MDM may change your life…or not
Breakthrough technology or just another acronym?
Comment According to (the normally more readable) Wikipedia, Master Data Management (MDM) “focuses on the management of reference or master data that is shared by several disparate IT systems and groups. MDM is required to warrant consistent computing between diverse system architectures and business functions”.
Great. An example may help. Most large organisations have multiple database systems, several of which may hold customer data (Sales, CRM, Marketing etc). It is highly unlikely that the customer data held in these systems is consistent. MDM is about taking data from these systems, cleaning and de-duplicating it until there is one definitive, accurate, standardised customer list. Of course, MDM is not just limited to customer information, it can be applied to any data within an organisation.
But, wait a minute. Doesn’t this all sound horribly familiar? Does the term data warehousing spring to mind? A data warehouse is a central repository of data taken from a range of disparate source systems across an organisation. Warehouse data is cleaned up into a consistent form that accurately reflects its meaning within the organisation.
So both data warehousing and MDM are processes concerned with ensuring that the data in the organisation is consistent and both extract and transform the data from multiple source systems and load it into a single location. Where they differ is what happens to the data next.
MDM is based on the premise that we write the clean, standardised data back into the source systems from whence it came. The huge advantage is that the data in those systems is now consistent (because it comes from a master data source) and so the analysis performed on the data in those systems will yield consistent results across the organisation.
In a warehouse, the clean data in the warehouse is the basis for further business analysis: it is not transferred back into the sources. The advantage here is that, within the data warehouse, it is possible to cross-reference data from the different sources. So we could, for example, plot sales (takes from the Sales system) against marketing spend (from the marketing database).
In addition, it is sadly lamentable, but nevertheless true, that many of the existing database systems have truly appalling analytical capabilities – so although MDM provides better data in the source systems, those systems may not have good analytical capabilities. Another advantage of a data warehouse is that we can analyse the data using modern, shiny, effective analytical tools and processes – such as OLAP and data mining.
Now, of course, MDM pundits can argue that the same could be done with MDM data before it is moved back to the source databases. Perfectly true. But, then, hasn’t the MDM system now become a data warehouse? And wait a minute, the only major distinguisher between MDM and data warehouse was that MDM allowed us to move the data back in the source systems if required, which has always been possible from a data warehouse: I know, I’ve done it on several occasions. So, either I have been way ahead of the game for years and have been using MDM before it was even an acronymical twinkle in its designer’s eye – or MDM is simply a slightly different flavour of data warehousing.
Pedantically of course, MDM isn’t an acronym because it is pronounced as three letters not as a single word. Perhaps its descriptor should be changed to MaDMan. ®