We’re not afraid to admit that sometimes we’re wrong. Well, maybe not “wrong” as much as not fully informed. After all, we’re only humans — despite all of the data we work with on a daily basis.
Such was the case, recently, when an astute observer read our recent piece on data governance. While our piece covered lots of topics under the umbrella of data governance, it perhaps missed some context and a forward-thinking viewpoint on the matter.
Luckily, this was no average reader. It was Ties Carbo, Consulting Director at Artefact — a Funnel Solution Partner. It just so happens that Ties is an expert on modern data governance, and he works with clients everyday to create best-in-class data governance structures. So, we decided to sit down with him and capture his hottest takes of what we missed.
Consulting director, Artefact
Three things people misunderstand about data governance
At the beginning of our conversation, Ties pointed out that there are three main elements that he sees his clients misunderstand most about the subject.
First, many people view this as just an IT problem. It’s not hard to see why. Data governance involves lots of potentially complex technical components such as models, processes and platforms. However, from Ties’ perspective, the implications of proper governance stretch far beyond the realm of IT.
“Sure, there are lots of technical components to data governance,” said Ties, “but it’s an organization-wide topic.”
This plays into the second misconception: people think it’s only about compliance.
The truth is that this topic covers so much more. Compliance around issues like GDPR are only one component of how proper data governance can affect your entire organization. In fact, Ties often advises clients to confront compliance issues as the final piece of new data governance structures..
Which brings us to our third misconception.
According to Ties, many teams work on data governance issues in isolation from each other. One part of the organization works on data governance but focuses solely on compliance, while another group ensures that data-in warehouses are of the right quality, and more.
The proper approach is to view all of this holistically, like one giant data ecosystem. When you begin to see data governance from a broader perspective, you quickly understand how it is a business-wide concern.
Michelin-starred data governance
Ties like to think of data governance almost like a kitchen. You need to keep your kitchen as clean as possible. This cleanliness (and the act of cleaning) can be thought of as your data compliance roles and responsibilities, policies, and processes. However, you still need to cook in that kitchen, right? Why else would you have a kitchen?
Ensuring that all of your cooking utensils are in place, your pots and pans are ready, your sous chefs are trained properly, your recipes are being followed, and your food is going out to guests encompass the rest of proper data governance.
You see? Having a clean kitchen will help you deliver a great food experience, just as data governance can help you derive value from data.
So, what is data governance exactly?
Ok. We know that compliance is just one part of the whole data governance concept. What about the rest? According to Ties, there are four main components: data structures, data policies, data operating models, data tools.
Think of this as your models and maps. They help you to understand where your data is and what it is. To go back to our kitchen concept, the head chef should know where each station is located inside the kitchen. Additionally, each cook should know where each part of their mise en place is stored.
These are the rules you put in place. Yes, compliance can be found here, but so can your policy on data quality. You’ll need to set your standards for how data is maintained. In the kitchen, a chef needs to set recipes, make sure those recipes are followed, and taste each dish as it goes out — making sure each is plated to perfection.
Data operating models
Let’s imagine you have your structures set and policies in place. You still need to identify who is responsible for each facet data in the organization. For instance, who is responsible for keeping sales data clean? Who can identify errors in the data? .
In our kitchen analogy, a chef needs to assign a cook to each station according to their current specialities. The cooks in these different areas of the business (the kitchen) then run their stations according to their specific needs.
This one is pretty straightforward. This focus area concerns the various elements of your data stack that you use to manage your data. For companies that just start their data governance journey, this could be Excel sheets and Visio diagrams. But the more mature players have dedicated tools, including data lineage, meta-data management, data quality management, and more.
The ideal model of data governance
During our conversation, Ties mentioned that data governance is often an overlooked issue in many organizations he deals with. Things seem to be slowly changing, though.
As business leaders are waking up to the importance of data as a strategic asset, they are beginning to see the value of good governance.
Even various departments are coming around to the idea that data governance is no longer just an IT issue. For instance, in marketing, we understand how valuable high-quality data can be. But only marketers can gauge the quality of the data in a marketing context.
“The IT team can’t tell you if the data is correct,” said Ties. “They can only tell you that the data is where it’s supposed to be. Specific teams need to ensure the data meets the standards of their use case.”
The ideal approach to data governance, according to Ties, is the data mesh approach.
“In a truly data-driven company, everyone is working with data simultaneously,” said Ties. “By following a federated governance structure, everyone has a vested interest in the data and with the ability to scale the structure.”
Data governance for the AI era
This modern approach to data governance, wherein everyone in the organization “owns” a piece of the data, is becoming increasingly important. Look no further than emerging generative AI tools.
While many people are generally aware of ChatGPT, its underlying technology can be used (thanks to its open source nature) to create proprietary AI tools. For instance, a business could take GPT-3 or GPT-4, load it onto an isolated server that they own, and feed it with loads of proprietary data to create a custom, business-specific AI.
A SaaS company may want to implement a similar system to create a highly knowledgeable chat bot to answer questions from existing and prospective customers. It could advise users how to best use a product, or convince prospects why one piece of software is better than another.
This AI tool will only be as good as the data you put in it, though.
“The data (and the quality of that data) you can feed these tools is your competitive differentiator,” said Ties. “Good data governance is a key component of this. Without proper governance, the data quickly turns into a mess, which leads to a less effective AI.
“The company with better data and data governance structure will have the advantage.”
Reap the rewards of good data governance
Every business requires its own unique approach to data governance. However, by using a data mesh approach as your North Star, you’ll set yourself on the path toward greater competitiveness and better data outcomes.
Whether you’re looking to improve the quality of your data, meet the compliance needs of a new market, implement new tools, or more, a strategic and holistic approach to your data governance will help you go further.