The new version of FIFA is coming on October 6th. Usually, EA Sports gives players a preview of the game…
How to set up an MDM in your company?Published on
With Master Data Management (MDM), we define the concepts and processes that govern master data within the company’s information system. This allows you to create, store, maintain, distribute and provide a complete, reliable and up-to-date view of all this data.
These activities are carried out in total independence from those usually carried out by a company. Nevertheless, it is important to use MDM well in order to reap all the benefits, which can be substantial.
Choose the right implementation style
An MDM solution brings many benefits to the company by encoding in one place the referential data of a system. These significant improvements compared to traditional systems require the adoption of good practices.
It is necessary to start by clearly defining the implementation styles. The “register” mode, which is rarely used, is largely non-intrusive and consists mainly of recording transactions and updating the repository.
In “consolidation” mode (this mode is the starting point for many MDM projects), the importance of data integration and quality functions prevails. The coexistence implementation style is intrusive, but not very structuring.
It is therefore the real-time integration and data deduplication functions that matter. The legacy system and the MDM software system must coexist in harmony. Finally, the so-called centralized style is relatively intrusive and quite structuring.
The importance of workflow and real-time integration functions is paramount. The activity of the system is as fluid as possible. These four implementation styles are each suitable for a given project.
Comply with certain good modelling practices
At the functional level, it is important to involve the various trades strongly in design. The golden record must be defined at the beginning of the modeling and must be the subject of a consensus. It is necessary to tackle the core of the master record, simplify it and then evolve by extending the processing. Readability for end users must be very high and the company glossary must be well documented. The design phases, including the creation of the keys, will be carried out with precision.
It is indeed more complex to have to make new changes after the event. Security data must be accessible, by means of roles and views. At the level of technical modelling, it is necessary to create a permanent internal key for master data: it is one of the keystones of the system. We therefore advise you to define these standards with high standards, and to stick to them. In addition, the same should be done for naming rules and only one modeling software should be used.
All templates, definitions, and rules are then reused. It will be necessary to anticipate the performance impacts on controls in the event of an improvement in propagation rules.
Take care of your data to have a quality MDM system
To achieve optimal use of an MDM system, great care must be taken to ensure data quality. Data profiling and data collection will greatly assist in good data parsing and validation.
It is essential to standardize these data so that they are representative for all MDM systems. De-duplication rules must be precisely established and refined as far as possible. Matching should be integrated into the whole process.
Once the quality of the data has been mastered, they will have to be integrated and then propagated. Choose batch, real-time, incremental, propagation or total loading processes as required. This is where you can integrate internal (ERP, CRM, HR, specific, etc.) and external (cloud and others) applications as well as big data.
Do not hesitate to organize data stewardship to make the master data operational. Data stewardship is to be organized by roles, it allows to manage the daily tasks of maintaining referential data.
You can edit master data, or define workflows for collaborative editions. Reference data can be activated, master data services are integrated in real time into processes, and the context is brought into enterprise, web and mobile applications as well as big data.
On the same subject