Is your Master Data Management solution creating new business opportunities through availability of integrated information? Is it implemented on a highly performance-oriented and scalable architecture? Does it accept and acknowledge market trends and pave the way for collaborative implementation?

Abstract

The session explores the benefits of having a Master Data Management solution in organizations. How it can be successfully implemented and leveraged? Panelists shared their experiences on the MDM initiatives in their organizations and what were the challenges they faced initially and how they tackled the same. Panelists concluded that change management is one of the biggest issues with MDM.

Discussion

Tom Thomas: We are going to debate about data, and data is something very personal to all of us. What we trying to address today is the challenge we have in terms of having multiple sources of data; try to bring them together into a single source. So I want to ask the panelists today all of us see MDM as an initiative.

Barry Briggs: I represent Microsoft IT and talk about our experience in master data management. Now we have many kinds of business; we have home entertainment, Windows, Office, server applications and on and on. We also have variety of different ways of contacting our customers. All those things have to communicate relatively, seamlessly for the eco system to work. Different systems have different versions of the truth. We created task forces for what is going on, where are the pain points, what can we do to fix data quality problem. We discovered we had about 37 different applications all of which they depend on the customer data. Naturally when you have an ecosystem like that there will be different snap-shots of data, there will be data inconsistencies and there will be duplicates. Naturally master data management is a solution which provides single source of truth, in fact, authoritative source of truth for customer information.

We realized that there are certain core elements of the data that were shared across all applications of the eco system. The obvious things: name, address, corporate hierarchy, stack codes, and so on. These are the things needed to be centralized in a single location. We put a group of information architects, data architects, examining what knowledge we had about our customers in each of the systems and finding out what was shared and finding out what was specific about each kind of business mission. There are number of different organizations in the world to build customer databases. We have matched the data we already have against that we got in the eco system. By that means we have shared the asset that is available to all applications of the eco system, and the idea behind that was we will then be able to, in real time, any time there is a reference to the customer's core information.

There was a crying need throughout all of line of business applications for clean data. Through various batch extracts from a database we actually were able to go and clean up the data in different applications. This has very significant business implications. It helped, for example, our licensing people to get a better view into what licenses the customer had of our software product and helped them to help our customers to be compliant, to get the full advantage of their enterprise agreement. We did the number of other initiatives with number of other applications on that base and the reason I am talking about is very important point is that by doing this we were able to return a quantifiable ROI back to management, of something like 5X, the total cost of what we perceived to be the MDM initiative all. There is this kind of notion is that you are providing the commerce service, this going to improve the quality of data in the eco system. But by the same token improvement also means change and managing the change, managing the processes of the change is something that is important not to be overlooked point in applying master data management solution.

Bob Legendre: I think Barry did a great job articulating the true need and experiences Microsoft brings to the table. But the key is it is not a myth; it is reality. It is important to have data to make decisions. When you are inside and looking at the different steps within your processes and to be able to communicate with your internal development teams, your supply chain management teams, your customers on the quality fronts, it is difficult to do with no data, let alone having pieces of data and trying to glue that together to make the right decisions. When we look at all the elements of reliability, quality and the supply chain, the productivity, previous supplier through your installations, it is important to have data. You are touching your ecosystem. We are trying to make decisions. What do you do with the data and the decisions are made. So to myself other industries, I have been in data is critical. That is the decision making of your company. If you don't have it you are not making right decisions. For engaging customers, suppliers, your internal processes, your internal processes as well as looking at your growth. It has got to be scalable. Whatever you implement needs to be scalable, because as you grow that data management tool needs to grow along with you.

Ulf Hjertberg: OMX started joined together with HCL in 2004 by outsourcing and market place trading systems. One year later we took another step and outsourced our legacy post-trade system to HCL. We see a number of trends when it comes to security processing and we see a need for cross border mid back office and custody processing platform. OMX is a company that strives for harmonizing the Nordic region when it comes to exchanges and listing. But we have a kind of a problem in the back end of the securities transaction. No one can take care of the cross border trading in the back end. These processes were the foundation for our design of a new back office application platform. And the vision is of course to reach 100% STP. What is this all about? It is all about managing data. I told that we were not the best ones in scaling up the development of the platform. We were the best ones when comes to business knowledge and how to design these from the functional point of view. So we were looking for partner at the time. And we found a perfect match with Penstock within HCL. That ended up in the beginning of this year in a true partnership agreement. In addition to this we are at this very moment launching the first step of a MiFID compliance solution to our customers in the Nordic region. This will take place all over Europe by the end of this week, first of November. This is also a solution created by Pramod Gupta and his team in Bangalore . What is this all about? Managing data!

