Master Data Management (MDM)
There are essentially five types of data in organisations:
- Unstructured - This data is found in emails, white papers, magazine articles, company intranet portals, product specifications, marketing materials and PDF files.
- Transactional - This data is related to sales, orders, invoices, receipts, claims, payroll, etc.
- Metadata - These are the ones about the data in the data warehouse. It may reside in the repository or in various other forms such as XML documents, report definitions, column descriptions in a log file.
- Hierarchical - Hierarchical data indicates the relationships between other data. It can be stored separately as descriptions of real-world relationships, such as company organisational structures or product lines.
To extract value from this data with analytics, MDM is a prerequisite.
Master data describes the business-oriented properties of the data objects that are used in the various applications across the organisation and their associated metadata, attributes, definitions, roles, connections and taxonomies.
Master data is the key to the business and generally falls into four categories:
People,
Things,
Places,
and concepts.
In addition, they are grouped according to domains, or types of entities. For example, within people, there are customers, employees and external staff.
Within things, there are products, shops, goods and assets.
Within concepts, there are elements such as contracts, warranty and licences.
Finally, within locations, there are offices, production sites, storage sites and geographical divisions.
Some of these areas can be subdivided. Customers can be further segmented by priority, credit rating. The product can be further segmented by category, sector and industry.
When several IT systems make up the information system, there are different definitions of the same object, creating inconsistent data and therefore inaccurate analyses.
To avoid this, MDM is a combination of applications and technologies that consolidate, cleanse, augment and synchronise the organisation's data with all applications, business processes and analytical tools.
In terms of data quality, it enables data integration and a single version of the truth. If the data does not accurately represent reality, then it is unusable.
MDM is not only a technological problem.
In many cases, fundamental changes to the organisation's business process will be required to maintain clean master data, and some of the most difficult MDM problems are more political than technical.
But this investment in MDM brings significant improvements in operational efficiency, reporting accuracy and strategic decision making.
"MDM is the latest attempt to solve the age-old problem of inconsistent versions of critical data at the centre of an organisation," said Andrew White, research vice president at the 2011 Gartner Summit.
An MDM project plan is tied to requirements, priorities, resource availability, schedule and problem size.
According to Roger Wolter, Microsoft Inc, most MDM projects include at least the following phases:
- Identify sources of master data.
- Determine the consumers of the master data.
- Collect and analyse your master data metadata.
- Appoint data managers.
- Implement a data governance programme and a data governance board.
- Develop the master data model.
- Choose a set of tools.
- Design the infrastructure.
- Generate and test master data.
- Modify production and consumption systems.
- Implement maintenance processes.