Skip to main content

Master data

Data in any organisation is shared between many different functions and operations.

Each of these functions and operations will use this source data in a different way to support processes, systems and outputs. They will all have different quality expectations and may move the data into departmental specific systems.This has a significant and largely detrimental effect on that data quality in terms of its use across the whole organisation.

Reference and master data management attempts to mitigate this issue by the reconciliation and maintenance of reference and master data. Reference data deals with control and publishing of standardised terms, values, business definitions, relationships; it is the ‘master copy’ of definitions for classifying all data items.

Master data management is the control over the values of those entities to protect data quality and create a single version of trust. Trust is a better word than truth as context and use will affect how a version of truth is established. Trust means simply the data under consideration is considered to be of appropriate quality and the lineage of any transformation is understood.

Reference and master data can be thought of as providing consistent context for transactional data. Clearly there is a very strong link to data quality management as the measure of successful reference and master data management is in the quality of the data it acts upon.

MDM is another component of data management that is best introduced iteratively. Many data management problems can be identified by a shared understanding of ‘who needs the data’, ‘where is it held’, ‘how different is it in different sources’, ‘how do we manage inconsistencies’ and ‘who has the most valid values’.

MDM is the cornerstone of solving this type of problem. By agreeing the business rules that define and scope the data in question an agreed set of values and quality metrics can be drawn up. Reference data can be external (e.g. postcodes) or internal (e.g course-identifier).  Master data provides the deeper context for the transactions which these entities are acted upon.

MDM is a large and complex subject. It should be guided by the principle that master and reference data are shared between the whole organisation, not individual departments or functions. From a data governance perspective it is the role of the data steward to take responsibility for controlling and co-coordinating reference data values.

A ‘single version of trust’ is a realistic goal for any organisation. While it is not an easy one, it has the huge potential to stop discussions around the efficacy of data and start those on what the data actually means. Creating this trust would – for example – stop once and for all an argument about ‘how many students do we have in our institution today?’