve needed the help with. But we really actually haven’t seen that. And we’re past that now. So generally speaking, they suggest that the product itself is pretty stable. When you go through Ramp up and you don’t hit major issues then you know you got a good product underlying the whole thing. The use of XI has also been extremely positive. Although it’s a new product that SAP’s brought in you might expect that it would have had some problems working with some of the other NetWeaver components. That’s not been our experience. It’s going extremely smoothly together, with both the Portal and the XI. The two major areas of NetWeaver we’ve chosen to integrate it with. And indeed so far, at least, it’s gone well with BI as well. Although we haven’t really explored those capabilities to the fullest extent yet. I think things to watch for, I mean clearly it’s on the leading edge. You got to make sure that you’re up to fairly high levels with the various service packs in your landscape. Luckily, we at Ericsson tried to keep fairly well up to date with the latest service packs in the NetWeaver area. We had to make some fairly unanticipated updates however, keeping our port carrier to make sure that everything worked. And even in some of the back end systems, that was a bit of a surprise.
That was probably one of the very sort of risks we faced by being in the Ramp up situation. It didn’t actually pose any problems, but it could have done. And one of the things we have found and this again is a bit of advanced warning for you. Tuning the Microsoft platform that we run on, to get the right performance, has been a bit of an issue. It’s always the case we’ve managed to get good performance, but you have to tune in one of two ways. You have to tune the platform for the mass load to begin with. You have to then further retune the platform for the sort of the incremental changes that come afterwards. And we had a few challenges to begin with to get the initial perform, the initial we wanted it to. And then having to retune the solution to beyond that. And as you bring on new data elements, it’s a constant sort of balancing act between. To attune for the initial load order, to attune for the ongoing performance. That’s something to be aware. It may be that we haven’t actually sized the platform correctly. It may be our own fault. But we have had some challenges in that area. Something to be aware of anyway. And make sure you focus on the severity set up quite early on because it’s going to take longer than you think to kick up your major SAP customs. What are our next steps
planned? We plan to upgrade to service pack four, as soon as it is possibly available. August we believe. Possibly, even sooner. We’re looking for the new capabilities that are listed here basically. Dun and Bradstreet integration, improve matching capabilities, and the performance enhancements that I mentioned earlier.
What we plan to do with it when we actually got this service pack in place, is that we plan to then go on and start to extend the thing. We want to move from consolidation to at least harmonization mode.
The rest of the discussion we should go straight to centralization mode. But that’s not something to be discussed. It’s the case of we don’t want to jump too far ahead. But clearly, we want to move from just being a consolidation shop as far as MDM into harmonization. Really driving this supply Master Data down. We want to get into using workflow in a much bigger way and to use the guided procedures and there’ll be forms in connection with that.
We need to extend the base. I’ve listed some systems before. We’ve got more procurement systems out there that we need to combine with. That’s what we’ll also be looking to do. And next, we want to look to extend to new data types. Customer Master Data is going to be our next area, more than likely.
We’re fairly convinced that we’ve got a job worth to do here and this tool can really help us in that area as well. And we would like very much with SAP’s assistance to get into modeling other objects. Indeed the financial objects that we’ve discussed as well. So therefore, we’re looking to collaboration. We don’t want to custom build too much ourselves. Ok? That’s all that I wanted to say from Ericsson. There will be time for questions at the end. But I think I’m going to pass over to my colleague, Gary, now to tell you about his experiences. Gary.

Gary Biggar, Data Quality Manager, Diageo
I’m Gary Biggar. I’ve been with Diageo which is I guess we can grab the Diageo slide to see where we are, I guess the data module within IS in Diageo. I’ve had a lot of experience, I guess looking at global product data issues. And I’m sure we’ve all got experiences of trying to align product data across your organizations.
And from that experience, Master Data Management for other objects such as customers and vendors and it came along. The last two years I’ve been spending time looking at SAP’s offerings in this area.
But I guess in Diageo it’s not all SAP and that is part of the problem. Similar to Ericsson, we have many legacy systems and we’re going to hook them all up together. I’ll just tell you a little bit about Diageo PSVN. Contrary, I do see many people here and I heard that there was a rumor going out that I was going to be giving away product samples at the end. And so if anybody’s here just for that, then you can go. Because there aren’t any on me, I’m afraid.
