Data migration and master data management
What are the differences?
Data migration and master data management are topics that have led to discussions on many of our projects. By clarifying these terms, this blog aims at aiding a common understanding and collaboration on projects right from the start.
Wing Lee & Burak Alper - 9 March 2017
Data migration (DM) and master data management (MDM) are topics that have led to discussions on many of our projects. Some use these terms as synonyms for the same domain, while others see them as completely separate fields. This especially adds complexity when projects are being set up, as the project scope, objective and skill set needed differ. This blog aims at providing definitions for both terms, explaining the differences, but also how they can complement each other. We have been applying these definitions and descriptions in our projects and we notice that it enhances the common understanding and the collaboration right from the start.
Definitions for DM and MDM
Data migration is a process of extracting data from system A (the source) and loading it into system B (the target), typically with some transformations in between to close any gaps between the two systems.
Master data management is the process of creating a single source for business critical data, which can be used throughout the entire organization for multiple purposes. Examples of master data can be objects such as Supplier, Product or Customer data.
The differences between DM and MDM
As the definitions above explain, data migration is focused on getting data migrated from system A to system B, while master data management prescribes how master data in system A and/or system B is best managed.
Main focus of data migration is on delivering the optimal technical migration solution that ensures data is migrated correctly in the target system, while master data management aims at defining the governance structure and data management procedures and processes to create, maintain and retire master data throughout its lifecycle. Ultimately, establishing a master data management organization, which could be a formal department or a virtual group.
Data migration is often a one-off process that ends after all data has been migrated into the target system. Master data management, on the contrary, is a continuous (business) operation with the objective to keep the master data details of suppliers, customers, items and the like accurate, consistent and without duplication.
When DM meets MDM
Despite the differences, data migration and master data management have a common interest: ensuring high data quality. As stated earlier, keeping master data accurate, consistent and without duplication is the objective of master data management. On the other hand, good quality data ensures migration readiness and minimizes fall-outs during the automated migration due to data issues. Because of this common interest, combining the two disciplines in practice can yield substantial business value.
Data migration is a golden opportunity to clean data and improve its quality. In fact, data cleaning is often a requirement to make a successful data migration possible. In parallel to this data cleaning activity, it is recommended to implement master data management principles to ensure the good quality of the (master) data achieved will be maintained after the migration. Appointed data owners/ stewards can be assigned to the data cleaning activity and be (and feel) responsible for the data in the target system from the very beginning.
Sophisticated master data management solutions often entail the use of data hubs. These hubs manage and contain the master data and provide a single source of truth to the business. It also makes integration with other applications considerably easier once implemented. When implementing such a solution, in addition to establishing an effective governance organization, data management procedures and processes, an initial load into the data hubs is required. The process of loading the data in the data hubs, including the data cleaning exercise, is basically the same as a data migration trajectory.
What if one is done without the other?
Although there is significant business value to achieve when combining both principles, in practice these are approached separately. The key here is to have a good overview of the risks of choosing one or the other without combining them.
The effects of a solid data migration implementation can be a one-off benefit that declines after it has taken place. Take for instance a company that goes from system A to B, spending time and effort on their DM approach. This will generally result in a smooth process during and immediately after go-live of system B. However in the long turn, the data in system B will lose its quality without a proper MDM strategy and approach in place.
On the other hand as explained earlier, when implementing a sophisticated MDM solution, there is generally a need for an initial data load into the data hubs. This initial load is in fact a data migration and should be handled as one. If the data migration principles and best practices are not followed properly, it could lead to incorrect/low quality data being migrated. Ultimately, although a good MDM strategy might be in place, but since the data is of poor quality the MDM solution will not yield the intended benefit. Moreover, heavy work is needed to ultimately have this corrected in the data hub.
Data migration and master data management are closely linked but different disciplines. Despite the differences, both have the interest in ensuring high data quality. Significant long term business value can be achieved when combining both principles in implementation trajectories.
More information about data migration or master data management?
If you have additional insights to this discussion, we are happy to learn about them. In case you would like to know more about data migration or master data management, please feel free to contact Wing Lee via +31882882902 & Burak Alper via +31882888752
Read more about data migration
Read the other blogs of Wing Lee about data migration: