«Det at det er en team sport; det tror jeg er i hvert fall noe data verden kan lære av (software development). / The fact that it’s a team sport; that’s definitely something the data world can learn from software development.»

Software development is ahead of the data world in many ways, but what can we learn from its methods? Audun Fauchald Strand, Director of Platform and Infrastructure at NAV, shares insights from building autonomous teams and balancing freedom with governance.

Here are my key learnings:

What is a platform?

• A platform is something you build on top of. Something that eases building applications, because if a common ground layer.
• A platform reduces the need for competency, since it predefines some choices.
• The capacity for change is higher when building based on a common platform, then when buying applications.
• It’s important to find a balance in how much you should standardize through a platform and how much room for innovation should be beyond that.
• A platform can provide an «easy choice» that is the best practice way, but it should also allow for alternative, innovative ways to get jobs done.

Software Development and Data

• Software development has been a role model for data work, but they are at different stages of maturity.
• What is happening now in data platform is something Audun has experiences in Software 10 years ago.
• Yet, the principles are largely the same.
• Data is not as mature as software. There is still something to understand when it comes to use of tools and open source, etc.
• «Verden har blitt god å lage software. /The world has become real good at making software.»
• There is a certain agreement on what good looks like in software.
• Software has embraced open source in a much larger degree then data.

Data Mesh and the need for scale

• Data Mesh was a description if the data world from the perspective of a software developer.
• Change in architecture of operational environments have led to more possibilities to scale and to speed up operational data processing.
• This changes the need for speedy analytical data environments that have traditionally been built to compensated for the reduced flexibility of operational environments.
• Scaling up in a Data Mesh fashion, requires more data and analytical competency in teams.

Teams

• Working in teams, and having team ownership of products creates a certain sustainability over time.
• In software, with open source utilization, only teams can take technological decisions. Who else could decide what library to use for what purpose, then the team working hands on?
• You have to believe in autonomous teams for them to work. If not you will create dependencies that minimize their authority as a team and thereby capabilities to deliver.
• To create alignment there are some steps to take, from communicating openly, creating a tech radar to ensure alignment between team, to describing deliberate choices, rather then general principles.

5 learnings from article on alignment and autonomy:

1. Insource data expertise to ensure ownership, security, and business alignment.
2. Embed data professionals in cross-functional teams to enhance collaboration and generate actionable insights.
3. Balance team autonomy with governance by allowing flexible tool use within standardized data policies.
4. Foster continuous learning through asynchronous knowledge-sharing tools like wikis and Slack.
5. Provide clear direction with self-service data platforms to support scalability, efficiency, and governance.

«I consider this Data Governance as a cure. (…) Data Governance can make things better.»

In this clarifying conversation, Finnish data expert Säde Haveri shares her 18 years of experience and introduces a practical framework consisting of five key elements that can guide any organization's data governance journey.

Säde, who is a Data Governance Manager at Relax Solutions and co-founder of Helsinki Data Week, first explains the important difference between a framework and a playbook. While many consultants offer ready-made solutions, Säde argues that a truly effective framework functions more like scaffolding, helping organizations uncover their own best path forward.

We dive deep into the five elements: the choice between a top-down or bottom-up approach, the balance between defensive and offensive strategies, how to define the right scope, identifying key stakeholders, and the strategic role of external consultants. Säde illustrates how these decisions affect the structure, implementation, and success of data governance, with practical examples from his own experience.

Here are our hosts key takeaways:

• Data Governance is at the heart of the socio-technical system - it requires a variety of skills.
• The experience for the end user has not change much in the last almost 20 years.
• There is a need for «group support» for data people.

What is a framework?

• There are ambiguous connotations of the word «framework».
• A framework is not a playbook.
• A framework describes the what, not the how. You need two adjust it to your reality.
• Think of a framework as non-prescriptive.
• Frameworks are related to best practices, but they are not the same thing.
• Use it to identify your strengths and build your data governance practices around those.

Top-down or bottom-up

• Can be a management awakening (e.g. GDPR), or a need for better data at a practitioner level that initiates the need for data governance.
• Top-down often materializes in conceptual approaches.
• You start at a conceptual level, you will create data governance roles around these conceptual entities.
• From a bottom-up perspective you are building governance around your tables or datasets.
• As a middle way, you can focus on data products as objects to build governance around.

Aligning strategy defensively or offensively

• Defensive vs. Offensive strategy - based on an article from Davenport 2017.
• There is no one-size fits all.
• You need to understand your motivation? Is it build due to risk mitigation needs or for business value creation?
• You need to understand your sector-driven differences.
• Look at this as a spectrum, where your approach can differ between offensive and defensive based on the criticality of the dat for the use case you are working with.
• You always have to show your value, the value needs to be measurable.
• AI ready data can be both offensive and defensive.

Identifying Scope & key stakeholders

• Identify your stakeholders and scope based on the strategic alignment on offensive vs. defensive.
• Use business stakeholders rather than IT, to gain a better understanding of the underlying problem.
• Data Governance is rather a business value enabler than a cost saving activity.

Determining the role of external consultants.

• You have to sell your solution. Selling is something you cannot outsource.
• If you are looking at tooling, consider if you can find consultants with the right knowledge and capabilities.
• Try to understand what experiences the consultants can bring to the table.
• Ensure that you are aligned on a methodological basis.

«Sometimes it feels like you have CIOs going on Julia Robe’s in Pretty Woman spending sprees.»

Have you ever wondered why data governance often becomes so complicated that no one really understands what it’s about?
In this episode, we take a refreshing deep dive with data expert Rasmus Bang, who shows how to make data governance simple and relevant.

We explore the delicate art of engaging middle management, which is often the key to successful implementation of data governance. Through data governance committees and a focus on concrete business challenges, you can create both transparency and accountability that drive real change. Rasmus also explains how to bridge the gap between process excellence and data governance—how these disciplines can reinforce each other rather than compete.

Here are our hosts' key takeaways:

• The level of regulation in your industry determines your approach to. Data governance.
• Spent time in exploring what is necessary and relevant from a business point of view.
• Identify pinpoints you have today, to tackle them beyond compliance. This is where you show your value to the business.
• Storytelling is essential in data governance.
• Be able to see and communicate how your work impacts the business.
• Data governance is a peoples game.

How to start your initiative?

• Do analysis to translate your data pain points to business pain points.
• Assess the size and complexity of the challenge ahead.
• Adjust your lingo to make it understandable for the business.
• Dont be dogmatic - use existing structures.
• You need to adapt to your environment instead of hoping that your environment adapts to you.
• Make sure you find a strategy to talk to the conflicting priorities of middle management.
• Watch out for the experienced professionals- they can become bottlenecks.
• Make the value proposition understandable for everyone, also talking to these priorities on middle management.
• How can my department become better, more efficient, etc.?
• Include middle management representatives activity around the data itself.

Process Excellence

• Processes are seen as a precondition, while they are actually a result of the organizations culture, organization and people.
• Process owners often have a broad network in the organization and are therefore important to make data governance work.
• «If you do data governance right you get de facto business process ownership.» - you get the right people in place.
• Processes can give you a repeatable structure that is foundational for your success in e.g. AI.

• Have an approach and build a model that your business stakeholders can recognize and see themselves in.
• Dont focus too much on theory. Focus on what matters for your business and for your stakeholders. It needs to resonate.
• Dont try to fix a complex problem with simple solutions.
• «There is a difference between quick fixes and fixing small things.»