But Diageo of course, is the leading, I guess drinks, brand of drinks manufacturer. And we have quite a substantial amount of money being made out of that activity, close to two billion sterling. And we kind of split our business pretty much across the world fairly equally. Between North America where people obviously drink Johnnie Walker, Europe where most people drink Gordons and internationally where most people drink all the rest. We’re reasonably well set up in terms of global reach. I must always say though we do promote responsible drinking. And as a Scot of course, when I got a kilt on, it’s quite difficult. But I do manage it. And we put a lot of our effort into educating the public about misuse of alcohol and so on. 1% of our pretaxed profits go in that, as you can see. Let’s look at Diageo’s data challenge. First of all, I must say that Diageo has a bit of strategy, which is a surprise to quite a lot of people. Having a data strategy, working out what you’re going to do with your data is not something I have experienced a lot in other companies. But what we’re attempting to do with our strategy is to
promote high quality data management processes. In many cases when people put in SAP, R/3 or they put in various solutions, they focus on their order to cash, their accounts and reports and purchase to pay. They expect the data to kind of just look after itself as you might imagine. There are five key themes in our data strategy. First of all, you have to design your data. And data management is about quality, enrichment, and so on. You have to have a strategy of how you do your data maintenance. Are you going to centralize it, are you going to devolve it? Are you going to make it self-serve? You have to work out that you’re not going to introduce protocol and even data migration. And make sure that you can support that activity. And then think about what infrastructure you need for that.
This strategy is, well as a Word document it’s about 30-40 pages. We measured our progress against that strategy around about two years ago and that diagram tries to show just exactly how good or bad we are in terms of our data management. One of the things that I should draw your attention to is the data management capability by geography here. We don’t necessarily have an evenhanded approach to how we do data management across the organization. And that’s part of the challenge. We in effect looked at that assessment and we certainly saw the chance to make a positive step change in data quality. What we realized fairly early on, was that if you want to increase your data quality, you have to find who owns each individual piece of data. That can go right down to granule level. But the key thing that we tried to do was to get at a senior level sponsorship of data objects. We took our procurement organization and we said you own vendors. We looked at our packaging organization and said you own the components and the bells of material in effect. In order to cash, we said you own customer. We then create a forum of senior business owners of those objects. And that means that we can take issues about data quality. We can report data quality to that student committee. In effect, challenge
them to make Master Data global. Cause if we don’t have that will from an organizational point of view, then all system efforts are futile. We looked at some of the issues that we’re facing. We believe that commandeer to our provider’s worth much reduced cost in terms of software design and systems’ design.
We said that data is not common. It’s captured in many and various legacy systems. We’d like to put it into single source. And then we can leverage the origination of that data in one place. Rather than have lots of people originating the data more than once. Data is manipulated a lot to get, to consolidate reporting. Especially within our Sarbanes-Oxley regulated business. Doing that isn’t compliant in Sarbanes-Oxley. We have to understand, normalize the terminology when we describe objects. For instance, mapping keys in MDM is one of the things that we would have to do. Our resource level to handle data maintenance, have risen. And I know that is a direct result of, I think, of implementing SAP. But when I say they’ve risen, they’ve just become more visible. They were always there. When you implement SAP, people realize data management or data maintenance becomes greater a key activity. Our data quality is not consistently measured. In fact, we weren’t doing any data quality measurement. I don’t know if anybody can say how clean your product data is. What would you say
50%, 60%? How would you measure it? We worked our way of measuring all our quality against our objects and I’ll talk a little bit about that in a minute. Data, the next are from, is that data facilitated in the context of single applications. Procurement, North America think of their vendors. The thought doesn’t enter their head that they might be using vendors that are also used by procurement in the UK or in Europe. And we’ve under invested in the technology support all elements of data. I don’t know how much we’ve invested in our ERP program and so on. But we’ve invested a fraction of that in data management technology. Therefore, it’s quite easy to justify that we’ve got these million pound systems and we’ve spent thirty-five pence on data management. What are we doing to realize our data strategy? Well our data strategy resulted in three key areas. First of all that flies in the wrong order but XI’s in that area of linking all those systems together. Our data dictionary is an in house solution and that’s to define the data. Because one of the things that we realize is that we don’t have consistent
pieces of data. So, we created our data dictionary.