Pramod Gupta: Penstock is a small product initiative that we have in HCL technologies for financial services. We do essentially the whole essence of our build is that we do foundation services and foundation data services on top of which we build solutions in many cases together with our partners. When the client came and spoke to us they found besides the knowledge of the industry and every thing else, we had the tangible solution which was essentially a master data management product. Because I just explained how we see the master data management as a technology product. They found that this actually kick-started their business application and took them to the market much faster. So essentially we hit success in the first year and just come back to market through MDM itself. What we do is we build a securities processing as a single example.

The second example which Ulf was giving was MiFID, just to explain to you all people who are not from the financial services area. MiFID is markets in finance instruments directive European regulative which is coming into effective in the first of November and just to explain the relationship we started working together on security processing. But the way master data is designed is that it helps you kick-off multiple applications on top. We see master data as a golden source repository, as a data model or comprehensive complete data model for a particular purpose or particular industry which we can use for other purposes also and the classical example is perhaps, without using the name, would have been ACP kind of a solution or CRM kind of application, they used to bring in data from other sources make a comprehensive single data source and then able to give different views. But they were horizontal business areas they were servicing. Now MiFID etc. are applications which are built again on top of master data but they are vertical industry solutions. And the industry solutions based on master data was left awaken at an earlier time. So what we did was took a definition of the data models and put the solution in very quick time to be able to taken to market and just very please to say that as it was last week OMX got message from Swedish regulators that they were the first in the region to comply and give the trade reports to the FSA of Sweden. So there was the first and also was possible I would say it again I would claim it back completely on the financial data model solution that we have. I might call it enigma at the best. It is an enigma because some people see it in a very simplistic manner; some people see it as a over complex and lot of people make an attempt to diving into MDM solutions.

Question (Audience): What I am eluting that what are the security considerations or requirement companies like Microsoft imposing on our partner and how we are going to manage those data of course in conjunction with PCI compliance and I would like to elute that what are the technology processes and technology policies and standard we are requiring from or we should require from our partner like HCL so that we secure our data?

Barry Briggs: I would say tenet really on in the development of our project that we would not keep e.g. PII data (personal identification information), in our database. So we said that is not what we are going to do, we wanted to avoid any semblance of us possibly even having the possibility of an exposure of such data because implications of such a thing are tremendous and the legal consequences are probably the least of our worries, it has been more the damage to the market. We were very careful to make sure that the data that has implications that is not publicly available is not in the Master Data Management Solution. What that means is that the mission of line of business application affords to the data, the reference data, they continue to be responsible for all way for particular missions where there legal compliance is required. They continue to mange those requirements.

Pramod Gupta: We are handling data and e.g. HCL has a BPO outlet where we handle client information, customer information etc. There are security standard BS 7799 etc., which enforce upon ourselves to address the issues of the security. MiFID Regulations has a complete section on outsourcing and that deals with the kind of question you are asking which how you actually deal with your client information privacy. I don't think the issues are so much also in terms of technology security. It is more about legal privacy and emotional privacy. Legal privacy is that some countries don't allow private data can be taken across borders and emotional privacy is that I don't feel for sharing information. If those issues are surmounted, then the rest is possible but outsourcing is now becoming (if I use the term) an extension of doing something by another department within your own organization. That is my answer I hope.

Question (Audience): I want to ask you what principle we use to decide what master data what just good old fashion today?

Barry Briggs: The fundamental principle we use in deciding what master data has to be shared. So it is not something specific, really the reflection of the philosophy that information which is specific to about particular business machine e.g., pipeline data in the sales CRM application, partner prospective information, various things like that really should stay in the line of business application. Information that is shared across entire eco system is really called core data - obvious things like name, address, state, tax id and so forth that something we keep inside the master data management solution. Really what it does is that allows the central kernel that we can correlate all the information across all the different applications together. Not only do we have the core data also we point to the augmentation of the data fuel in all the other kind of business applications. We have the core data point after different line of business applications and they are fully empowered to maintain information and evolve their systems independently to complete their missions.

Ulf Hjerfberg: One example that is very complex for us is when we are handling core protection, we have to deal with data that we feed in externally from different vendors and we have to one way or the another create to copy that we already use for handling core protecting events. So that is crucial kind of data for us to handle.

Barry Briggs: It is easy to think about adding to core data model. And these changes must be resistant at all costs and so one of the key elements of MDM that we have not talked about is governance. And keeping the notion that the core data model stays pristine in some sense really stays as just data which is shared, because it is so tempting to say I can really use the data over here and what is to put it in. That is not the right way to do it. Remember what the enterprise architecture of MDM solution is intended to be "Stick to your Guns" and the governance is absolutely critical for that.

Tom Thomas: So Barry what is the governance structure that you put in place?

