What is Master Data Management?

Master data management (MDM) is a process-based approach to data integration. MDM's goal is to provide organizations with a golden record of their most essential business data, such as customer and product information. 

This golden record is the SVOT, or Single Version of Truth. Data can exist in multiple places across an enterprise network – for example, a CRM and a billing system might hold different records about the same customer. If there is a discrepancy between these records, the organization can refer back to the SVOT for the definitive version. The MDM process ensures that the SVOT is always accurate and recent. 

What Data is included in Master Data Management? 

Master data describes the entities that are critical for business operations. Essentially, this means that the master data describe things, not transactions. Master data typically has four distinct domains: 

  • People: Everyone organization holds data about three main types of people: customers, suppliers, and employees. Salespeople are often categorized as a separate data entity, as they are employees, but they have relationships with customers. 
  • Products: For organizations that sell physical goods, this can include product name, unique identifier, price, and location. For other businesses, this data domain may describe services, and these could consist of details like include hourly rate and billing.
  • Locations: Generally, this refers to the physical locations of the business properties, such as stores, warehouses, and offices. Customer and supplier location data falls within the People domain.
  • Other: Master data may need to describe other business entities, such as financial ledgers, licenses, or digital assets. 

The difference between master data and transactional data is like the difference between verbs and nouns in language. Consider a sentence that describes a business activity, like, "Carolyn sold a widget which Mr. Wong bought and our Tulsa branch shipped." The verbs soldbought, and shipped all refer to transactions, and are therefore not described by master data. A transactional database records these actions.

The nouns – CarolynwidgetMr. Wong, and Tulsa – are all discrete business entities. Master data would describe these entities. 

What is the Process of Master Data Management? 

Master data management requires the right combination of people, processes, and tools. 

1. Define and Categorize Sources

Master data comes from disparate sources. A typical enterprise will have several platforms with production databases, such as their CRM or ERP. The enterprise might also have data repositories, such as a data warehouse that stores historical information. 

The first step in MDM is to assess each of these sources, evaluate their contents, and decide whether the database is a suitable source of master data. Other important factors include the reliability of the source, the recency of the data, whether the values clash with other sources, and the structure of the data. 

2. Establish MDM Business Rules

Master data management is one of the core pillars of good data governance. As with all aspects of data governance, the MDM process requires codification by a series of internal business rules. These rules must answer questions like:

  • Which sources are valid?
  • How often are MDM records updated?
  • How will the business work with SVOT?
  • How is data quality assessed?
  • Who is responsible for updating and publishing MDM rules?

These rules need to cover most possible scenarios that might arise when working with master data. They must also be easy to understand and easy to implement. This will allow all staff to follow the rules, not just those staff who understand database technology.   

3. Appoint Roles

When organizations implement an MDM process, they generally appoint three people to oversee it. 

  • Data owner: The data owner has the ultimate responsibility for master data. They oversee the business rules and ensure that everyone is following them. They will also ensure that the master data process meets all compliance standards and fits the organization's data governance strategy. The data owner will also be responsible for any funding and resource allocation related to the process implementation. 
  • Data steward: Stewards are responsible for the data itself. They oversee quality projects to ensure that is master data is a functioning Single Version of Truth. They can respond to any errors, concerns, or handle any changes to data sources that may impact data quality. 
  • Data custodian: Custodians are IT team liaisons who work with stewards to ensure master data quality. A custodian will focus on the tech elements of the process, such as the ETL data pipeline or other master data solution. They will help implement rules, deal with any system failures, and ensure users have appropriate master data access. 
  • Quality auditor: Auditors provide an independent analysis of the master data's quality and reliability. Their work may involve comparing master data to source data or analyzing relevant business outcomes. If they find an issue, they will raise it with the data owner and notify the data governance team. 

 4. Implement ETL Workflows 

On a tech level, MDM is very similar to the ETL process. Some organizations may use ETL to implement this process, while others use a dedicated MDM solution. 

In all instances, the process follows these steps:

  • Extract: An automated process pulls data from the relevant sources. 
  • Transform: The ETL process transforms the incoming data to fit the master data schema. ETL cleanses and validates data before moving to the final stage. 
  • Load: The new data integrates with existing data in the master data repository. If there is a conflict, then the MDM business rules will tell the process which version to discard. 

Master data is not as volatile as other forms of data. However, it can change frequently, so regular updates are essential.  

5. Align Business with SVOT

When the business has a Single Version of Truth, all other processes must work with this golden version. For automated processes, this can mean a reconfiguration. For example, a data pipeline can update the CRM with customer information from the master database. 

But people and processes also require alignment. For instance, some salespeople might use written records for their clients. They will have to adapt to the new process, which means entering the correct contact details in their local system. That system then pushes the updated contact details to the master database.

Share This Article
facebook linkedin twitter

Glossary of Terms

A guide to the nomenclature of data integration technology.