SAP does not have a Master Data definition solution. We put our in house system in place for that. We then created; what we realized is what we need is something to keep our data quality up. So, the data cleansing tool kit, things like the base of quality, integrity, and harmonization and so on. MDM teaches for that very well. But it’s key in that area that we have the KPI reporting capability too. MDM does not report on itself. We use say BI to do that. As far as the Master Data maintenance and management, MDM fulfills our requirements in that area. And if you look at those scenarios or components in detail, let’s take a look at them. The data maintenance, our objective for data maintenance is to gather all the data about be it customer, product, or whatever into one place centrally. To do that, we have to get it all into that one place and Sunil talked about consolidation and so on. But for instance, vendor data
maintenance we do internally and customer data maintenance internally. It is done by different groups in different places to different levels of quality and so on. If we want to get consistency to that data, then applying a single process to that makes a hell of a difference. Our basic steps with that would be to consolidate our sources, cleanse them, harmonize them and create some data capture interface.
Now, MDM itself, you can maintain your data in MDM. But our feeling is that we need maybe a webbased interface to do that. You map the data to the outbound receiving targets. Create your workflow and this great strong workflow support in MDM and distribute that on a scheduled basis. Be near real time or on a nightly basis. Then you begin your data maintenance; seems easy. From a data-cleansing point of view, there are occurrences when a business rule changes. You might find then in your own organization that because you’ve started your own process differently then the whole of your customer file has to have a certain field changed and rules.
Now that’s quite difficult in an R/3 environment for instance. And requires somebody with some
programmatic capability I suspect. However, you can easily make a change to your global file on a certain aspect, according to some rules within MDM. You could do that if you had already consolidated your data in the MDM repository and then release to the R/3 environment. Or for instance if you want to do it as a one off change, you could do just exactly those steps that I’ve outlined there. Use it as a tool to make a global change for certain aspects of data. For a data quality point of view, we use our data dictionary to set quality rules. So, you may say for example; well there are some of the rules up there. The invoice to customer number must have a trade class of zero. Don’t know exactly what that means. Don’t know the detail of them. But these are rules that we’ve set and we measure those rules in a system that we call our Anchors. But in effect, you have to assign owners to those and create the KPIs against them. That is where we take those KPIs from that Data Steering Group I was talking about before. And we can actually say to them, by measuring all the various quality rules that you may have about a Master Data object, you can then give a percentage. Any single number that says data quality
for product is at 61%. You can then draw a graph and show it to the stakeholder. From migration point of view, we foresee using MDM in a migration environment. To take the data into MDM and then to use it to load on extra parts of our business into the R/3 environment. So for instance, we may acquire a company or we want to expand the amount of in market companies that we have using SAP R/3. Our last of the steps would be to define the object and extract the data. Do the cleansing if you like, in MDM and then release that over time into the R/3 system.
Global Data Synchronization, being in the consumer goods industry we are keen to do that. This will indicate what our architecture would look like were we to do that. Our PLM and bar code management processes would probably still be in some sort of product data management area. And then MDM and MDM DDS would be used to serve up the data-to-data pools. By using the MDM product database to feed into that GDS environment, it can also synchronize that into R/3 and legacy environment. Let’s just talk about journey and some bullet points. Establishing the organization first is certainly the way that we found success in getting BI into the MDM, the Data Management roadmap. In establishing a data steering group established the stakeholder group lower level to actually make those forums for people to discuss data issues. Then the infrastructure came next. The data dictionary is a huge part of our
infrastructure. And we have already installed MDM at our foundational area, a layer in Diageo. But there is a great future for MDM in Diageo, there’s an MDM roadmap covering for instance vendor data maintenance. That project is up and running. We are looking at data quality cleansing and consolidation if you like, already in progress. And supply chain data alignment as you might imagine with GDS in mind is a major project for us too. We’re quite busy. But our vision is based on a single source of the truth and other MDM presentations have seen this highlighted too. We’ve all arrived at the same thing.
One single place to get an authority, definition of your products, your customers, is a very useful thing to have. We established a data strategy to achieve that. It was focused on a forum to transformation. And to make it happen we had to target the business support. It’s not an IS load activity. It’s not a systems load activity. So we told them, retold them and we’ll tell them again. As Johnnie Walker might say journey of a thousand mile steps might seem as that we’re a long way to go but we’ve got a good starting position with SAP. That’s all I’ve got to say.
Thanks. Thank you very much.

Sunil Gupta
Thank you very much. I think we have a few minutes to take a few questions if you’re interested. We’re also around here if you want to ask some questions. Okay? Thank you very much.

• Read the solution brief (PDF, 98 KB)
• Read the IDC report (PDF, 155 KB) Log-in required  Our Server

• Read about IBM's solution (PDF, 1.4 MB)
• Read about HP's solution (PDF, 281 KB)