This piece is a community contribution from Tiankai Feng, a data leader, writer, TEDx speaker, musician, and meme creator, with a growing list of creative works in the data domain. Currently at Thoughtworks Europe, Tiankai has spent over 14 years shaping projects and strategies that drive data-driven initiatives. Known for breaking down the most complex data concepts into simple, often humorous insights on his LinkedIn, Tiankai makes the tough look easy. We highly appreciate his contribution and readiness to share his knowledge with MD101.
We actively collaborate with data experts to bring the best resources to a 7000+ strong community of data practitioners. If you have something to say on Modern Data practices & innovations, feel free to reach out! Note: All submissions are vetted for quality & relevance. We keep it information-first and do not support any promotions, paid or otherwise!
Putting People First: Communicate Communicate Communicate!
Data Governance is a people business. Despite the focus on policies, rules, and guidelines, it’s people’s mindset and behaviour that truly make Data Governance effective and valuable.
Only if people understand their responsibilities and what actions need to be taken, can Data Governance prevent and mitigate risks, as well as enable the generation of value with data through proper data management.
But with any action for improvement comes change, and the most important discipline in any kind of change management effort is always: Communication.
The Friction in Communication
One of the most common problems in data management, especially in large organizations, is the lack of transparency and communication between stakeholders along the data lifecycle.
When data producers, data processors, and data consumers are not talking to each other and do not provide any documentation for context, then bad and irreversibly damaging decisions can be made.
What We Can Do
We need to make communication an integral part of any Data Governance strategy or framework by including aspects like:
The main goal is to use communication to understand motivations, pain points and root causes - and to not just assume and thereby define solutions that do not solve the actual issues or potentially even make matters worse than before.
Communication can not just be an optional side task, it must be a prioritized mandatory task to ensure Data Governance can work - and everyone involved in Data Governance needs to act accordingly.
Roles & Responsibilities a Star Data Governance Team Should Have
Data Governance is always a cross-functional effort, so when we discuss the roles & responsibilities needed in a data governance team, we don’t mean “team” in a hierarchical unit sense but in a sense of cross-functional collaborators with a shared purpose and clearly defined contributions from everyone.
Think less Fantastic Four and more Avengers! Quite literally it requires people with different expertise and skills to make Data Governance successful.
Before we jump into the descriptions of the required roles, it should be noted that there is no one-size-fits-all approach to data governance, and that includes what roles should exist - so everything described next should be more inspirational, and it’s up to the reader to adapt those into the environment and context of your own organization.
So, what are the roles needed for successful Data Governance?
Data Governance Manager
A method & framework expert who can quickly connect the dots about what’s required for effective Data Governance, a master in Networking and managing Stakeholders, as well as facilitating cross-functional agreements and decisions related to data. This person oversees that Data Governance is set up right and that the right problems are being solved.
Data Quality Manager
A problem-solver with a deep passion for structurally identifying, mitigating and preventing data issues, and who can translate between business and technical requirements, as well as build a network with responsible stakeholders along the data lifecycle and value chain. This role should define, own and enforce the overall data quality management processes.
Data Discovery Manager
A metadata management expert who is passionate about describing, documenting and discovering data assets. This role should own the data catalog as a platform and should engage with all data owners, stewards, product owners and other responsible roles in populating, curating and sharing metadata in the catalog and beyond.
Data Owner
A business leader with a good understanding and passion for data, usually with higher seniority in an organization, is accountable for the overall quality and compliance of data in a specific data domain and has the final authority to make decisions about the data.
Data Steward
A business expert who acts as the right-hand person of the data owner, operationalizing decisions that are made by the data owner, and continuously overseeing and steering issue prevention and resolution over a given data domain.
Data Custodian
A technology expert who knows how data in a given domain flows and is managed across systems from a technology and architecture point of view and who steers issue prevention and resolution with technology stakeholders
Data Product Owner
A product manager who provides data assets to a defined group of users with specific value-driving use cases, usually ensuring specific Service Level Agreements (SLA), Service Level Objectives (SLO), all measured by Service Level indicators (SLI), usually a close collaboration with data consumers
Platform & System Owner
A system expert who is accountable for a specific system that creates or processes data, usually by acquiring and adapting an external proprietary system and, in many cases, operational systems where transactional data is created for the first time in a data lifecycle
It’s important to note that these are typical roles in data governance defined by a range of tasks, and not all of those roles need to be one full-time employee - depending on the size, maturity and industry type of your organization, some roles might be taken by the same person simultaneously, while others might not be necessary at all.
Initiation: Lay out a plan and find a champion to lead the governance program.
Every product/startup requires someone who champions its value and consistent sponsorship from the leadership. ~ Excerpt from “Functions as Startups”
Before diving into the technicalities, define the goals and scope of your governance program.
It’s crucial to identify a champion who not only understands the data landscape but also has the leadership skills to inspire and rally the organization around this initiative.
