Oops! The input is malformed!
Originally published 4 April 2007
I have talked in previous articles about three kinds of master data management (MDM) systems you can deploy. These are a registry based MDM system, an MDM data hub and an enterprise MDM system. The registry based MDM system uses global identifiers and data federation to create a virtual master data system of record (SOR) where no persistent master data store exists. Instead, the master data is integrated at run time and provided to requesting applications and processes that require it. An MDM data hub is a is a system that propagates master data changes between disparate operational systems and can persist a complete integrated set of master data in a master data store as a system of record. In this case, the operational systems that maintain the master data remain the systems of entry (SOEs). The most advanced type of MDM deployment is an enterprise MDM system which has the following characteristics.
A diagram of enterprise MDM is shown in Figure 1.
Figure 1: Enterprise MDM
One of the major differences with enterprise MDM is that this kind of MDM system is both a system of entry as well as a system of record. This means that master data is updated centrally before changes are synchronised with other systems. Getting to enterprise MDM is a significant effort for many companies and may take several years if you are in a position where master data is heavily fractured and there is no single system of record. One way to evolve toward enterprise MDM is to start with a registry based MDM system, then progress to a data hub with a persistent system of record and then finally move on to an enterprise MDM system.
Admittedly, this is a lot easier to say than do. However even if you have worked out how this can be done, it is not just the mechanics of trying to centralise master data that is of concern to many organisations. One of the real challenges associated with deploying master data management is that of change management that may result from introducing an MDM system. In recent consulting engagements and in talking to several of my clients, it has become clear to me that some organisations consider the integration of master data as the easy bit. What is actually causing the caution and the slow adoption of MDM is an even bigger issue that is much more difficult to get right and that is the change management process needed to correctly implement full enterprise MDM. In other words, working out what has to change and in what order when introducing an MDM system is what many are struggling with. Looking at this problem more closely, it is not a surprise that people are having difficulty considering that an MDM system may result in:
The data piece is just part of the story here, and so it is often the case that data professionals may find their skills to be good in the mechanics of master data integration and data management but weak in areas like changing processes, user interfaces, applications, documents and people’s roles and responsibilities. Certainly, from the above list it is changes to people’s roles and responsibilities and changes to human/document workflow and system processes that are often the most challenging. One thing for sure is that you will struggle to implement the necessary change unless you know how your existing processes and applications work with master data.
In my opinion, therefore, a significant contributor to success in MDM is that IT professionals tasked with implementing MDM must start an MDM project by leaving technology aside and taking the trouble to learn how their business works. By this I mean that it is necessary to conduct a business analysis of existing processes to identify what process activities are associated with master data. This means understanding the processes associated with creating new customers, for example, and understanding what operational and analytical applications, collaboration tools and documents are used in such processes with the objective being to identify all core business documents, process activities (tasks), applications and application functions, application screens and online forms, reports, data integration jobs, etc. that currently access and maintain disparate master data and also to identify all data stores holding disparate master data (e.g., files, RDBMSs, cubes, content management systems).
Figure 2 shows an example order entry, fulfilment and tracking process with activities highlighted that are associated with master data.
Figure 2: Order Entry Example
In order to do this well, a number of key questions need to be asked for each master data entity when following a business process. These include:
The objective of this kind of questioning is to create an inventory of things that may need to be changed when implementing an MDM system. This inventory could include documents, processes, application functions, screens, paper and online forms, data models, etc. Once this has been done you should be able to identify overlaps, duplication and conflicts within this inventory and zoom in on who performs duplicate activities. The objective is to identify duplicate, overlapping and conflicting:
Having got this far, the next stage is to create a matrix to track conflicts, duplication and overlap so as to identify what must be changed in order to eliminate these. Then you can determine the order of priority in which changes need to occur. Factors influencing the order of priority on MDM change management include return on investment brought about by the change, change complexity, cost versus budget available, human resources required and their availability and time to implement.
At this point, you should be in a position of control over change management for MDM because you know what has to change and in what order. You may find that the “to do” list you have created is somewhat daunting. However, at least you know what it is and that you can nevertheless make it doable. Transitioning systems to leverage centralised common master data may involve changing line of business SOE applications to remove/disable master data attributes from data entry screens and online forms. Once this is done, new forms and screens will need to be created to directly maintain the master data in central MDM data hubs. This latter capability may be available out-of-the-box as some MDM systems ship with pre-built standard portlets that can run in any enterprise portal server to allow users to maintain master data directly from an enterprise portal user interface. We will talk more about this later in this article.
The big change is the move from many systems of entry to one SOE. In order to make this happen, you have to understand the impact of that kind of change on existing line of business applications. For example, line of business SOE systems may update more than master data in a specific transaction scope. Therefore, transaction logic in the line of business application may also have to change to remove master data changes from transactions. In addition, duplicate or overlapping functionality in existing systems of entry has to be identified and removed/disabled to stop master data conflicts. There is no doubt that major transition begins once an MDM system shifts from being a system of record to a system of entry (see Figure 3).
Figure 3: Transition to System of Entry
When this shift happens, you need to already have a thorough understanding of the impact an enterprise MDM system will have. It may be the case that changes may need to be made to:
In Part 2 of this article, I will look at the potential change impact on each of these in more detail.
Recent articles by Mike Ferguson