Master Data Management

Data warehousing systems are inseparably connected with another question - Master Data Management (MDM). To begin with Master Data itself - by definition, it's stated to be a single source of main business data, which is used across all systems, processes, and applications. If extend a definition to a whole enterprise scale, Master Data still refers to the single source of main data, but its range includes all departments and systems across the whole enterprise. Then, what's the point of Master Data Management?
MDM itself might be boiled down to a range of processes and tools responsible for managing and defining basic data of a company. According to that, "basic" means usually general non-transactional data. In brief, the point of Master Data Management is collecting, aggregating, transforming, and connecting data, simultaneously caring of its quality and unity.

Thank to MDM, all departments across the enterprise get an access to the same, timely data and have a possibility to make the most efficient decisions.

What does it mean in practice? An example should resolve the doubts.
Think of a car dealer. While buying a new car, name of a customer is put into a database. Then, after a few days, a client - happy with his newly bought car - receives a packet of advertisements, oriented on convincing a client to buy a new car. Seem pointless? Indeed. But it happens as an effect of disarranged data systems. Information about customer put into a database, but not categorized.

Master Data Management is important especially for enterprises following expansive business scenario. Acquisitions and mergers lead to duplicating data systems and - as a consequence - lacks of data integrity and unity. Every enterprise have its own DW systems, and - when it comes to a merger - Master Data from each of them gets duplicated. It's pointless to remind what consequences might be caused by data multiplicity. Theoretically, all the doubled data should be immediately highlighted and removed, but the real-life experience shows that hardly happens.

Usually the data model for the MDM servers can be grouped into three domains: Party, Product and Account.

Integrating only two systems demands initiating complex reconciliation processes. But it's still the easiest case. Usually, any acquisition brings up at least a few diverse systems, sometimes even more than ten. How to integrate them? It's not an easy thing to do, but well-prepared Master Data Management systems reduce more than a half of work, time, and costs.

While MDM applies to reconciling even the most differing systems, its efficiency demands involving diverse processes. Firstly, identification of a data source. Following processes apply to data - its collecting, transforming, and normalizing. Then, it comes to errors highlighted before - their accurate detection, and - obviously - correction. Finally, prepared data might be consolidated, stored, and distributed dependently on specified needs.

Even though the efficiency and usability of Master Data Management solutions are irrefutable, MDM systems aren't as popular as they deserve. What is the reason of this situation? Mainly, costs of implementing and maintenance of those solutions by the world-class vendors. Finally, low rate of investments return doesn't ease the business justification of MDM implementations.