There is a pervasive belief that a single version of truth–eliminating data siloes by consolidating all enterprise data in a consistent, non-redundant form – remains the technology-equivalent to the Holy Grail. And, the advent of big data is making it even harder to realize. However, even though SVOT is difficult or impossible to achieve today, beginning the journey is still a worthwhile business goal.

The road to SVOT is paved with very good intentions. SVOT has provided the major justification over the past 20 years for building enterprise data warehouses, and billions of dollars have been spent on relational databases, ETL tools and BI technologies. Millions of resource hours have been expended in construction and maintenance of these platforms, yet no organization is able to achieve SVOT on a sustained basis. Why? Because new data sources, either sanctioned or rogue, are continually being introduced, and existing data is subject to decay of quality over time. As much as 25 percent of customer demographic data, including name, address, contact info, and marital status changes every year. Also, today’s data is more dispersed and distributed and even “bigger” (volume, variety, velocity) than it has ever been.

SVOT: The Road Less Taken in Health Care Data Management
Sustaining a single consistent, non-redundant data

A single version of truth is a worthwhile but lofty goal. This is true in every industry, especially in health care as massive changes are being dictated by the new federal Affordable Healthcare Act (ACA). Cost-containment and improved outcomes are key objectives of ACA, and patient-centric data will be just one aspect necessary to achieve both. For example, primary care physicians who have implemented electronic medical records and share data with hospitals and other physicians will no longer need to order expensive lab panels for patients who already had them done. Clinical encounter data from providers and payer claims data has historically been siloed, but there are initiatives now to consolidate some of that information to provide a holistic view of patient encounters over time. While big data technology provides the ability to digitize unstructured text, images, audio and medical device readings, it represents a silo that will be challenging to consolidate with existing structured data. The number of n
ew data sources required to integrate and maintain an SVOT is daunting.

People, Process and Technology
Considering the roles of data governance and master data management

The number of master data management and data governance initiatives have also escalated demand for applying SVOT as a necessary business mandate, especially after an EDW implementation. The typical response to the continuing business requirement for an enterprise SVOT is to propose a “project,” a discrete outcome that is usually led by IT. Projects require a business case for funding and staffing, which is usually approved by management riding the wave of high expectations of the business.

Multiple articles, conference presentations, blogs and Ph.D. theses have stressed the importance of considering all of “people, process and technology” for successful MDM and DG initiatives. But all too often, people and process are afterthoughts, ignored in favor of technology. This is common because people and process are the hardest to measure and get right.

For MDM, the project scope is defined by subject area, typically customer and product, with customer being the most challenging due to myriad data sources. The landscape is littered with a large number of unfinished or poorly implemented MDM projects. In many cases, an MDM project is spearheaded by an IT group that believes MDM is no more than a surrogate key and a set of ETL code. The outcome of many MDM projects is nothing more than a customer data mart. Perhaps establishing transactional match engines, which ensure that data integration occurs consistently across platforms, is too hard to implement, not interesting enough, or suffering from insufficient time and/or money to do it right? In many cases the result is to build a customer data mart and call it MDM. Sadly, the outcome of countless projects doesn’t resemble the initial objectives, but management either is unable to measure the difference or doesn’t care. They will simply issue a pass or check mark on a status report that claims to be complete. The people and process components of the project are ignored after project initiation in favor of a pure technology delivery. Even when MDM projects are successful, they don’t morph into ongoing programs that are required for enduring success.

Consider that technical silos are also a reflection of both organizational culture and technology limitations. Previously, the decision to implement data quality initiatives was assessed primarily against the risk of monetary penalties for regulatory and compliance mandates. However, today committees are established to review data quality, as though that’s the essence of data governance. When convened, however, these data quality reviews are typically reactive and do not address standards, processes, and metrics for discrete data elements, not to mention audit, balance and control, data modeling, metadata and profiling.

Unfortunately, it’s still uncommon for organizations to see beyond the regulatory risk inherent in reporting. It’s important for organizations to prioritize a focus on establishing and maintaining an operating model that incorporates policies, processes, business rules and metrics to ensure that IT actually implements them on an ongoing basis. New organizational structures often do not empower participants with either accountability or enforcement authority, resulting in disillusionment and skepticism by the business.

As soon as MDM and DG are recognized as having equal standing with other programs in terms of funding and staffing, real progress can be made toward realization of a sustained SVOT. It takes enlightened management and a committed workforce to understand that successful MDM and DG programs are typically multi-year endeavors that require a significant commitment to of people, processes and technology. MDM and DG are not something that organizations should undertake with a big-bang approach, assuming that there is a simple end to a single project. SVOT is no longer dependent on all data being consolidated into a single physical platform. With effective DG, a federated architecture and robust semantic layer can support a multi-layer, multi-location, multi-product organization that provides its business users the sustained SVOT. That is the reward.