The Information Technology (IT) pendulum is turning from technology towards information, and data management is one of the 7 key ingredients of digital business that I focused on in my previous blogpost. It is no coincidence that these map to the phases of enterprise architecture frameworks, such as the TOGAF architecture delivery model’s: vision, business, information, technology and governance. These are relatively static, the difference is how we apply their contents and execute to reach “target state”; which in our time is digital business.
Without all the pillars, the architecture and vision falls, so there is really no way of promoting one as more important than the other – that being said, data management is really important.
When talking about data, briefly consider that the word ‘data’ is often semantically interchanged with the word ‘information’. This is at best imprecise, but for the purpose of this post whichever of the two that you prefer will suffice. For the record though, and for the theme of this blogpost being data and semantics, I’d like to state that information really is data put into context. In the sense that data has potential value, and information has concrete value. For simplicity I will continue to refer to data, but keep in mind how value is derived from data:
We have all said, heard and know that data is an asset – always has been, but now perhaps more than ever before. If we agree to this reflection, the consequence is a necessity to manage data as we do financial and physical assets, in order to retrieve and secure its value. This does not happen by chance, or as a side effect of something else we prioritize – it will only come as the result of enterprise commitment and leadership. Together with an agile and practical approach, continuous ROI is feasible, but acknowledge that it is a demanding undertaking that will require time and resources to get right.
To give an impression of the potential magnitude of a data management program, here are some of the central capability areas involved:
Some are closer to strategy and process, others to technology – but all together important to get full value from data assets.
The high level reasoning for the importance of data management, as these knowledge areas implies, is to enable the enterprise to accurately communicate across the value chain. Executing business processes or making business decisions based on low quality or even faulty data, involves business risk and loss. Likewise access to high quality correct data, that is seamlessly integrated across the complete value chain, enables effective innovation and digital business opportunities.
Capabilities also imply activities, and they are continuous efforts – circular processes. Initially you will seek to mature to a level of productivity, but the result must be governed to upheld quality. As such, a starting point is to decide and define the policies of data governance. Basic policies should be the corporate glossary, that semantically unites the enterprise entities in an agreed business terminology. I.e everyone and everything involved interprets data the same way. The naming conventions, that makes sure the enterprise data dictionary is implemented equally over time. The information security policies, that ensures data is handled according to sensitivity, confidentiality and proper authorization. And as you come full circled with a produced outcome of data management, the established policies and metadata must be applied as part of operations change management process – upheld by mandated roles and forums specific for data governance. Top down center of excellence is a recommended model.
Once the data governance policies are in place, the data model that they will govern can be designed. Data modeling is typically done using an ER or UML design tool. The first level of data modelling is conceptual, and only models the top level business entities, and the relationship between them. Business understands and owns this model. The modelling is typically done per business domain (domain model), and the complete result is the enterprise data model. Domain Driven Design (DDD) is an effective methodology to accomplish this task.
The next level is the logical data model, where all relevant details (attributes) of the data entities are included together with technical metadata. The purpose of the logical model is to specify how the data model should be implemented, but still technology-agnostic. Finally you have the physical data model which contains all the technical details for a specific database system.
As you can begin to appreciate, data management is a potentially large topic when done correctly, and still only one of seven areas in focus for digital business. Even so, it is possible to get traction when approached correctly with agile methodology and a practical, iterative processes. But a clear vision, strategy and executive management ownership, is a crucial success factor. A classic pitfall is to establish a huge waterfall project for data management, or even worse – seven silo projects; one for each pillar of digital business. Some planning and preparations are of course inevitable, but ensure to align initiatives as an enterprise program effort, where early and continuous operational results is a clear and measurable requirement.
Coming back to digital business and the digital transformation that data management is part of, it is hopefully more clear now how this goal only can be achieved when data accurately describes the entities of business process, the interpretation of data is exact in the context of a given domain, the integrity and quality of data is guaranteed and trusted, and the security protects this valuable asset. Or in the opposite case, how the target state of digital business cannot be reached without proper focus and implementation of data management.