Master Data Management (MDM) is a data governance process in which business units and IT departments collaborate, cleanse, enrich, publish and protect common information entities that must be shared across an enterprise. More broadly it is a cross-organization program consisting of People, Process and Technology. The purpose of MDM is to make sure that the data is uniform across the various disparate systems in an organization. We will see the challenges when using multiple systems and the benefits of using Master Data Management, how it helps in unifying the data present in disparate systems.

Challenges from Disparate Data:
Several problems are encountered due to data being present in disparate systems. This is the motivation behind using a common system called the MDM. The following IT and Business problems are usually faced when the data lies in multiple disparate systems:

IT Problems:
Compliance – Compliance issues arising from many single versions of the truth

Data accuracy – Data accuracy decreases when owners lose control to access the data

Data quality – Difficult to control data quality when we have many SaaS applications

Integration – Integration of public clouds, private clouds, and legacy systems

Revenue recognition – More difficult to realize revenue

Business Problems:
Sales Operations – Increased difficulty in booking orders, fulfillment, and payment

Account Receivable – Poor data quality often leads to problems in collecting/realizing payment

Account Payable – Entitlements are not clear with many single versions of the truth

Customer Care and Support – Entities are not clearly represented

Finance – Revenue recognition is impacted

Benefits of an MDM System:

As the number of endpoints increases, the complexity increases. MDM is a centralized data hub which reduces the cost of integrating new systems as the number of integrations are less when using multiple endpoints. This is because all the integrations go via MDM as it eliminates direct point to point connections. The data is stored as a centralized MDM database as a golden record, which shall be sent to output channels for updating the output systems.

  • Reduce the cost of integrating new systems
  • Turns an exponential problem into a linear one
  • Central “copy” of data allows for removal of the “data bounce” effect

MDM Terminology:

Domain – A domain is the data used to track a particular “noun” of a company’s business — for example, contacts, employees, partners or vendors

Model – Representation of a domain which identifies the fields, match rules and optionally the data quality services steps it consists of

Repository – The MDM engine and physical database storage of golden records

Data Quality – Set of services that are used by the MDM repository to validate, cleanse and enrich data

Golden Record – The approved data or single source of truth for a specific instance of a domain. Golden records are normally a composite of information collected from multiple systems which have been run through data quality services

Source – A system that will contribute information and optionally receive information from the MDM system

Channel – An access point that publishes record with create, update and delete requests to a specific source

Operations in an MDM System:

The following are basic processes that need to be done in MDM applications:

Define: Model the master data domains defining validation, enrichment, and matching policies
Deploy: Deploy the models in MDM repository and identify which source systems interact
Synchronize: Design processes that enable data to flow between MDM and integrated source systems
Steward: Steward your data as it flows into MDM system. Resolve duplicates, fix data entry issues, identify and correct erroneous data

Source Entity Processing

We use Dell Boomi MDM to illustrate the data processing in the MDM as it is easier to understand the basic concepts. The current section explains the processing of source entity data and how the records are either moved to Quarantine or to the outbound delivery channels. This is the process explained for a source entity and the same has to be applied for all of the source entities.

Whenever the data arrives from the source system, the following operations are done

Validation – The data is validated by checking the structure and schema of the input XML
Enrichment – Here business rules validations for the different fields are done. Data Quality check is done using either respond to a request from an external data service, a result of a process or custom business rule
Incorporation – After the above steps, data is matched against the master database and it is updated or inserted into the master database by automatic merging. In case of duplicate records, the records are quarantined.
In the below diagram, the inbound processing of the record is shown as a flow for easier understanding.

The data record arrives from the source application and the validation, data quality, match rules, approval steps are done in sequence.
Depending on the match rules and approval, the data is stored as the golden record or rejected and sent to Quarantine.
Quarantine records are later sent to source application after rejection so that the source application can look into the exact data issue and handle/resend the data.

Once the data is stored as a golden record, they are retrieved and sent to the output applications via outbound channels. For instance, there are 2 outbound channels for the App1 and App2. The outbound channel can perform either create or update based on the specific application.

To summarize, having an MDM system in an enterprise provides the following advantages:

MDM maintains data consistency across systems. It also eliminates duplicate or incorrect data. This is especially applicable in scenarios when the organization grows through mergers and acquisitions.
Provides options for data viewing, manual approval or rejection (quarantine) by the business user
Automatic data validation and merging can be done using stored business rules
Options to use external service for data validation. Eg. Dun and Bradstreet for identity matching

Having an MDM with consistent data increases customer satisfaction, operational efficiency, decision support, and regulatory compliance. Enterprises today must have an MDM system in their business roadmap so as to leverage the mentioned advantages.