In my career, which has spanned both sides of buying and selling IT, one thing hasn’t changed: Data people and business process people don’t talk to each other. Business process people assume that data simply exists and care little about how it is “managed”, while data people focus more on the bits and bytes stored in repositories than how the bits and bytes are created and consumed. With data and process sitting on different layers on enterprise architecture diagrams, the prevailing view is that everything will work as long as the interfaces are well defined.
In reality, data and business process are far more intertwined. In today’s service-dominated economy, most business processes operate on data rather than physical goods: They consume data, perform a set of tasks, and produce data that feed into the next set of processes downstream. In other words, data is consumed by business processes as input, and data is produced by business processes as output.
These relationships lead to a couple of important observations:
- Business processes should be responsible for the quality of the data they produce. Ironically, data people don’t produce data and shouldn’t be held responsible! The right solution to chronic data quality problems is to either modify the business processes that produce the data, or put additional process controls in place to improve the quality of the data output.
- Business processes suffer a great deal from bad data. Wrong or missing input data leads to rework and manual intervention, ultimately driving up cost per transaction. Business efficiency depends on good data.
To improve efficiency, every core business process should have a codified set of policies that define the requirements for input data as well as accountability for output data. Getting these policies defined and communicated is the first missing link between data and business processes.
In addition to the producer and consumer relationships described above, data and business processes are linked in another important way. The act of managing data assets needs to be instituted as a set of business processes in their own right. As an analogy, managing human resources, another key enterprise-level asset, requires a strong set of processes. Data assets should be treated with the same rigor. When a piece of data fails to meet validation rules, who should be notified? What are the steps for correcting it? What’s the escalation path? These tasks have to be automated using workflow that enables a broad range of business people to actively manage data assets.
These missing links, data policies and a new set of business processes for data management, are what data governance is all about. In all the chatter on data governance, business process has been sorely missing. Too many vendors talk about data governance simply as another use case for existing IT-centric tools without addressing the links between data and business processes. A true data governance solution needs to bridge these two realms that have been apart for too long. In my next blog, I’ll explore how to do this in practice.
What do you think? Do you agree?