By Debo Mukherjee,
Senior Consultant - Data Engineering, Clarity Insights. A Data Enthusiast with extensive knowledge in Information Management and Data Architecture. Well versed with traditional On-Premise Datawarehousing system and Big Data Hadoop on Cloud platform.

When John Smith is Not John Smith. A Cautionary Tale of Unmastered Data

We’ve all experienced this familiar scenario: Your company prints and mails a catalog to your customer, John Smith. You also send one to Jon Smith at the same address and another to his old address. That’s three catalogs sent. Two are a complete waste. In the aggregate, data errors like this can cost your company a lot of money. Master Data Management (MDM), the practice of creating a single master data set from which everyone works, can help eliminate these problems. 


Understanding the High Costs of Bad Data 

Simple mistakes involving data can be costly to fix. According to a 2018 Experian survey, up to 30% of customer data is inaccurate. That means if you have five million names on your customer list, you might be paying to print one and a half million catalogs that will go straight to the landfill. 

Think about what you pay to print and mail a catalog. Typically the cost is a few dollars a piece. A catalog mailing at this rate will cost you between three and six million dollars. Do that on a monthly basis and you’re burning up between 36 and 72 million dollars a year! Not to mention all that wasted paper. That’s not even counting the wasted time your people will run up dealing with merchandise returns and lost orders that arise from having the wrong data on hand. 

So what is the root of the issue? Your company likely lacks consistent, accurate and maintained master data. 


What is Master Data Management?

Master Data Management (MDM) is a discipline of practices, policies and tools that come together to build and maintain your master data set—ideally, a single source of truth for all of your business operations. The goal is to ensure that any time your business creates and uses that data, that it will have high integrity, will be complete and will be accurate. 

MDM isn’t a single point of time when it comes to data management - it is an ongoing effort. There’s always new data to ingest and conform to MDM standards. Applications constantly require new MDM rules and adaptations of policies to keep the organization’s data clean and correct.


Establishing Master Data Management Best Practices

Building a program around MDM requires key fundamental operations.

1. Institute Master Data Schema

First, you must establish the parameters that will be used for the data records, also known as a master data schema. Essentially, this is a set of business rules that the data must adhere to. This might include details like whether you use middle initials in people’s names, 9 vs. 5-digit zip codes, abbreviations for “street” and so forth .e.g. “Street” vs. “ST” vs. “St.” John Smith’s address from Source 1 might be “1000 Center Street” while, from Source 2 it is “1000 Center ST.” Now, both point to the same address and usually standardization/normalization rules make it the same.

2. Ensure Data Correlation and Integration

Then comes data correlation and integration. When data comes from multiple sources, it’s common for data to live across systems, but might vary depending on the purpose or use case of that record. Often, data users will have different ways of describing the same thing, After you’ve standardized, you have to ensure that there is matching and linking of data records. In our story, this would start by identifying that John A. Smith, John Adam Smith and Jon Smith in Cleveland, Ohio are all the same person. Making the identification requires a process known as correlation. Each of these three seemingly different people share the same address. Therefore, you can infer that there are three separate records for the same person. 

3. Perform Record Consolidation

With successful correlation and matching, you can consolidate the records. Again, rules come into play. To consolidate three records into one, without losing any useful information in the process, you need clear-cut master data policies. If the data needs to remain in different systems, there will need to be a process for ensuring that when changes to the record occur e.g. John Smith changes his address, that it is updated in all systems - whether through management processes, or through a data governance platform.

At this point, unless every application that uses John Smith’s data follows MDM best practices, the data will quickly devolve into a disaggregated mess. 


Building an Enterprise-wide MDM Program

MDM principles are easy to understand but performing MDM at scale can be a big challenge. This is where MDM tools come into play. The market offers an abundance of options with virtually all the major database providers offering MDM functionality. Several standalone MDM platforms also enable advanced MDM at scale. 

Putting MDM tools to work is not a push-button process, though. You need to implement the tools in alignment with business goals and enterprise-wide objectives. Here are a few things to consider when creating your master data management strategy.


  • Take a business-oriented approach: It’s a good practice to bring data stakeholders into the master data management process early. Working with them, a team can establish a business case for MDM, e.g. it will save money on wasted postage and unnecessary administrative tasks like tracking down orders shipped to the wrong person. The business case can create much-needed buy in from the business as well as a justification for the investment in MDM tools and people. Executive sponsorship is essential.
  • Maintain focus: MDM needs to be a cohesive program and not a one-off project. Keep objectives in clear view while building out a framework. What does your organization want to achieve? What does long-term success look like? How will the program guide new decisions within the organization?
  • Don’t let software drive the program:  It may be tempting to buy a suite of new software and plan the program around the tool. Technology is only as good as what it enables. A best practice is to gather MDM requirements and then work backwards into selecting the right tool which meets program objectives.
  • Define and implement your governance policies: This can be a big or small undertaking, depending on your organization’s use of data and who creates, maintains and approves it. Data governance can involve things like how your company uses SIC codes, but it can also be driven by your organization’s regulatory compliance, privacy laws or a need for a clear audit trail.
  • Build a timeline with phased rollout: MDM has a learning curve. A resulting best practice is to take a phased approach to the MDM rollout. One example of a lesson that often gets learned the hard way has to do with affiliations or relationships between people and business entities. Even in a phased MDM rollout, it’s a good practice to plan for capturing relationships. Consider a pharmaceutical company that uses MDM to build a master physician database as a single source of truth for its sales team. What you can miss in this process is the hospital affiliations that doctors almost always have. It may not be in their official data record, but it’s valuable data nonetheless. It is necessary to plan for this relationship and map to it as you roll out your doctor database with MDM. 


Now What?

Master data management offers many business benefits. These include reductions in data-related problems and their associated costs, e.g. time spent resolving mismatched records. It’s a simple idea to grasp, but a relatively complicated practice to put into place. The choice of MDM tool is important, but so is the adoption of proven best practices. MDM should be business-oriented, with a stated business value and data stakeholders across the organization should be engaged early on.  

Clarity Insights has worked with many organizations on the development and implementation of MDM programs. To learn how we can help your business build an enterprise-wide MDM strategy and mitigate risks, let’s talk

You may also like:

Machine Learning Data Management Artificial Intelligence

What Will Augmented Data Management Mean for Your Organization?

To understand augmented data management, it’s first necessary to know what is involved in basic data management. “Data m...

Data Management Data Architecture

No More Data Fabrics. It’s Time for the Data Trampoline.

Another day, another data management metaphor... Gartner, for example, cited the “data fabric” as one of their “Top 10 D...

Data Governance Data Management

4 Data Governance Best Practices You Need to Follow

In the rush to get exciting, advanced analytics projects off the ground, don't overlook the importance of data governanc...