Data Domains & AI Adoption: A Match Made in Heaven

Introduction:

Data is the lifeblood of any modern enterprise organisation. It flows through each and every business across the world and has by its very existence, become a modern day commodity. Indeed, in 2022 the World created close to 92 zettabytes of data. That suffix certainly wasn’t taught when I was saving homework to a 100 kilobyte disk back in the… ahem, 90’s. ??

However, as organisations amass increasing amounts of data, they also face new challenges around data ownership and management. These ownership challenges are exacerbated by the rise of digital transformation, cloud computing, and the growth of data-driven business models. In years gone by, I often commented about “born in the cloud businesses” or “cloud native organisations”. However, with the advent of artificial intelligence (AI), there are businesses and disruptors who are by all intents and purposes “AI-Ready” organisations.

As a result, enterprise organisations are looking for ways to utilise AI to make money, save money and reduce risk. However, with data sprawling across private and public cloud infrastructures, business leaders must get a better handle of their data to ensure that it is of a sufficient quality and integrity to support AI driven use cases.?

In this context, data domains have emerged as a foundational solution to the problem of data ownership in the enterprise, to support the adoption of AI in a safe and secure manner. In this blog, I will cover what data domains are, their key characteristics, the roles of data owners when defining domains and some simple steps to follow when defining them in your business.?

What are data domains??

Don’t worry, we won’t be going to another galaxy with this definition! Though, the very word domain does invoke memories of Star Trek…

In simple terms, data domains are logical groupings of data. Data domains are all about ownership, governance and responsibility. In respect of good governance, data domains refer to a specific area of responsibility within an organisation for managing data assets.?

Typically, domains are defined by a specific subject matter, business function or service/product for which data is collected, stored, and used. By defining data domains, organisations can establish clear ownership, accountability, and control over the management of data assets and ensure that they align with the organisation's data strategy, business objectives and regulatory commitments. This is crucial when considering ethical obligations around the use of AI. Being able to trust where data has come from and prove its origin is crucial, particularly for regulated businesses.?

Upon defining their respective data into domains, organisations can provide a clear and structured approach to managing data quality, security, privacy, and compliance. Ultimately, this aids to remove any question marks and ambiguity about who actually owns the data and what they are responsible for; in terms of maintaining data quality and accessibility to other domains across the organisation.

The role of Data Owners in a domain driven model

Within any given domain, data owners play a crucial role in controlling the use, storage, and maintenance of their data-sets and data products. They are responsible for ensuring the accuracy, quality and security of their data and should be made responsible for deciding who and how to access the data under their control. In simpler terms, a data owner is the person or entity that holds the rights to a set of data and is responsible for managing and protecting it.

Indeed, a single data domain can consist of many different data sources, data products and have many data owners. That aside, there are 7 key responsibilities that data owners should be accountable for when managing and protecting the data in their domain. Namely:

• Data Governance: Ensuring that data is managed and used in a responsible and compliant manner, which can be difficult to achieve without clear ownership and accountability for data within the organisation. This is often done in collaboration with a centralised data governance function and is ideally managed in a way where governance controls are automated and audited.?

• Data Privacy: Protecting sensitive data from unauthorised access, use, and breaches, which if not, can have serious financial and reputational consequences for the organisation.

• Data Quality: Maintaining the accuracy, completeness, and consistency of data, which is critical for data driven decision making.?

• Data Integration: Integrating data from multiple sources, systems, and departments, which can lead to inconsistencies and errors without clear ownership and standards.

• Data Security: Securing data from internal and external threats, including cyber-attacks and data breaches, which can have devastating effects on the organisation’s reputation and bottom line.

• Data Monetisation: Monetising data through licensing, data products, and services, which requires clear ownership and control over the data.

• Data Sharing & Accessibility: Sharing data with other domains and data teams, partners, customers, and regulators, which requires clear ownership and data sharing agreements. This typically involves making data available via discovery tooling like a data catalogue. Equally, using API’s to connect to specific source systems or domain aligned data products is becoming an increasingly common requirement for data owners.?

