Reference Data, I’ve been involved in multiple Data Governance and Master Data Management initiatives across different industry verticals. A lot of customers (and to some extent a lot of MDM applications) place a lot of importance on business processes, workflows, data stewardship and business rules. Business processes, workflows and data stewardship all suffer from one basic problem which can lead to bad data in the long run – human intervention. Human intervention, although important to create and manage a masterful MDM solution also leads to exceptions being made on an on-need basis. System rules on the other hand do not suffer from such gray areas.
Other than business rules there is one more potent tool to ensure the data quality in your organization – The reference data. This Reference data ensures that the subsequently created master data always conforms to standards and there are minimal chances of exceptions which ultimately lead to data quality issues. Even though reference data is an important piece of the data governance puzzle it does not get the attention it deserves.
Maintenance of reference data is usually left to simple lookups and data sources. Data cleansing always addresses the cleansing of the master data at the entity level but not enough time and effort is dedicated to cleansing of reference data. This usually leads to a situation of bad data quality building up over time as the reference data used to create the master entities itself is incorrect.
The 4 types of Master Data Entities currently are People (Customer, Vendor etc.), Thing (Product, Services etc.), Concept (Recipe etc.) and Location (Warehouse, Well Master etc.). We would argue there is a 5th type of Master Data itself which is Reference Data. This kind of data is as important if not more so to get its own category. What qualifies as reference data is dependent on the business use case of the organization.
So what are the potential points to consider for this perennially underprivileged area of data?
- Understand the different reference data that entity master data is going to use.
- Ear mark designated time and effort for cleansing of this reference data even before the data cleansing of master data. In fact the cleansed reference data will provide great insight into invalid data at the entity level.
- Do not underestimate the maintenance of reference data as an out of the system business process. Create a viable workflow for maintenance of reference data just like one would for master data.
Riversand understands this potentially understated need and has specifically provides functionality in the MDMCenter application to maintain the wealth of reference data.
- Treating simple reference data as simple lookup tables in the system.
- Treating important reference data as entities with associated workflows and approvals to create and maintain reference data.
- E.g. instead of modelling or treating Product Brand data as just a reference lookup data model it as an entity which allows for workflows on addition of new brands, maintenance of brand name or any other attributes associated with the brand.
- Treating reference data as a Glossary of Terms, with each term linked to an attribute at master data entity level.
- Treating reference data using this methodology allows for security privileges both at the reference data level and the valid values within a reference data set level.