We talk about data governance a lot…what it is and why it’s important. But what does it look like in practice? Better yet, what needs to happen before the implementation phase?
My colleague Abbie Byram explained why governance is the key to your business’s data-driven future. But in short, data is only as powerful as the people who wield it — so if you have data, but you don’t know the right context around it, then you can’t pull the right insights. The only way to get the right information out of that data is to know who knows about that data and to really understand what it’s describing. That context is key. And that’s what data governance enables.
I earned my undergraduate degree in English and am fascinated by this work. The process of creating a shared data language and data sharing strategy across networks and organizations blends technology and writing in a way they don’t tell you about in school. It’s exciting (and frankly, pretty cool) to use my background in English on data projects.
Here’s how we approach discovery on data governance projects:
#1 Build the foundation for solid data governance
Data governance is highly dependent on the culture of the organization. Effective governance requires deep discovery, buy-in from business leadership, and rethinking the organizational design. When leaders fail to recognize the value potential in data governance, it becomes another set of policies carried out by IT and ignored by most employees (who are expected to uphold and maintain it).
We meet with stakeholders at kickoff and build a solid foundation before entering discovery with the larger team. We facilitate a conversation to…
- Identify pain points.
- Look at current data practices in the organization.
- Brainstorm the future state and map where they want to go.
Then, we organize this information into pillars of governance to determine what is the most crucial to accomplish first.
Once we have a framework, it’s time to dive into people and processes.
#2 Focus on people and processes
Meet with people. Ask questions. Dive deep to learn more about the organization and its existing processes.
During this phase, we invite the client’s teams to meet for interviews, then to connect us with different stakeholders to gain a comprehensive representation and avoid bias. We ask questions like:
- What is your experience with data governance?
- Where do you run into problems?
- Where do you get data from?
- How do you access data?
- Where are the points of friction?
- What is the handoff process between groups?
- What does your ideal future state look like?
- What didn’t we ask that you feel like we should know?
The last one is key. Asking an open-ended question empowers people to share valuable input.
#3 Draw insights from raw notes during discovery
This is an iterative process. There’s no one-button solution when sifting through a large volume of notes and feedback, but here’s an approach that works for us:
- Read the raw transcript and take notes. We look for early themes and patterns in our conversations. For example, what are people asking about when talking about relationships? Or, when discussing engineering, is it in a positive or negative context?
- Take that transcript and turn it into a spreadsheet to see patterns and themes. In this stage, we’re establishing a foundation for analysis. This is also exploratory, setting up a data structure we can tag, filter, and sort, and performing quality control checks (e.g. checking for misspelled words, missing values, etc.).
- Run the data through a series of tests. What words come up most often? What are the most prevalent word clusters? When a word comes up is it positive or negative? Were people in agreement? In this step, we cross-check the results against our initial notes, asking questions and applying what we learned.
- You might need to revisit the original dataset to structure and clean the data. In this stage, we make case-by-case decisions that help us refine our results. For example, applying stop words to filter out less important information in the text. In some situations, we may replace contractions to get whole words. As we explore the text and continue to check our understanding, these decisions enable us to make broader thematic connections across the data set.
The more you explore, the tighter the narrative gets.
#4 Uncover sources of inertia and recommend data governance solutions
After discovery, we read out to the kickoff group sharing what we found: similarities, differences, and more. For example, everyone agrees it’s difficult to access the data they need or everyone wants a shared process to avoid the risk of data privacy breaches and misalignment of information.
What happens next?
We strategize and build a roadmap of the organization’s governance plan. Then, upon approval, we segue into the implementation phase, putting our discovery into practice. Stay tuned for Part 2 of this article! I’ll share our approach to implementing data governance strategies.