|
When it comes to MDM, talk is cheap. Almost every company we’ve been at, we’ve heard the acronym at the water cooler, in the hallways and in the boardrooms but we haven’t seen one project kick off with ribbons and champagne.
The reason why so many data architects and managers fail to convince business to fund a MDM project is because MDM is like the plumbing in a house. People usually don’t care about the plumbing in a house until it affects them in some way. When they shower and the hot water runs out, that’s the time they will think about the plumbing. The same goes for MDM - the business does not care about how the data is stored or managed until something happens and they are responsible. For example, answering to the CFO why two numbers are different when they represent the same thing such as revenue.
We sat down with the Forzani Group in August and again in December 2007, to discuss the possibility of implementing MDM there. They recognize the need for MDM because of the number of data silos they had and growing. With inventory spread across systems, it is hard to tell whether they are overstocked or backlog. We recommended to them, to produce a ‘holistic’ data model to represent their business on an enterprise level. Once their physical database is up and running, start by building adapters between the database and each data silo. These adapters would know how to match entity to entity from its domain values. They need to look at their reporting requirements and determine which data silo represents the greatest breadth in their business. We told them to begin with that and slowly merge and integrate the data into the enterprise data model.
Another company that plans to implement MDM is BMO. The sell was a statement that Henry Ng used, ‘Since data is an asset. It is only fitting to look at ROA instead of ROI”. That’s an interesting statement because data is only considered an asset if it represents information valuable to the company. If data exists in several places, it may carry different attributes making it difficult to report on. Data is not considered information until it has been cleansed and mastered. That’s the end effect of MDM.
|
