From silos to synergy: rethinking data ownership in banks

Welcome back to the tech blog. This time I will look into one of a bank’s greatest competitive advantages – its ability to manage one of its most valuable assets: information, or in other words, data. How can this be achieved in the most effective way within a large company such as a bank.

There is probably not just one way to do this, but there are several central elements that most companies consider essential in managing their information assets. One of these central elements is the approach you take to governing your data – that is, the operating model you adopt to define ownership of the bank’s various information assets, as well as the responsibilities assigned to each respective data owner. In this blog, you will read about several models for organising data and data ownership, along with some advantages and disadvantages associated with each approach.

Domain-driven data ownership

This model is based on each business or operational area owning its data, and a data owner is appointed within each domain - for example, Sales, Finance and HR. The data owner’s responsibility is mainly data quality management, handling definitions and concepts within the defined domain, as well as responsibility for classifying, making data available, and defining the use of data within the domain. The model provides clear business responsibility and proximity to the significance of data in the business. However, the disadvantage is an increased risk of silos and more difficulty coordinating between departments.

Centralised data ownership

In a central data ownership model, a central unit such as Tech, Chief Data Officer or Data Office owns all data, and the business is to be seen as a user rather than an owner of the data. In this model, the data ownership responsibilities include standards, operational responsibility for data quality, data platforms, and access to data. The advantages of such a model are a greater degree of unified governance and strong central control. The disadvantage is less business anchoring and risk of bottlenecks.

Federated Data Ownership

In a federated data model, responsibility for data management is shared between the business and a central data function. This model is widely used in modern data organisations. A typical setup is that the business owns its data – or, more precisely, the data created within that area of the business. The ownership includes defining what the data means, responsibility for data quality management, and how the data is used. The central function, typically a central data office, owns central instructions and standards specifying how data is stored, protected, distributed, as well as rules for documenting lineage and common routines for data quality management. The advantage of this model is that it provides a strong balance between control and flexibility, and it scales effectively in larger organisations. However, for the model to function well, roles and responsibilities must be clearly defined.

Reflection

There is no universal answer as to which model works best in practice. Every organisation needs to consider its unique circumstances before choosing how to organise data ownership. Factors such as the business operating model, the company culture and the size of the organisation affect which model is most suitable. What can be said, however, is that banks often choose a federated model where the business owns the data content and central functions govern standards, data quality procedures, and sets the compliance related internal rules.

This approach is common in organisations like SEB. The background is that banks are relatively heavily regulated companies with extensive data management requirements on business processes, confidentiality -and risk management derived from regulations such as BCBS 239, GDPR, and regulatory requirements related to risk management and financial crime prevention.

Ultimately, the most important focus should be to work with data ownership responsibilities consistently and persistently over time to ensure its implementation. To reach high level of Data Management maturity may take time in a large organisation. However, banks that succeed with this will build a strong competitive advantage in a world dominated by players fully utilizing automation and Artificial Intelligence.

Öystein Graasvoll, 
Head of Group Data Management