The Covid-19 pandemic has meant that data-driven decisions have influenced all our lives over the last two years. But decisions made without proper data foundations, such as well constructed and updated data models, can lead to potentially disastrous results.
For example, the Imperial College London epidemiology data model was used by the UK Government in 2020 to justify lockdown policy decisions based on a forecast that 500,000 deaths would occur if no action was taken. But questions have been raised about these data modelling predictions and the need for such stringent lockdown measures in the early days of the pandemic.
The problem was not the data but how it was interpreted. The same issues occur for businesses.
All decisions are underpinned by data so it is important for the right data to be available to decision-makers and for that data to be high-quality and trustworthy. Data modeling enables this through a series of data models (conceptual, logical & physical) that start at a high-level, driven by the decision-maker’s business needs, and evolve into greater technical detail for how data is to be stored, organised and managed.
This creates a common language across an organisation which is the starting point for a “single source of truth” for data and the effective flow of data within an organisation.
Data modelling therefore:
- provides consistency in how to treat data across an organisation (which improves data quality), and
- unlocks the real value of an organisation’s data which exists in the relationships between different types of data. It assigns rules to identify those relationships, again, all based around the decision-maker’s business needs.
This consistency is essential for any data analytics, business intelligence or artificial intelligence application that supports an organisation’s business operations. Without it, an organisation’s data foundations will be fragile.
The detail behind data modeling is highly technical and complicated and it is recommended that organisations turn to subject matter experts that have a deep understanding of metadata (which sits at the heart of enterprise data management) and data modelling tools. For example, Envitia’s Data Modelling Toolkit has been used extensively by the UK Hydrographic Office to deliver a complex digital transformation project around maritime data. The benefits pay dividends once completed. This includes:
- Exploiting Data as the Most Valuable Asset in your Organisation: Data has significant value and, like any other asset, it needs to be managed, maintained, protected, and utilised to exploit that value. Too many organisations are sat on massive enterprise-wide data holdings simply not knowing where or how to start.
- Faster & Better Data-Driven Decisions: With a complete overview of an organisation’s data holdings, the data model maps how a business leader’s requirements are being fed by the right data (or not). Data can be found much more quickly but, conversely, it means that:
- Redundant or missing data can be identified reducing the decision errors based on missing data.
- Poor data quality is improved which means better decisions. Errors in decision making based on poor quality data is often a significant yet hidden cost in any organisation
- Underpin Digital Transformation with new Digital Business Models: With more flexible access to the reliable data, new business models can be developed as part of a digital transformation process.
- Providing a Common Language: In a world of fast-moving technology and blurred lines of responsibility between a Business Unit, CIO, CDO, & CTO, using data modelling as a tool to strengthen coordination and communication, with a common language and understanding, should not be underestimated.
- Creating a Data Driven Culture: Data modelling is not a one-off exercise but needs to be part of a data-driven culture. As the organisation evolves, so do business needs, and therefore so do the data models to ensure the data is organised (or re-organised) in such a way to continue to deliver those evolving needs.
This journey does not need to be a long or expensive process and the benefits can quickly outweigh costs.
This article was first published in The Data Administration Newsletter.