‘Master Data’ is not as easy to define as other data sciences. Relational theory is mathematical in nature. Dimensional design is based on ease-of-use and taking advantage of system architectures. Master Data doesn’t have the same foundation.
There are attempts to define a philosophy floating around the internet. Most of them identify master data as ‘critical business nouns’ – people, places, products, etc – and some ideas go deeper, defining all the possible values of an attribute as ‘reference data’ or ‘reference master data’ or any other absurd name. (meta-reference mastered data nouns – Ha! Beat that one)
None of these philosophies are perfect because the idea of ‘master data’ comes from the single, urgent need of thousands of organizations across the globe:
Data they can trust.
Data that is always up-to-date.
Data that never has human entry mistakes.
Data that is retired when it becomes obsolete.
So I would like to propose my own definitions:
“Master Data is data that must be complete, accurate, and reliable.”
“A Master Data Management tool ensures completeness, accuracy, and reliability of a data set.”
Different tools serve this function in different ways. Some focus on one specific topic, like people or products. Some can bounce addresses against Google or Bing to validate them. Some focus on de-duplication and fuzzy matching, or controlling data entry and modification. Some do all of the above and more.
So if you’re tasked with choosing an MDM product, or find yourself in charge of one, don’t fret about learning what master data is. It’s about what your organization needs it to be. What do they need to master. How does the tool in question master it. Definitely keep philosophy out of business meetings about MDM. If someone asks ‘what is master data?’ a heavy, complex answer can cause the whole thing to flop. Stick with my simple definition above and focus on what processes will add, edit, delete, and consume mastered data.
On a side note, many of the tools I’ve looked at are… greedy with their data. They’re often difficult to integrate and focused on BI use. But mastered data is something that any system can benefit from. So keep an eye out for how easily it integrates. If they support web services, have the salesperson set up a technical demo with SoapUI. Have them do a POC that uses SSIS, or DataStage, or whatever you want.