The original version of this article was written in Spanish by Rubén Obando and translated into English by ChatGPT.
Those of us who enjoy working with data have been fortunate to witness the accelerated evolution of the industry and the fascinating increase in technological capabilities that enable us to make data-supported decisions.
I remember when we used 3.5″ floppy disks to install any application and even to enjoy a slow and pixelated — but entertaining — game. Not to mention the early web pages we navigated, where it was almost a lottery to ensure that our homes wouldn’t receive calls because the dial-up line shared internet access. Let alone the fact that the most privacy and protection we could afford for our desktop PCs was to save our most valuable information on a removable floppy disk.
Recalling moments like these involves managing a large part of the information we have in our minds, and that is precisely what I want to address in this writing. Many times, we think that when we talk about data governance and management, we are only referring to technical areas or, at most, a dictionary of business concepts, metadata management, or the indispensable data quality. But today, we really want to go beyond that.
The strategic value that companies from all industries have given to data in recent years is undeniable. It is very common—almost a commodity—to see, hear, and feel that more and more companies have the goal of being data-driven, considering data as the most valuable asset, and emphasizing that decisions should be based more on data than on intuition, “gut feeling + experience.” While there is already awareness of these issues at the highest levels of many companies, the real challenge today lies at the cultural level.
It’s good to have that clarity at the executive level; however, the next step should be for the tactical and operational areas to internalize and materialize the importance of true data product creation to close the life cycle as expected.
By this, I mean not only ensuring the best data ingestion, processing, and visualization but also transforming this into a product that generates a competitive and sustainable advantage for the organization and/or department, beyond remaining a simple project like others. It is at this point where we can truly approach value generation with data. In other words, we can begin to find the ROI (Return on Investment) in the investments made in team training, solution development, and technology acquisition.
Today more than ever, our companies need (and should) define mechanisms in their strategic plan to govern and manage all data interacting with their processes appropriately.
The Pillars of Governance
Below are, in my view, the pillars of a Data Governance strategy:
#1 Data Strategy
The data strategy must align with the organizational strategy. Let’s think about each time we go on a trip:
1° The first thing is to know where we are going and why we want to reach that point.
2° The second is to know what resources we have to do it.
3° Finally, we have to think about how we are going to manage these resources to reach our destination efficiently, appropriately, and in compliance with necessary regulations.
We think of this triad as articulated in such a way that we can scale and generate added value while ensuring sustainability and growth.
#2 Structure
Everything, absolutely everything, in the fascinating world of data has grown exponentially: teams and profiles, platforms, languages, needs, and opportunities.
But have we defined or assigned a team to manage this “growth”? Have we faced the question of who is responsible for this or that feature? To answer these questions, the first step is to define a structure that sets the direction for this management.
Let’s call it an office, committee, forum, group, etc. But always think that data governance and management today consist of a group of people with the authority to define, validate, and promote a culture of care and awareness of data.
Only then can we start talking about profiles, roles, positions, etc. Forget whether we call them the same as in one book or another—call them what makes sense within the company, what is functional for us. However, let’s be clear about what their function will be in this selection (whether it’s a goalkeeper, defender, midfielder, forward, equipment manager, doctor, etc.)
#3 Eating the Cake in Pieces
Just like in data projects where we think big but execute in short bursts to achieve tangible and value-generating results in short periods, the same happens in data governance and management projects.
Our organizations cannot afford to wait for theoretical and lengthy implementations where, in the end (if we’re lucky 🤞), the documentation is read once and, if we’re still lucky, will stay updated for a few months.
Data Governance today expects tangible results beyond one or a thousand documents.
It expects us to add value in the short term (pipelines, dashboards, predictive models, bots, etc.), and we can achieve this by prioritizing and executing domain archetypes one by one, defining, developing, and implementing archetypes one by one.
#4 Extensive and Excessive Documentation vs Tangible Results
The era of deliverables where we had documents as the great outcome of data governance initiatives is behind us. Nowadays, organizations need functional artifacts that generate value at the pace required by markets, companies, and both data and functional areas. Therefore, rather than a challenge, I see it as a latent need, as we have evolved the way we deliver value. As an example, with a client, we took two domains, and in just four months, we managed to have control dashboards with: self-diagnostic results, domain map, glossary of terms, data catalog, and identification of data quality rules.
As a result of the persistent thinking of always being at the forefront of opportunities to add value with data, our #BigPeople managed to showcase, on the client’s messaging application, the first implementation of Open AI’s GPT services in Data Governance, responding in natural language to users’ questions associated with the aforementioned archetypes. 🔥
#5 Speaking the Same Language
All individuals within an organization should definitely speak the same language and master the same functional terminology. It’s very common to ask our clients if all areas understand the same things by terms such as “user,” “active product,” “direct service,” “derived service,” “profitability,” “gain,” “loss,” “inconsistency,” etc.
When I ask that, the response is usually affirmative. So, I ask them again: “Does this mean I can open the door and ask the first person I see about any of these terms, and they will respond the same way as any of you?” Here, they typically respond with a smile that can be evasive, uncomfortable, and, of course, in some cases, affirmative. We then conclude that the first thing we must do is standardize the language and terms to make them visible and internalized at all levels.
#6 Data Quality
What I recommend at this point is to delve deeper and move from the attractiveness of the term to its real application. A very relevant phrase that still holds true is:
“There is no governance without data quality, and likewise, there is no data quality without governance.”
I recall a project from 2016 in one of the largest companies in Colombia, where we initially assessed issues related to user information duplication. However, after a methodological exercise, we discovered the real pain point: user distrust and dissatisfaction due to the quality of contact data.
During the implementation, we collaborated with Product Owners to define three specific and concise business rules that ended up revealing that 65% of those considered users were actually duplicates or people who could not be contacted (due to incorrect physical addresses, phone numbers, and emails).
To our client’s surprise, we presented that out of 4.3 million presumed users, only 1.5 million were genuine and accurate because they met these three rules (which were actually quite basic). For those more technically inclined, consider the deletion of 206 tables and 179 million records that were eliminated.
Although some years have passed, I still use this example to make it clear that to generate ROI in data solutions, we don’t necessarily have to develop and incorporate the most complex machine learning models or pay for the most sophisticated artificial intelligence. It often involves activating our natural intelligence better to:
- Think in functional terms of the trust in the data we have.
- Analyze and seek to correct them from the root.
- Think about monitoring strategies so that, after rescue, they don’t degrade again.
***
From the days of 3.5″ floppy disks to current implementations of artificial intelligence, we have come a long way in data processing and usage. However, the most relevant issue we face today is not technological but cultural. Although companies have recognized the strategic value of data, the real challenge lies in making all parts of the organization understand and materialize the importance of turning data into products that generate sustainable competitive advantages.
Data Governance is about turning data into a strategic asset that creates value and competitive advantages. This not only involves implementing advanced technology but also instigating a cultural change throughout the organization.
In future articles, we will explore not only these pillars but also how experience leads us to revolutionize the vision of such initiatives today and how companies can move towards a more effective approach in data management. Stay tuned to discover how Data Governance can drive business success in the current era. 😉
