A question stumping many organizations — how does one embark on the complex journey of data governance?
While many different facets of data management could act as the starting point, this article puts forth an argument for Master Data Management (MDM) to be the foundation upon which an organization can build its initial data governance maturity. An MDM implementation can enable you to bring together a core set of data governance disciplines, apply them to a limited yet impactful scope, learn lessons and set standards, and then scale them out across the rest of the organization afterwards.
This point of view will do so by first examining the aspects of a typical MDM implementation, and by then mapping the MDM implementation components to a standard data management framework. A case study will be used to describe how a leading insurance company used MDM to kickstart their overall data governance journey.
Master Data Management (“MDM”) is a capability that ensures accurate and consistent data on key business entities across an organization’s systems, applications, and processes. MDM is crucial in providing a single, reliable source of truth for master data attributes such as customer, product, vendor, or raw materials data. (For a more detailed definition, see here.)
An MDM implementation project typically involves the following phases:
- Strategy, policies, and standards: The intended outcomes of the MDM solution and process are described, and the rules and expectations are articulated on how critical master data is to be governed.
- Discovery and Analysis: The current state of data within the organization is assessed and understood, and the master data entities that need to be managed are identified.
- Data Governance Framework: The data governance framework is defined, including the roles and responsibilities of data stewards, data owners, and data governance forums, as well as assigning these roles to “real” people and teams, and mobilizing the governance forums.
- Data Modeling and Mapping: The data model is created that defines the relationships between the master data entities and the attributes that describe them, and the master data is mapped to the systems and applications that consume it.
- Data Integration: The data integration is built from multiple sources into the MDM solution by creating connectors to extract data from the source systems, transforming it to conform to the data model, and loading it into the MDM platform. Similarly, feeds are established out of the MDM solution for consuming processes and use cases.
- Data Quality Management: Defining and executing data quality rules, calculating metrics, establishing reporting, and identifying the process for resolving data quality issues. This may also include entity resolution and survivorship analysis.
- Training and Change Management: Training is provided to stakeholders to ensure that they understand the MDM platform and their roles and responsibilities in the data governance framework. A change management process is established to manage any changes to the platform or framework.
- Monitoring and Maintenance: In this phase, a process for continuous improvement is established, including data quality monitoring and reporting, data governance audits, and identifying opportunities for improvement. An issue management process can be followed to log and resolve issues as they are observed.
Of course, implementing an MDM solution and operationalizing the processes around it will drive maturity in the dimension of Reference and Master Data. However, throughout the above-described 8 steps of a typical MDM implementation program, there are connections with the wider data management framework as well.
Let’s describe these connections explicitly to identify exactly how maturity can be built up in these capability areas.
- An MDM solution will not drive value if it does not come accompanied by a wider strategy that explains why mastering certain critical data is paramount. If a data strategy is already in place, this can be used to reinforce it, but if it not, it can actually serve as the organization’s first articulated data strategy.
- For the specific platform implementation, use cases and consuming processes should be identified ahead of the program to ensure that corresponding requirements are incorporated. This can turn the MDM solution into a true data product, and enable the data office to start managing a portfolio of data products and tracking data-driven use cases (and value creation!).
- Around the MDM solution and the mastered data, foundational governance is established, including through defining a set of policies and standards that outline the expectations around MDM in general, as well as specific roles and responsibilities. These can enhance existing data policies and standards, or if they don’t exist yet, form the first version.
- A data or domain owner is assigned that has accountability for the mastered data. Data stewards could be identified that have been given the responsibility (as delegated by data owners) to look after the data. This includes responsibilities to ensure fitness-for-purpose of the data, and can enable the creation of a wider data governance RACI-matrix that can be expanded to non-MDM areas later on. Hence, data governance roles can be introduced and defined for master data, enabling a culture of data ownership and accountability, which can then be scaled across the organization, ensuring consistency in data governance practices.
- Metrics are defined to track the health and usage of the platform, and to track risks. These metrics can form the first set of data governance metrics that can be expanded later on. If there is none already, a forum can be convened to monitor these metrics and resolve issues. The forum can be expanded or serve as inspiration for new forums to cover other data governance topics later on, driving critical data maturity in governance practices in general.
- For the MDM platform itself, a solution and data architecture is created. This can be informed by a set of explicitly identified architectural standards. The best practice can be established that any solution architecture should be informed by applicable data standards. The actual standards themselves are also reusable for other, future transformation initiatives.
- MDM only drives value if the platform becomes a trusted source that is used across the organization’s transactions. This should not be optional — if there is mastered data, then its consumption should be mandatory for anyone consuming this data. This helps to establish a wider reference data architecture, where guidelines, principles, best practices, and implementation patterns can drive the robustness and consistency of architectural design across the organization, as well as the rationalization of vendors and technologies.
- For the MDM solution, it is captured from what sources data is ingested and how the mastered data is made available for downstream consumption. The documented data flows and lineage can form the starting point for capturing the wider organization’s data landscape.
- A critical component of MDM is to measure and manage the quality of the mastered data. Without controlled (and quantified) quality, there can be no trust, without which adoption will falter. The process built around the identification of business rules, creation of executable data quality rules, and compiling of data quality metrics is very reusable for any future data quality efforts.
- If a specific tool is used for the measurement of data quality, it can be deployed for additional scope later on. Similarly, a dashboard that summarizes data quality for the mastered data can be easily expanded to include data quality metrics for non-MDM data.
