As technical data experts, a question we often hear from current or potential customers is, “Why do I need a data governance strategy if I’m already using master data management?”
If there were a simple answer, we wouldn’t need to write an entire blog post about it. However, data governance coach Nicola Askham does an excellent job explaining the reason with her comparison of how we treat our own physical health.
“Consider your Master Data Management repository is the human body… if you put
good, clean, healthy data into it then it's going to work really well. You are going to have
the right data in the right place, the right processes will work, you will make the right
decisions on the data.”
Nicola wrote a piece on the importance of having data governance even while using master data management—and after sharing it amongst our team, we felt inspired to build upon her idea with our own thoughts.
Below, we share our experiences with the situations Nicola explains on the data governance topic. Continue reading to see if you fit one of these scenarios.
What is Data Governance?
In technical terms, data governance is defined as “a system of decision rights and accountabilities for information-related processes, executed according to agreed-upon models which describe who can take what actions with what information, and when, under what circumstances, using what methods.”
To simplify things, data governance is really just the change management of data that ensures the effective and efficient use of information per the requirements, standards, or rules that an organization has set for its individual business goals.
When an organization has implemented a master data management system, the data governance ensures the privacy, quality and purpose of the master data within the initiative.
Many people think data governance is a separate process from master data management when in reality, master data management requires data governance. Just because they are labeled differently does not mean they are mutually exclusive.
Keep in mind, you can't truly "master" data if it’s ungoverned. Mastering a source of truth requires both solid information architecture and clean data for an individual master (e.g. product, customer, supplier, etc.) in addition to the points of interconnection between masters.
Companies that master data well do so holistically—and companies that are best in class with data governance take a holistic approach as well. Data governance is an entire business function, just like master data management.
No End Goal In Sight
One of the most common problems Nicola sees companies run into when working in a new classification system is not having thought about the future of their project.
She explains how even when your data migration is seamless, when there’s no data governance in place to protect and manage the data, as time goes on, your world will begin to turn upside down—and if you’re a Stranger Things fan, you know the upside-down is not the place to be.
This reminded me of a time when one of our customers had us complete a large-scale data enrichment project but did not have a long term governance operation in place. The project was completed successfully, but over the next year and a half, their data began to drift as they acquired, classified and launched new products, categories, and attributes without strategizing how to properly implement those changes as an ongoing, “steady-state” governance operation.
The data became messier over time. Two years after the initial project, they were forced to tackle another large-scale cleansing and enrichment project to address the drift when they could have saved 75-85% of that budget by maintaining a data governance program instead and redirecting resources toward other priorities and operational needs rather than a large-scale cleansing initiative.
Without a data governance strategy in place, the quality of your classification system becomes messy—and let’s face it, no one likes to work in a dirty environment.
Failed Data Migration
For some, Nicola expressed how they’re able to still get through the data migration process without a data governance strategy in place, but for others, they can’t even get that far.
Oftentimes it’s because there hasn’t been enough engagement with users and the company hasn’t done enough analysis for the data to be migrated properly and successfully. Nicola hammers in that without a proper data analysis, you’re bound to find endless issues with the data not being mapped correctly or it being too poor of quality to actually fit into the system.
I have seen several customers of ours end up in this situation. Migration projects can hit failure points when the staging environment and the target system are not in sync in their configuration—it can be as simple as not realizing there are differences in field lengths, for example.
Looking closely at how the data will be made ready to fit the target system is critical. What happens if units of measure are not mappable on a 1-to-1 basis? What if there are other gaps in the target system that have not been identified in how they relate to your data model?
If your staging environment handles fractions or multiple attribute values just fine, the question is, does your target system? Proper data governance must engage on where the data will be used, how it flows across the digital landscape of the organization, and any constraints or issues that emerge along the way.
Classification systems are a great investment, but when you don’t do the grunt work to use your new platform to the best of its ability, your time, money and efforts will all go to waste.
Lazy Project Management
Have you ever tried shoving so many clothes into a suitcase in hopes that it will somehow shut all the way? This is exactly what I thought of when Nicola explained what happens when companies migrate all their dirty data into their classification system without cleansing it beforehand.
Once all your clothes fit, you’ll just fold them better once you get to your destination. Or once all your data is in the system, you can just validate and normalize it then—no need for it to be done as a part of the project.
Not only will this lazy project management cause stress among your employees, but Nicola let us know that your system is likely worse off since you’ve shoved dirty data into a new platform that is set up differently.
An iterative approach works best for these initiatives. Validation from business stakeholders before loading data to a target system is critical in a classification, cleansing, and enrichment project. It can result in identifying changes that need to be made to the category structures, attribution schema, selection of filter attributes, and more.
It can be a lot harder to attack these data problems at scale if you’ve already “loaded the mess.” Planning for change allows for the opportunity to adapt as you go.
Yes, You Do Really Need a Data Governance Strategy
A big shoutout, again, to Nicola Askham for providing our team with great insight on the data governance and master data management relationship. If any of the above scenarios sound familiar to you, it’s likely you’re operating on a system with no data governance strategy. Contact us today for support in getting data governance up and running in your organization.
