Who would argue against the importance of having a ‘single version of the truth’? Certainly not the IT vendors and service providers who make their living out of the careful, laborious work of integrating data from business systems and databases. Surely only someone who was reckless with their company’s future, or brave with their career, would be so foolish.
But like many things that sound quite reasonable (“no one ever got fired for buying IBM”), there’s a dark side to the truism of single version of the truth.
That’s because when this very reasonable sounding goal is being driven by IT-led initiatives on data quality, governance and master data management, there is often a huge amount of money and effort being wasted on the pursuit of a journey with no hope of completion, and little worth seeing on the way.
For some, this might sound an extreme view. In fact, it’s long been supported by industry experts – and acted upon in businesses across the world.
Leading industry commentators such as Frank Buytendijk (see e.g.‘The Myth of the One Version of the Truth’) point out that any real business has many legitimate views on what, for example, it means to ‘keep a customer’, or ‘earn revenue’. The salesperson and the bookkeeper, for instance, might have separate – but valid – viewpoints.
The answer to the confusion that this causes, Buytendijk would say, is not to force pre-determined, rigid, IT-inspired governed hierarchies of data onto the business, but to put these multiple interpretations into the context of the parts of the business operation that they relate to. This offers a quicker and more effective route to a better functioning enterprise, which can then be clear about how and why it is trying to achieve certain outcomes.
At the level of the data industry as a whole, huge numbers of people have been inspired to try empowering technologies such as self-service business intelligence (“I can take charge of presenting and analysing my own data”) or open-source, file-based alternatives to data warehousing (“I can fill up a ‘data lake’ without having to think about what this all means at all!”). These movements are arguably a response to the same issue of needing to be free from having to agree a pre-determined single version of the truth. However, there is also an increasing awareness that without the business context Buytendijk highlights, these contribute to, rather than decrease, the confusion see e.g. Gartner’s Merv Adrian who believes ‘through 2018, 90% of data lakes will be useless’.
If the goal is to improve enterprise effectiveness, there might seem to be a stark choice between rigid, time-consuming, expensive data governance, and quick to access, free-flowing but ultimately expensive self-service and data lake technologies.
Taking a steer from Buytendijk, the answer has to be technologies that can be explicit about business context. Industry analysts are predicting that the next wave of data and analytics technologies will be dominated by artificial intelligence – technologies that can identify patterns, reason about trends and provide guidance. These are inspired by data science rather than data reporting, and have the key feature that they function exactly by putting data into some kind of meaningful model – a kind of business context. The issue with these technologies currently is that they are heavy on the data science, appropriate for specific applications or projects, and must be implemented and integrated by highly skilled people.
The challenge is how to make this approach work at an enterprise scale, so that it can be used to prioritise data governance activities, is driven by and offers value to people across the business and ultimately lets an enterprise be clear about the how and why concerning the achievement of outcomes.
So there is a nuance to the concept of single version of the truth. This is not simply about integrating data, using data or semantic models as a point of reference. It’s about enabling business users to understand each other’s versions of the truth by putting that data into a business context. And that, surely, is a journey worth starting.