This Champion would later evolve into the role of a Data Product Manager and should possess a mix of technical acumen and the ability to communicate the value of data governance in business terms, bridging the gap between IT and the broader business units.
Foundation: Create a culture that breeds governance heroes.
People are at the forefront of enabling success for any and all initiatives. This makes culture, which is often overlooked, the prime puzzle piece that makes the big-picture goal much more achievable.
Create a culture that celebrates and supports each governance hero for the value they create. This means transparency in metrics that enable governance practitioners to be seen as impact-drivers who can affect the North Star Goals.
Define metrics relevant to the function's objectives and build experimental processes that help bring these goals closer. Assign owners to different initiatives. You can't grow or alue what you can't measure. ~ Excerpt from “Functions as Startups”
This also involves fostering cross-team collaboration, ensuring that members from different domains—be it finance, operations, marketing, or IT—are included in the governance team. This also plays out from a human angle where the team does not just enforce, but more importantly, is able to influence transparency in metrics and limelighting the true owners of initiatives.
Autonomy: Empowerment through Self-Sufficiency
Set up for success: decentralised decisioning and ability of the function to fulfil its goals with the right permutation of resources. This autonomy naturally ensures a high-speed culture, which gives experimental ground, speed in experiments, and eventually helps discover high-value processes.
Ensure that your governance team has the skills and confidence to succeed.
Fostering a continuous learning environment where team members can share knowledge and experiences will keep the team agile and adaptable.
The likelihood of the team’s ability to take up ownership is directly proportional to knowledge. Which again comes from experiencing projects on the field. Essentially, it also boils down to trust while assigning ownership and the initial leeway to make mistakes while acquiring experiences.
It’s essential to recognize that building a star data governance team is a consistent and continuous process. Regularly review and refine your strategies. By staying adaptable and committed to continuous improvement, your governance team will not only excel but also set the standard for data excellence across the organisation.
Where do these Governance Heroes Create an Impact in the Data Lifecycle?
Having a team of Governance Heroes is great - you found the right people, and they have the best skills, but how do they generate impact?
Far too often in Data Governance, the approach is to dogmatically set everything up first, then apply a comprehensive framework with policies on everything at once - this “boiling the ocean” approach just doesn’t work in today’s world.
So to really show impact from the get-go, Data Governance should start solving problems from the get-go - and a use case based approach can really help to do that, and simultaneously build up a data governance framework and operating model use case by use case.
So how should use cases be identified and prioritized? From a Data Governance point of view, a use case should be only classified as a “governance-ready” use case when two conditions are fulfilled:
There are clear business requirements behind the data issue. There should not only be a complaint of what is wrong, but a well-defined expression of how “correct” should look like.
There is a quantified business impact. Value of solving a data issue should always be quantified in one or more of the following value types:
increasing revenue,
decreasing costs,
or avoiding risk.
The requirements help the governance heroes know what needs to be done and how complex the job will be - and the business impact helps motivate the heroes to take action and prioritize use cases against each other.
Once there is a list and a sequence of use cases defined for this team of governance heroes, the real work begins, and it all starts with understanding where the specific issue came from - also known as a root cause analysis.
Leveraging the detective and analysis skills of these heroes, investigations will follow the path of data lineage to trace an issue back to the origin - and the more complicated the data landscape and infrastructure are, the more time this could take.
Which is why having roles such as Data Custodians or Data Stewards who are already knowledgeable and have the right network of people in the organization can really accelerate the process.
No matter where you finally identify the root cause error, there are three main types of errors that lead to different actions for solutions:
Human error - an error that was caused by a manual task by specific individuals, which has led to significant problems downstream
Process error - data processes and business processes not being synchronized or aligned, leading to a structural problem of data not being useable at given critical process steps during a business process
Technology error - an error that was caused by a bug, failure or missing reliability of a specific system, pipeline or other technological aspects
While a human error might mean more training and guidelines are needed, a process error might mean some more process alignment by senior leaders is needed, and a technology error might mean an adaptation of SLAs or development of some technological failsafes - every organization should proactively define what solutions are needed to what kind of errors, this is how the governance heroes can be the most effective.
Arming Your Heroes with the Right Weapons
Always view technology as an enabler instead of a director.
How Technology Impacts the Day-to-Day
Self-service data stacks are designed as perfect enablers - facilitating objectives and governance strategies instead of deciding them.
Traditionally, governance teams might have spent significant time managing data access, ensuring data quality, and responding to ad hoc requests. With the rise of self-service analytics, the nature of their work shifts from gatekeeping to strategic enablement and guidance.
Instead of being bogged down by operational tasks, they can focus on setting the guardrails—defining the policies, standards, and frameworks that empower users to access and use data responsibly.
Self-Service: What’s Considered Valuable by Governance Teams
Self-service should enable business users to explore data, generate insights, and make decisions without constant oversight. However, it must do so within a framework that ensures data consistency, quality, and security.
