You know that multi-year project you’ve been working on trying to consolidate all your organization’s data into a central hub to establish a single version of the truth? Yeah, you can stop now. It’s not going to happen.

In point of fact, there is no such thing as a single version of the truth in today’s post-modern world, and the advent of Big Data is only reinforcing this notion. As Frank Buytendijk, who teaches and writes about performance management and organizational behavior, put it during his keynote address at TDWI World Conference this morning, “We are further away from the one version of the truth than we have ever been.”

Rather, truth is relative and is dependent on perception. Consider a simple example, like the definition of revenue. Depending on whom you ask in any given organization, revenue might mean gross revenue, booked revenue, net revenue or something else entirely.

Now, the instinct of most BI professionals (and the goal of most BI tools) is to reconcile these competing views into a single report or view of “the truth” and then impose that definition onto the rest of the business. But this, as Buytendijk points out, isn’t how businesspeople work.

Frank Buytendijk, CMO at Be Informed, Teacher, Writer

Businesspeople adopt as “truth” those data definitions and concepts that allow them to do their jobs most effectively. They base what is true on their confidence in the data sources and on real-world experiences, and then make judgments and decisions based on those perceptions. It’s not a perfect scenario, but then in life nothing’s perfect, including (or maybe especially) your centralized enterprise data warehouse.

It is time for IT (and vendors) to catch up with their business colleagues and stop trying to force a hierarchical, one-size-fits-all approach to data management and analysis on the enterprise. Instead, as Buytendijk points out in this excellent blog post, IT should take a process-oriented approach, recognizing that there are multiple versions of the truth throughout the information lifecycle and each is potentially valid in its own right.

Donald Farmer, VP of Product Management at QlikView, offered me a simple but very effective illustration. After the devastating earthquake and tsunami in Japan last year, QlikView customer Este Lauder performed some analysis to determine how the natural disaster was impacting sales of its products in that country. For this exercise, the company defined Japan as a market.

During the analysis, however, they also found that an upcoming, highly advertised product launch in Australia was in jeopardy due to the earthquake and tsunami because certain raw materials needed from Japan were now unavailable. With this insight in hand, the company took steps to delay the product launch. Here, Japan was defined as a supplier.

So which is it? Is Japan a market or a supplier? Of course it is both! Luckily for Este Lauder, it had both the tools in place to make this discovery and the mindset in place to accept both views of the data as equally valid – or equally “true” — for their own purposes. Most enterprises aren’t so lucky, but they need to adapt.

When you apply the construct of the above simple example to a more complex scenario, you can see how it isn’t realistic to boil down all your data into a single view of the truth. When you begin taking social media data, machine-generated data and other high volume/high velocity data into account, the concept of a single, centralized data temple is quaint.

Further, multiple views of the truth can actually foster dialogue and can produce even more important insights when compared and analyzed against one another. Instead of settling on an imperfect but single view of the truth and presenting it in a report that the business ignores, Buytendijk  argues, IT should create rich reports and other tools that present multiple views of the same data, side-by-side, in a single interface to encourage further data exploration.

ServicesANGLE

All of this is not to say that anybody within your organization should have carte blanche to determine his or her own data definitions. Of course sound data quality procedures and data governance frameworks should be implemented to ensure everyone is playing by similar sets of rules. But IT should give up on building a holy data temple to house a single view of truth and instead provide the business flexible, agile services and tools for comparing and analyzing differing points of view.