- Ideally, the concept of “data quality by design” is embedded into the architectural fibers of the MDM solution. That is, data capture, integration, and transformation are processed in such as way, that data quality is — to the extent possible — guaranteed. For example, for country fields, only options are allowed that appear in the respective ISO-standards. This establishes a pattern, best practice, and muscle memory that is relevant as well for any future modernization and transformation efforts.
- An issue management process is created to ensure that issues in the mastered data are resolved. The corresponding process for identifying, tracking, and resolving data issues can then be scaled across other data domains.
Modeling & Design and Metadata
- As mentioned above, a data model is required for any MDM solution, to define the relationships between the master data entities and the attributes that describe them, and mapping the master data to the systems and applications that consume it. The conceptual and logical model (and the underlying metamodel) that are created for the master data (e.g., customer, product) can be expanded to include other domains (e.g., finance, HR, supply chain, transactions).
- As part of the MDM, you can adopt a technology to design and capture the data models and metadata. The technology as well as the process to use it, can be used for data models for subsequent solutions and domains.
- The more data models and metadata are captured, the less the incremental effort will be for future scopes. For example, if data elements are to be described and metadata captured for a customer onboarding process, it will take a lot less time if key customer data and related business requirements have already been defined at a conceptual and logical level.
Data Storage & Operations
- During an MDM program, there’s a critical need to consider the storage and operational requirements of master data. This need drives the implementation of data storage best practices and the development of standards for data operations. As part of this, organizations would design and implement an optimal data storage strategy, taking into consideration factors like data volume, velocity, and variety, as well as the need for scalability and resilience.
- The operational processes developed for MDM, including data backup, archival, and recovery processes, can become the basis for similar processes for other data management areas, thus paving the way for building maturity in data storage and operations.
Integration & Interoperability
- MDM, by nature, requires the integration of data from diverse sources, thus fostering a strong foundation in integration and interoperability. The procedures set up for the consolidation, deduplication, and synchronization of master data across different systems can act as a blueprint for future data integration initiatives.
- Establishing data standards, creating common data models, and enforcing consistent data formats during the MDM implementation can set the groundwork for increased interoperability among systems. The integration patterns and practices built as part of the MDM process can be reused across the organization, helping to establish mature integration and interoperability practices.
- Specifically, there is an opportunity to define a set of interoperability standards — for example in terms of defining data using a common data model and using a selected set of API and ETL technology — that can be adopted more widely throughout the organization.
Warehousing & Business Intelligence (BI)
- A well-implemented MDM program ensures that data fed into a data warehouse is clean, consistent, and reliable. It provides the single source of truth needed for effective reporting and analysis, thus laying the foundation for a solid BI strategy.
- The data quality rules, metrics, and processes established during the MDM process can be used to maintain the integrity of the data warehouse. Moreover, the master data’s robust metadata management can drive efficient data lineage, impact analysis, and improved understanding of data for BI purposes.
- When implementing an MDM program, organizations establish and enforce specific security policies for data access, modification, and deletion. These policies are applied uniformly across the organization, enabling tight control over who can access or manipulate the master data.
- This offers an opportunity to lay the foundation for a more comprehensive data security framework. While setting up the solution, organizations can define security roles and responsibilities and develop protocols for encryption, anonymization, and pseudonymization of sensitive data, thus building the initial blocks for broader data security maturity.
As a practical illustration of how MDM can drive broader maturity, let’s consider a case study involving a leading insurance company that I had the opportunity to work with. This company, one of the top three in terms of market share in its primary market, embarked on a project to implement an MDM solution for its customer data.
As part of this initiative, my role was to ensure that the technology solution, which was being handled by another vendor, was not only implemented but truly operationalized within the enterprise. This involved establishing the right processes, documenting these processes, creating a robust data model, developing a data quality solution and dashboard, as well as a metadata model and basic metadata management process. We also identified and established crucial roles such as a data owner, data stewards, and data quality analyst.
In terms of practical application, the main targeted consumer of the master data in this case was the company’s marketing department, with use cases including customer segmentation and marketing campaign management. Prior to the MDM implementation, the company struggled to achieve even a basic 360-degree view of its customers due to difficulties in reconciling customer data from various sources.
A year after the implementation, the successful business impact was clear. With a consolidated and reliable view of customer data, the marketing team was able to discern whether a customer engaged with one business line also had products from another. This facilitated a deeper, data-driven approach to customer segmentation, replacing the previous rule-based analysis. The newly accessible data enabled predictive modeling for determining the “next best action” and allowed marketing campaigns to be activated on a much more personal level.
An unexpected yet significant benefit was the improved management of contact data, such as email addresses, phone numbers, and physical addresses. With updates from various input channels (like call centers, physical branches, or customer emails) integrated into the MDM solution, all consumers of this information had access to the most current details. As a result, the success rate of reaching the intended recipient with written marketing materials or via phone increased by >20%. According to the team’s estimates, the return on their investment in marketing campaigns increased by as much as 50%.
MDM can serve as a powerful launchpad for an organization’s data governance journey. It offers a comprehensive approach to establish a reliable source of truth for key business entities, build a strong data governance framework, and drive maturity across various data management capabilities. And by focusing on MDM for a given domain as a manageable scope, organizations can introduce key data governance concepts, test them in a controlled environment, and gradually scale them across the organization.
All images unless otherwise noted are by the author.