Five reasons for master data management failures

William Dorner

William Dorner

A successful master data management (MDM) implementation relies on many work streams collaborating as one team. A few of those work streams include: data integration (i.e. ETL), data quality, web services, architecture, data governance, and data profiling and matching, amongst others. When companies forge ahead with an MDM implementation without proper planning, the MDM program ends up not meeting business needs and not being used, and often needs to be re-implemented or modified, costing additional time and money.

To help MDM program sponsors and leaders that are beginning their MDM journey, here’s five of the most common pitfalls of MDM programs—and solutions for how to avoid them.

Not built to address a business problem

Before you begin your MDM project, it’s imperative to understand why you are developing a master data management program. Ask your sales, marketing, operations, and finance teams which business issues they’d like to resolve or improve with an MDM program. Reach out to these departments in your organization, understand their pain points, and build a prioritized list of problems the MDM program can address.

Without understanding your business goals, your MDM program isn’t likely to be effective. When an MDM program is driven with purpose, it will be much more likely to succeed—and it will also help you gain support for funding to build and expand the program.

Related issues that detract from a successful MDM program include:

·       Too broad of a business case

·       Constantly changing requirements

·       Built strictly from an IT perspective

Lack of data governance

Many organizations initiate MDM implementations by simply laying out the initial tasks to build an MDM hub, such as: gathering data from different systems, consolidating the data into a repository, matching the records and outputting a “golden record.” These organizations don’t focus on understanding or defining their data and its usage, procedures, and policies, or defining roles and responsibilities to ensure data quality flows in to and out of the hub. This failure to consider governance leads to a complete lack of confidence in the organization’s MDM program—without governance, the program cannot resolve data issues as anticipated.

As a precursor to implementing an MDM program, organizations should build a data governance work stream that has executive sponsorship. This will address defining the critical attribute standards, data policies, procedures, and roles across the organization to ensure a consistent understanding for all contributors and consumers of the MDM hub. Other benefits of data governance are:   

·       Issue-resolution process

·       Improved confidence of your data

·       Optimal operating model

.     Meeting coordination plan

Exclusion of data quality

Data quality needs to be prioritized early in an MDM program just as data governance should be. For many early MDM implementations, executing standardization rules and ensuring data consistency were not considered critical success factors. It was only important to identify the data attributes to be utilized for matching and identifying records.

As the master data management space has matured, it has become clear it is necessary to build data quality into MDM programs. There’s been a shift toward focusing on the initial data quality as well as the ongoing data quality of the master data attributes. In conjunction with data governance, organizations now put their data quality first, understanding the repercussions of poor data in an MDM hub.

Ignoring your data quality results in the poor matching of records, increased workload for data stewards,  decreased user confidence, and increased time and cost due to rework. It is essential to include data cleansing to defined business standards as part of any MDM implementation.

The wrong tools and architecture

Just as important as implementing MDM to solve a business need, the right software and architecture needs to be selected. Not all tools can support all architectures. It is crucial to understand what is required for your implementation prior to selecting a vendor and/or consultant. For example, you’ll want to review the business needs (current and future), the data integration plans and requirements, the data quality rules, and the different master data views required. This is just a short list, but the critical point here is that your requirements will drive the best architecture and tools for your implementation. Be sure to not get trapped by determining your tools and architecture and then squeezing your requirements in to fit.

Lacking the right team

Lastly, an MDM program is ever evolving as businesses change and grow. Don’t only staff your program with consultants, but with employees that can carry the torch as the first implementation is completed. These individuals can work side by side with the consultants to gain valuable knowledge and insight to understand what will be required to maintain and grow the MDM program. The most successful programs have been led by consultants, but driven by employees that will continue to adjust with the ever-changing business world to meet the needs of their organizations.

About William Dorner
William Dorner is a Solutions Manager in the Information Management practice in Slalom Consulting’s Chicago office, focusing on Master Data Management, Data Governance, Data Integration and Data Quality.

One Response to Five reasons for master data management failures

  1. Madhu Kiran Ivaturi says:

    This is a good one William. Share any best practices around Data Governance best practices as well..

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: