This post is the seventh in my “5 Keys” series covering the core principles of data management. Check out the others:
1 – Introduction to the “5 Keys” Series
2 – The “5 Keys” to Data Governance
3 – The “5 Keys” to Data Architecture Management
4 – The “5 Keys” to Data Development
5 – The “5 Keys” to Database Operations Management
6 – The “5 Keys” to Data Security Management
Today’s topic is master data management. DAMA technically calls this topic “Reference & Master Data Management,” but I think reference data is a specific type of master data, so it can all roll up into one subject. Why make things more complicated, right? Master data represents the most important data entities in your organization – those ideas like “customer” and “product” that go beyond a single system or line of business.
Figure 1: 10 Data Management Disciplines (adapted from the Data Management Association)
KEY 1: If it is frequently changing, it probably isn’t master data
Sales data is important – nobody should argue against that. Sales data is not master data, however, because it is constantly changing. If I asked, “What is our sales data for Customer A?” it would be hard to answer. Am I talking about annual sales? Daily sales? Because this is data that is measured over time, the answer will vary.
By comparison, customer A’s mailing address is likely master data. Something important is likely master data when it can be shared without a time-based qualification. “What is Customer A’s address?” “123 Main St.” “That’s some great master data right there.” “Thank you.”
KEY 2: Not all master data candidates are worth being treated as master data
Just because it meets our first criteria doesn’t mean it is worth the exalted status of master data. The more master data an organization chooses to manage, the more effort and complexity (and cost!) will occur. So before deciding that all 200 attributes of your customer should be considered master data, think carefully about the implications. Decide which are the most valuable: start with the top 5 or 10, and get that right first and then loop back to the others when the resources are available. Otherwise 5% will get done, and then the resources will run out and the whole program will fall apart with no real progress. It happens all the time.
KEY 3: The first function of master data management is to consolidate and unify disparate data
Business units usually have some autonomy, and it is this autonomy that leads to their unique data making everybody else miserable. The catch-22 is that business unit autonomy is an important part of a business making money. If a business unit couldn’t go to market to make money until they adopted data standards that apply to all areas of the business, we wouldn’t have a very dynamic or responsive business, would we?
Master data management, in conjunction with data governance (obviously), looks at the enterprise and figures out how to bring data together so that it can be reasonably useful to everybody who cares. The downside is that sometimes business units have to give up a little of their autonomy, and sometimes the corporate standard needs to accommodate a few things that not everybody cares about. The idea is that these decisions should be made deliberately, and with an eye towards what is best for the organization as a whole.
KEY 4: The second function of master data management is to communicate in two directions
Master data management nearly always communicates unified data to downstream systems, but the Holy Grail is in the feedback loop. If you can feed more comprehensive, standardized data back to the transactional/source systems, data integrity improves everywhere! If the source systems feed better data to the master data environment, then fewer cleanups will be necessary and more refined cleansing processes can be instituted. It’s magic!
The other point to this key is that if you go through the trouble of capturing and cleaning master data then make sure people can access it. There is no greater tragedy in this world than master data going underutilized! That might be overstating it (slightly), but it IS important.
Great communication alone will not guarantee success, but poor communication alone will guarantee failure!
KEY 5: Master data management doesn’t always have to have a separate repository
It’s hard to talk about master data without mentioning a master data repository, but a separate repository isn’t always necessary. Sometimes it makes sense to keep the “official” master copy of data in the system that uses it (and changes it) the most. Considering using Salesforce as a customer master data platform? Why not? Chances are it can capture everything you care about regarding your customers (and prospects). You might need to be a little more diligent around data entry and how changes get made, but these are solvable problems.
An organization’s master data is fundamental to the characteristics making the organization unique. The solutions that can be employed in managing master data can be as unique and creative as you like. Consistency of communication and execution of the standards across the entire organization are the non-negotiables. Never lose sight of those, and your master data management initiative will be successful!
Anthony J. Algmin leads Data Governance in West Monroe Partners’ Information Management Practice.