September 24, 2012 | By Rey Villar
My last blog detailed measuring DG program success using Data Governance Metrics. This blog addresses that timeless chicken and egg question – which comes first – MDM or data governance? Let’s also explore some of the difficulties implementing MDM solutions, and I’ll provide some insights from my recent experience implementing multi-domain MDM for a healthcare client.
Can MDM be implemented successfully without robust data governance in place? On the other hand, isn’t it true that data governance policies and procedures support MDM by focusing on master data, reference data, and data quality? Like yin and yang, I would argue that one cannot exist without the other. Organizations that try to build MDM without data governance, and vice versa, have all-too-often failed to reap the benefits of either.
Gartner and the MDM Institute have projected that thirty percent of companies implementing MDM programs will run in to serious challenges by 2013. Why? What kind of challenges are there, and what kind of action is required for success?
The MDM Institute defines Data Governance as the processes, policies, standards, organization, and technologies required managing and ensuring decision rights, ownership and accountability of data in an organization.
Master data is the information that describes core business entities such as customers, products, locations, suppliers… and so forth. Master data typically is non-transactional data, shared by several applications, static in nature.
Master data management is the required organizations, processes and tools to ensure that every master data element …
- Is captured accurately & consistently thus enabling master data quality
- Is stored in a way that guarantees integrity and a single place of reference
- Is made available to those who need it, whenever they need it, both internally and externally.
Why are MDM & DG PROJECTS so Challenging?
MDM projects require a paradigm change in business process. Creating a single version of truth for the enterprise is not easy. Many organizations take the approach to look to their analytics platform as the source of master records for MDM. In fact, the single version of truth has to be solved where data is created, not where it is consumed. The analytics platform should not be the source for MDM. It can’t be treated as IT project with bunch of business rules to match and merge, apply trust score and create a golden record. This lack of ability to change thought processes makes MDM projects so hard to implement.
Other reasons MDM and DG projects can be tough:
- Difficulty in obtaining executive and business buy-in of the MDM concepts
- Cross-enterprise scope and impact causes angst
- Difficulty to properly scope and execute the project
- Heavy reliance on a strong business/IT partnership
- Extent of organization changes required to succeed with data governance, data quality, and business process re-engineering
- Lack of measureable benefits
- Lack of understanding about long-term implications of the program
The right mix requires a business-driven initiative with IT firmly behind it and commitment from both sides.
Key Success Factors
MDM and DG initiatives need to start with a strategy and roadmap. This includes setting realistic scope and expectations and teasing out some ROI to show value.
Start with a pilot project that becomes the enterprise project without adequate business investment (funding, people, systems access etc.). If you fail to recognize that Data Governance, Data Quality, Data stewardship and Metadata are prerequisite for MDM, this impacts the delivery of the project. Implementation must show ongoing value of the project. A big-bang approach should not be applied to MDM and DG projects and organizations should accept that MDM and DG is an evolutionary process.
|Category||Key Success Factors|
|Clarity of Purpose||
|Think Globally;Act Locally||
|Communication is Key||
Implement Data Governance and MDM as a bundle. Start with data governance and queue up MDM as a part of DG initiatives to address data quality and single version of truth for Master data.
In my next blog, I’ll explore the various architectural concepts of the MDM implementation as foundational building blocks for a successful MDM and DG program. If you’ve already walked this path and have some learning to share, feel free to engage.