Governance teams find value in self-service tools that come with built-in compliance features, automated data lineage tracking, and role-based access controls. These features allow them to monitor usage patterns and ensure that data is being used appropriately.
For governance teams, the impact of these capabilities is significant: they can spend less time on routine data requests and more time on higher-value activities such as refining data governance policies, analyzing data usage trends, and identifying new opportunities for data-driven innovation.
Samples of how Self-Service Improves the Lives of Various Data Citizens
📝Resources If you’re building your own self-service layer or validating ready-made one’s for adoption, here’s a community-driven standard for Self-Serve Data Stacks you can use as a guideline: datadeveloperplatform.org
Align your Team’s Strength to Your Org’s Governance Roadmap
As we established, a Governance Roadmap should consist of building and rolling out Data Governance capabilities through use cases, solving one after another - following more of a “continuous delivery” approach rather than a “waterfall and milestones” approach.
Nonetheless, the skills and strengths of the Data Governance team should always be leveraged and further grown, specifically from the different aspects of the 5Cs of a Humanized Data Strategy (established in Tiankai’s book):
Competence
The world of data is constantly changing and Data Governance professionals need to continuously keep up with the newest trends, for example getting more upskilled in the mechanisms of AI and Machine Learning, the new trends in Data Engineering or following up with new regulations.
Collaboration
Data Governance only works through cross-functional collaboration, and with new use cases new collaborations are always needed, and the most effective is one rooted in trust and with co-creation with stakeholders in mind. Manage a stakeholder network proactively and with intention, so resistance can be addressed in a timely manner and advocacy can be leveraged to accelerate issue resolution.
Communication
Even with the best solutions to data problems, the value has to be communicated in the right way for it to get attention and buy-in from the wider organization. Having good communicators with tailored messages to different audiences is key.
Creativity
Creativity is in every human being and is the source of innovation, especially in Data Governance. Giving the governance team not only the space but also the encouragement to be creative and experiment with new ideas can lead to innovation that can take Data Governance to the next level.
Conscience
The more data is being used for advanced use cases, the less experience and lessons learned exist to deal with it - requiring a higher level of human oversight and applying more human critical thinking combined with a moral compass. The data governance team should apply their critical thinking consciously and should involve colleagues from other areas for morally ambiguous use cases (e.g. legal, ethics, compliance).
Keeping the skills and satisfaction of data governance professionals in mind will lead to the intrinsic motivation from the data governance heroes to keep improving and optimizing data governance organization-wide.
Best Practices to Help Your Heroes Get Better with Time!
Continuous Feedback and Learning
People: Engage in regular training and seek feedback from data users across the organization.
Process: Regularly review and refine governance practices and policies.
Technology: Implement strong feedback loops for governance models and policies by establishing a Product Lifecycle for Data with inherent feedback loops.
Collaboration and Knowledge-Sharing
People: Encourage collaboration across departments and with external data communities.
Process: Foster knowledge-sharing through workshops, forums, and retrospectives.
Technology: Leverage a semantic layer to standardize data definitions, facilitating collaboration through a common business language.
Automation and Self-Service
People: Free up team time by automating repetitive tasks, allowing focus on strategic activities.
Process: Implement proactive governance by automating data classification, quality checks, and compliance reporting.
Technology: Enable self-service of policy control, policy modeling, and quality & observability resources.
Leadership Support and Recognition
Ensure leadership actively supports the governance team by providing resources, recognizing achievements, and promoting a governance-first culture.
Thanks for reading Modern Data 101! Subscribe for free to receive new posts and support our work.
A Treat from Tiankai! 💝
Tiankai Feng has just launched his latest book, Humanizing Data Strategy.
In this book, Tiankai offers his unique perspective on the “human” side of data strategy. He shares expert insights and experiences on how organizations can maximize their data strategies by focusing on the often-overlooked human element.
Here’s your own copy of the Actionable Data Product Playbook. With over 500 downloads so far and quality feedback, we are thrilled with the response to this 6-week guide we’ve built with industry experts and practitioners. Stay tuned on moderndata101.com for more actionable resources from us!
I believe most of these roles can be replaced—and the work distributed across the organization—by implementing well-defined data contracts that are programmatically deployed, along with effective data modeling of operational and transactional systems, including master data management. In my view, everyone in an organization, from salespeople to AI engineers, should be responsible for maintaining data quality, rather than delegating most of the effort to a specific group. This approach is not only more cost-effective but has also proven to be more efficient, at least in my experience.
I believe most of these roles can be replaced—and the work distributed across the organization—by implementing well-defined data contracts that are programmatically deployed, along with effective data modeling of operational and transactional systems, including master data management. In my view, everyone in an organization, from salespeople to AI engineers, should be responsible for maintaining data quality, rather than delegating most of the effort to a specific group. This approach is not only more cost-effective but has also proven to be more efficient, at least in my experience.