How to get high quality master data


How to get high quality master data

A paper by Cliff Longman published December 2009

The idea of “Six degrees of separation” is that any human being can be connected to every other human being through six relationships at most. Data is a little like this: through data relationships, all data is connected.  When it comes to managing shared data (the primary focus of master data), you should plan to include all data types within your scope because, ultimately, it is all connected. However, placing all shared data under management is a significant task involving many parts of the business, changes to people’s ways of working, new processes and potentially risky technology implementations.  Not an exercise that can be undertaken in a single step, it needs to be done in stages.  At present many companies are conducting pilot MDM projects without giving sufficient thought to the endgame: how to control all their master data. Cliff Longman has identified four different “strata” of data: transactional data (e.g. a payment is made), dependent data (e.g. sales regions), independent data (e.g. customer, product) and policy and reference data (e.g. categories of products, currencies) which have different characteristics.  There are then four alternative strategies which ultimately end up with all shared data under management through a system of data mastering.

  • One data domain at a time
  • One business intelligence need at a time
  • One category of data at a time
  • One business process at a time

Far from being just one approach to an MDM program, Longman believes there are these four main viable approaches with very different cost, risk, and benefit profiles.  Each shares a common end-goal: that of a managed master for all data types across an organization, but MDM thought leaders should select a strategy appropriate for their organization to get there. The full white paper, which discusses these approaches in detail, is available for purchase below.

The white paper has 15 pages.