3 questions to consider before starting with data governance  

Ana Marija Galic


In our previous post, Jason talked about the role of federated data governance in data mesh. In this post, I would like to tackle some of the questions you should have always have the answers to, before deciding how to start with data governance.

Let’s dive right into the questions we ask our clients to help us both determine their potential next steps in data governance implementation and work prioritization. With each question, or group of questions, we will also explain why it should be asked, and how it can help.

1. What are your biggest data related challenges?

We usually start with this question because it is the easiest for our clients to answer. This question should help you to determine your priorities and therefore, the next steps when it comes to data governance.

Now, depending on the data governance approach that one follows, one can argue that there are clear next steps for data governance, like assessing current state and defining scope, defining and filling roles, and following with a defining data governance mission and vision statement, however we find this to be too abstract for most. You see, the issue that most of organizations have, as Lauren Maffeo, author of “Designing Data Governance from the Ground Up Six Steps to Build a Data-Driven Culture” nicely puts it, “[Data governance is] an admittedly nebulous term that can seem to mean everything and nothing”. What we try to do with this first question is “personalize” data governance to you and your organization by bringing attention to the issues you know you have, and check if there are ongoing solutions, instead of formalizing governance and building it from the ground up. Later, we will use these answers as initial points when building a governing framework.

We believe this approach brings you head on into thinking about data ownership and accountability, two important enablers of data governance. In the case of our clients, it makes them face the realization not understanding data governance sooner, so we can be there to help. Basically, this wall comes sooner than in traditional approaches where in the initial phases you have a feeling that you are doing something governance related (after all, you are hiring, writing perfect mission statements etc.) when in reality you are postponing the “what now” moment when it is time to turn theory into practice. It causes clients to think about what their definition of governance is, and what they want to highlight and formalize.

Another reason is the agility of this approach – doing it now and fixing it later when it is centrally formalized, rather than waiting to create it once and for it to be “perfect”. In this later approach it often happens that teams just follow rules blindly instead of understanding the why behind that “perfect” approach, which again impacts accountability (“I followed what centralized governance told me, so I can’t be blamed” instead of “What is the impact to this, with my data/consumers/domain”).

Lastly, thinking about different issues in the organization allows teams to prioritize what they need, and therefore be able to have it right where they are in their governance journey. If we were to follow a centralized approach, then everyone should end at exactly same point, which for some teams is harder than for others – as some teams are more mature than the others, some have greater technical debt, and some are more impacted by specific issues compared to others (motivation). In conclusion, different teams have different struggles, and following the same approach is not going to yield the same results.

2. What are the drivers for this initiative?

This question is not necessarily explicitly asked, but finding answers to it is important for understanding the state of the organization and its maturity, as well as to some extent if there is (and if so, where) a real desire in the organization for change. It also helps us find out if there are some burning issues (i.e., legal) that prompted it but that did not come out through answering question 1.

Answering this question will also help you to better prepare for the organizational pushback that will inevitably come. For example, if the data governance initiative was a push from the top that usually means good support from upper stakeholders and faster buy-in, but will probably cause turmoil on the level you are operating. Interestingly enough, there are 2 different scenarios in cases when governance is a top-down initiative: The first, people in the organization embracing the change but not really being ready or knowing how to change, and secondly, those that will show you they don’t want the change on every possible step of the way.

Now my personal preference is the later type of people. Although it is exhausting to explain (or even defend) every word you say, at least they vocalize their concerns and fears, giving you room to address them.

The former ones on the other hand will often actively participate without complaining until it is time for takeover when they report to higher-ups (and whoever wants to listen to them really) how unsatisfied they are with the proposed solution. They might even create some benchmarks that this data governance initiative has not satisfied, but that were never before disclosed, and that are often not even relevant for the initiative. As an example, maybe according to this assessment, data mesh is not way to go for them, because the data catalog tool they chose can’t show lineage well for their main system.

This is where question 3 “What do you see as a result?” and the conversation around expectations comes into play – making employees talk about their expectations and bringing them onto the same page, as well as explaining concepts repeatedly and keeping everyone engaged and responsible -because they are not lacking inclusion but rather engagement. However, escalations will happen, since question of ownership always brings out fear of change, authority, and autonomy, and fear causes a fight (direct escalation), flight (avoiding responsibility), or freeze (do nothing and let initiative sink) response, with each response causing an escalation in one way or another.

In any case, be aware not to fall/run into the pitfall of having too much management. Sometimes certain people in the organization will try to micromanage the whole initiative, which on the surface may seem as if you have embraced the concepts, when the reality is that you are still where you were at the beginning because there was no real autonomy (or even time) for the teams to really participate in the way they found suitable for them.

As you can see, the idea behind question 2 is to help you and everyone involved, understand your organization and stakeholders, and prepare for the social part of data governance. In the end, organizations go with major changes like data mesh for one of two reasons, number one, reaching the ceiling of the current architecture and being unable to proceed without creating bottlenecks or technical debt, or two, wanting to proactively change before reaching that ceiling. In any case, change is inevitable.

3. What do you see as a result?

The final question we will present here is about expectations and results. As aforementioned, we want to know what questions or issues you and everyone involved expect to solve at the end of this initiative (and during our collaboration). This question helps to get everyone onto the same page – comparing what was said vs what was heard, while simultaneously helping you to determine how deep you should go into your process or what the scope should be. In some cases, you can’t go too deep because the organization is not ready, so it is better to have a few simple principles for data governance for the whole organization, than to go into too many details. In other cases, you will go much deeper with the details but will do it for a few data products. It all depends on the information you gathered through answering questions 1 and 2.

One thing that happens surprisingly often is that the goal the organization sets up is to get data from one source. It sounds simple and not wholly related to governance, but then turns out to be an issue of someone holding data hostage. Such individuals or teams are afraid they will lose authority or relevance if they “release” the data. While in some rare cases this can be true, in most cases it is just the contrary, with governance formalizing the authority while also, within the context of data mesh, opening that data up for self-serve. In any case, be aware of that pitfall as well.

The bottom line is that data mesh and data governance, although very powerful, are not silver bullets for all your data and organization related issues. They are just tools you must learn to utilize to become an effective, true data-driven organization.

Have these questions helped you to assess your organization? Have you found them useful? Let us know your thoughts! If you have any questions or comments, feel free to contact us.