What are the key characteristics of a well defined data domain?

A well-defined, understood and governed data domain typically exhibits the following characteristics:

• Ownership: The data domain should have a designated owner who is responsible for its governance, maintenance, and protection. Indeed, there could be many data owners within a single domain. Though, ideally, there should always be a single figurehead who is accountable for cross data product/source governance within their respective domain.?

• Clear scope: The scope and boundaries of the data domain should be clearly defined and limited to a specific set of data source’s, elements and their supporting attributes.

• Relevance: The data elements within the domain should be relevant to the organisation's business needs and goals.

• Accuracy: The data within the domain should be accurate and up-to-date to ensure its usefulness and reliability for data driven decision making.

• Accessibility: Access to the data within the domain should be controlled and restricted to authorised and approved personnel only.

• Consistency: The data within the domain should be consistent and follow established standards, rules, and policies.

• Transparency: The rules and processes for managing the data within the domain should be transparent and clearly communicated to all stakeholders.

• Compliance: The data domain should comply with relevant regulations, laws, and industry standards.

• Security: The data within the domain should be secure and protected against unauthorised access, use, and alteration.

• Monitoring: The data domain should be monitored regularly to detect and prevent any issues or violations of governance policies.

Establishing these fundamental characteristics and principles can certainly help organisations to better manage their data, which in turn unlocks opportunities to safely explore AI use cases. This is especially so in businesses that are experiencing data sprawl across hybrid cloud estates. Based on personal experience, once your domains are defined the most crucial step is assigning ownership to them and perhaps most importantly making sure that your data owners understand what is expected of them in their roles.?

Simple steps to follow when defining data domains

Needless to say, if you ask 5 people to define their views of data domains within your business, you will likely get 10 different perspectives!?

As such, I’ve outlined some simple steps to follow when trying to determine the data domains within your business:?

• Define data objectives: Start by defining the objectives of each domain including the data elements/sources that need to be governed, the intended stakeholders, and the desired outcomes that each domain serves (e.g., customer fulfilment, order processing, customer complaints)

• Outline data domains: Determine the data domains by defining the specific data elements and their attributes that will be included, and group these domains into logical clusters. I typically find this is easiest done based on the products, services or functions within a specific business unit (e.g., Customer Onboarding, Sales, Marketing, Regulation).?

• Assess domain relevance: Evaluate the relevance of each data domain to the organisation's business needs and goals.

• Determine domain ownership: Assign ownership for each data domain to a specific person or team who will be responsible for its governance, maintenance, and protection.

• Define rules and standards: Establish rules, standards, and policies for managing the data within each domain, such as data accuracy, consistency, and security.

• Implement access controls: Implement access controls to ensure that only authorised personnel have access to the data within each domain.

• Document processes: Document the processes and procedures for managing the data within each domain, including data input, storage, retrieval, and deletion.

• Communicate policies: Communicate the data governance policies and procedures to all stakeholders, including data owners, data stewards, and data users.

• Monitor compliance: Monitor the data domains regularly to ensure compliance with established rules/policies and to detect/prevent any issues or violations.

• Review and update: Regularly review and update the data domains and ownership as needed to ensure their continued relevance, accuracy, and effectiveness.

In Conclusion:?

Over the course of this blog, I’ve tried to demonstrate that data domains can vary greatly in scope and complexity, but the underlying principles remain the same: to manage and protect data in a way that supports the organisation's goals and values. Indeed, getting the boundaries and ownership of your organisation's data domains clearly defined and understood can be an enabler to increase data quality and accessibility.?

This in turn can support the future adoption of AI enabled capabilities in an enterprise organisation, by ensuring that solid foundations are in place to support the traceability of data’s origin. Which in turn prevents “garbage in, garbage out” syndrome when trying to introduce AI in an enterprise context. Arguably, data domains are a match made in heaven when it comes to AI adoption.

I hope you’ve enjoyed the read and thanks for taking the time to listen to some of my perspectives in this regard.?