BlogikirjoitusKirjoittaja: Iiris Lahti
Data governance unlocks the value of data
"Imagine a situation where your employer does not have an idea how the data flows in the organization, how it ends up to reports or feeding algorithms, if the data can be trusted or in case of issues, from whom to ask. Now imagine a situation where your employer has clear data ownership and contact persons, a data catalog for searching data assets, data quality monitoring that helps to identify potential data issues and clear guidelines for responsible, ethical and sustainable data usage. Now consider which one of these companies would you want to work for?"
We’ve seen numerous waves of data hype in businesses and are reaching a point where data is produced from numerous sources in our organizations. Conventional wisdom states that data in itself is worthless – but how do we wield data to our advantage?
There is no silver bullet for capitalizing on the value of data, and the issue needs to be addressed at the systemic level. Many organizations need data governance, whether they realize it or not.
Here’s how to build working data governance in organisations.
Why data governance?
As data becomes ubiquitous, we need processes for making data work for our business goals. Data governance takes us beyond just producing data – it enables organizations to make data a success factor for them.
Fundamentally, data governance is about assigning the roles and responsibilities, building the tools and defining the processes for data in an organization's. It enables organizations to build an unified set of data that can be utilized in various parts and functions of the organization.
Data governance issues
Typical issues with data governance in organizations is the lack of responsibility and roles relating to data ownership.
When you’re building data governance, the first question is: "Is the ownership of data clear?"
Like any other aspect of your business, data needs to be owned and managed. This happens at two levels:
- Data engineers are responsible for collecting, processing and making data available, in a technical sense. Business stakeholders need to understand the way data is collected and used in the organisation, and use it as a tool in their work and decision making.
- Data strategists bridge the gap between IT and business. They can increase understanding and competence, fostering ownership of data and building workable processes.
Quality of data
The second big question of data governance is: “What’s the quality of your data?”
’I don’t know’ is the worst possible answer. It means that governance practices and ownership of data are not clear, or that the tools to manage data are not sufficient.
Data catalogs and data mesh platforms are often seen as solutions to various data challenges, but but either is a poor stand-in for proper data governance. Especially data mesh – the concept of the data platform as a distributed network of interlinked platforms specific to various business functions – is on the minds of many organizations looking to boost their use of data.
’I don’t know’ is the worst possible answer. It means that governance practices and ownership of data are not clear, or that the tools to manage data are not sufficient.
Data catalogs and data mesh platforms are often seen as solutions to various data challenges, but but either is a poor stand-in for proper data governance. Especially data mesh – the concept of the data platform as a distributed network of interlinked platforms specific to various business functions – is on the minds of many organizations looking to boost their use of data.
As a rule of thumb: "The more distributed your data is, the stronger the governance models need to be."
Proper data governance should be seen as a prerequisite for building a data mesh or data catalog and utilizing them properly.
How to build data governance
Building functional data governance is practical work. Each company is unique and needs an approach that fits their needs.
The unifying feature in successful data governance initiatives is that the change has been cultural, and fundamental. The ubiquity of data means that data governance can’t simply be tasked to IT – it needs business ownership, as well, and ownership and responsibility over data and its governance needs to be there throughout the organisation. This is the way to make data work for your organisation and reaping the benefits.
The process is typically as follows:
1. Model data flows
- This is Ground Zero. your organization needs understand how data flows through your organization and who own it. In the best scenario, company as a Data Catalog tool that support automated or assisted data lineage visualizations. However, organizations should not limit themselves in case a proper Catalog is missing – this can be done in any documentation tool, as long as it increases the user understanding of the data flows and origins.
2. Clarify responsibilities
- Many organizations have given the ball to IT when it comes to data, but it needs to be a shared responsibility. Data strategists and data engineers help organizations build data governance, but assigning responsibility to both business and technical personnel is key. Many organizations name data stewards – specialists who know where data originates from and how business specialists can use it in a responsible way. Often data is also governed by data domain owners. For instance, the sales director might be the data domain owner for sales data – they ensure data is produced and utilized in sales operations.
3. Build the capabilities
- In addition to ownership, data governance requires the competence to gather, sort, analyse and draw conclusions from data. This might mean educating people, building collaboration models and building or fixing tools for data management.
4. Clarify development roadmap
- A clear development roadmap is key. This means that an experienced data strategist builds a roadmap and development plan for improving data quality. This is then put into practice by data engineers and developers.
5. Build data practices
- Active management and documentation of data is key in a data platform, and automation helps ensure effective handling of data. Strong DevOps practices concerning data help make data an effective tool for your organisation.