Pramod Gupta: It is very important to keep and tight control in terms of what is in a common data and the governance issues are very important from internal. I was just speaking all the rules earlier that if you want take such a solution to market it is littered with many initiatives which haven't really succeeded as a commercial venture. You try and keep it so tight and that you are not value adding too much for a potential barrier or prospect to make a decision to buy. The other problem is by its very nature that MDM kind of solution sets in the middle of many other applications. It is a very invasive kind of implementation you are touching and feeling a lot of applications around you and you are sitting in the middle. I have a solution which delivers not so much value but little value in the financial services area that not so many people buy unless it is regulation driven. We have pure static data which is like customer information, product information, etc. We have business reference data; we call it different mainly because business reference data like market data, pricing information etc. is not static. Therefore you have to handle it differently. Then we handle transaction data also, which is common data pool for handling all the information that you got vis-à-vis your customer relationship that you have with them, position and so and so forth and then you have a business data which is a business view for various internal uses that you have risk managers, sales managers, etc. So in our case I must say that we push them lot further because that what means success for us. If we keep it very tight we are not able to excite enough people in terms of getting intensive. That is just another view.

Question: What is the governance structure that you put in place prior to implementation and post-implementation?

Barry Briggs: So we have an organization which in fact is responsible for quality of data in the IT eco system that is called data management organization, and the fundamental charter, something called information quality that they run in monthly review, all the information quality council, IQC and they are responsible for viewing and reviewing first of all how we use master data. They are responsible for tracking the improvements. They are responsible for helping to guide the plans for expanding the data model as we think about additional applications.

Bob Legendre: I come from different industry. I am looking for data mining. The question is looking at hard core facts so you can make decision from. So in my world I am looking into we see the information and connectivity of all the resources that are important. We look at the performance of our tool, which is the most real time dynamic set of information. So collectively there are data streams that are put together using multiple resources. So the connectivity is important to us as an industry where we are making decisions. We are generating power of certain regions depending on whether we give different results. So the connectivity is important.

Pramod Gupta: Data ownership is distributed. Therefore is always a challenge in larger organizations especially they are more global and more multi-delivery multi-business divisions. It has been a challenge people have tried to co-opt and get sponsor from different organization and that has succeeded to a certain extent, otherwise most of these initiatives can be difficult to implement. We have created a position called chief data officer. I just try to circumvent the issue by having the central role which is more business sponsored role as you suggested while ago and hoping to circumvent the problems in that manner.

Audience: Manufacturing people and supply chain people need to understand that pricing is connected to the right partner or with the right supplier and once I know that supplier is performing and I can run a report. Is he performing or delivering on time? I think that master data management is going to take a different spin depending on who you are, where you come from within your organization and then all link to the customer. To me information is power. Is it the right information that is allowing you to make the decision vs. I am just having the information for the sake of checking a box.

Bob Legendre: You still have to provide the data your customer wants to make the right decision. If they are lacking data, then they make their choices which are changing the provider, changing the bank whatever it might be.

Pramod Gupta: We gave certain instances of financial services, but no means I completely agree with you 100% that when we are talking about MDM as a solution, what we are saying is quality to handle central pieces of common information and if it is a customer relationship which is in the energy industry that is as much of a central data. The only good thing is the treatment of such common data, whether it is financial services or it is your industry, the technology treatment is very common.

Tom Thomas: Ulf and Barry talked about importance of master data in business systems. And Bob you talked about really in terms of decision support area just curious how you, Berry using it master data in decision support you see any applications there?

Barry Briggs: We see master data as a mechanism by which we can keep the eco system clean. One of the applications in the eco system is there are various data warehouses we have, and it is critically important that the data inside the data warehouses is absolutely pristine, so we use our master data management system to cleanse the data inside the warehouse. That gives high level confidence through decisions that are actually driven out of support system which are based on the data warehouses and the models that are connected to that.

Question: I want to go back to Berry as you doing now in the user world or operational world when you are touching the customer that data input has got to be pristine. So the cleansing mechanism you are doing other than the users going in and looking at the data. What are the processes/mechanisms to cleanse the data?

Barry Briggs: We have a mantra actually we talked about inside the IT that we use, and all our programmers saying and all our business people saying to look after for create, and the idea is that for creating any new customer entities in any application, the idea is to build the infrastructure to first look at the inside of the MDM repository, find out for all exists. We take information and we look into master data management repository and is that already there with high confidence. Sometimes there maybe slight variations in street addresses, zip codes. If it is there we turn back their id and corrective version of the data in place of the line of business application. So look at up first. If it is there we use the master data and multiply that across all the systems that in the past have been creating data and we have an overall net cleansing effect.

Tom Thomas: How do you build business case around MDM? What kind of investment involved? How do you justify the investment?

Bob Legendre: I can share point from operational approach. Again when you have data you can make decisions. And you are always trying to improve your eco system. All the way from your supplier through to your customer and the key is as you making decisions you are improving. You are improving flow of information to your end customer; you are improving the cost of your systems and tools and your end product. To me that is critical. The partnerships are establishing even more so. If you make decisions real time dynamic, there is a change. And the change is critical to your end customer as it starts to flow, so to me it is huge. And we make them everyday. I mean if you got to able something to achieve deploy the field and what that back development in your organization to show improvement to show what those changes need to be. So again you give another outlook in performance on your products. So to me it is huge investment and return on all your dollar spent absolutely.

Question: What are the challenges when trying to implement an MDM?

Ulf Hjertberg: I think this is a close cooperation between the business and the IT organization. The business must take the first responsibility for the data, information need for running the business. On the other hand, they must cope with the kind of architect people, how to store it, where to store it, security is essential.

Tom Thomas: I would put this question to Barry in term of challenges any organizations that face in implementing MDM.

Barry Briggs: There are numerous challenges. There are technical challenges of first identifying systems that should be improved by the existence of the MDM application. There are technical challenges of integration those are not to be underestimated. It is actually building a set of web services that are connecting applications to each other. All the requisite and consequent problems are building web service space infrastructure, testing SLAs, exceptional handling so far. Change management is probably the largest single challenge before MDM implementation. Why is my data changing? Why the control of business processes taken away from me? I think that mitigate the fact that if you can demonstrate that all wins available with master data like gold standard data will result in significant ROI, even through quick wins we are able to demonstrate.

Pramod Gupta: I think, implementing central data structure and MDM kind of solution, first of all needs a very strong arm support from within the organization. If you give this implementation to anybody who is the weak or not somebody who can carry the organization through, chances are that it will fail. But that case the moment you are restyling the applications, the ownership of the applications from what I hold in my control to moving it into a central utility. Unless you have complete conviction that is the way to grow, chances are it has failed in the past and it will continue to fail in the future.

Question: Barry did you have any challenges in building common coding schemes to be used by all the units? We are implementing MDM in an apparel company. Apparel companies operate with different brands. Each brand is a division. And these divisions are using their coding system for the past 20-25 years. And they had built intelligence into those product codes. So when we started MDM implementation we said it is not possible to have division specific codes. We need to move into a just running serial number. We have major challenges in convincing each division to go for a just running numbering system. Did you have such challenges?

Barry Briggs: We had precisely that exact challenge. I would say we are lucky in this regard that we have strong executive support for we called single customer id. And so to plan exactly, our implementation essentially added notion of new attribute to each customer record. That actually was because of id and we did mandate applications that going to be using our system would also add that attribute that provide common id space. There was a terrific problem that a quite bit bind to the concept already.

Question: The first part of the question is to do with master data management. I think one of the guys said it is central repository master data. When we look at the financial services industry you are looking at a industry that has need to obtain information on a global basis and the need for the reference information to be updated on a global basis. How do you deal with that? Second part of the question is how do you guys deal with new forms of information that eventually need to make into master data repository? What are the technology organizational challenges dealing with globally distributed organizations and changes master data that are coming to you guys?

Pramod Gupta: I think this is fallacy which happens that this is single technology implementation paradigm for master data management solutions. And that is all wrong. The question that you are asking is so true to life kind of a question. Yesterday we had a session which was about single instance implementation and everybody was saying the world is moving toward single instance implementation. And that is where things were moved to. I am a complete non-believer in that kind of approach. I believe that the solution would or physical implementation would continue to be in hub and spoke manner. And you will have to see because all these especially financial services besides all that you have said. One important issue which comes in handling advance application is latency. Everybody talks about milliseconds, microseconds, sub-seconds and all that stuff. We are talking about that stuff you cannot have everything across so many hops. And be able to get your kind of performance you are looking for. You have different geographies where you are actually doing your business and you have different distributed systems. You will have to make your physical implementation architecture which is suitable for that business. We are doing in implementation and it is very hard to give single line answer. So that is the correct way to go. So third degree normal form has been talked of, fourth normal form and fifth normal form and stuff like that. There are means of handling changes is a fact of life. It is a curse and it is a blessing at the same time. Without change we won't be in business actually.

Tom Thomas: I think you all convinced that you all going to implement MDM. Who should be and what should be the profile of the CDO. Can a database administrator, graduate to become a CDO. What should be his background?

Barry Briggs: I think actually that information architecture is a discipline. It is one that is not as highly valued as it should be, and you think about the person who is data architect, who understands what the key entities in your eco system are in your business. Because what you really do is modeling your businesses. So the CDO and the people who work for the CDO has to be very, very sophisticated, very, very senior, very, very mature individual who get not only the technical aspects of the eco system, And what artifacts actually in place, but also understand business, direction of the business where it is likely to evolve. And to build those data models actually turned out to be fairly sophisticated and fairly IQ intensive activity. I think that the CDO's profile is going to be one that is in great demand in next few years.