POSTED : May 8, 2015
BY : ProKarma

What does James Madison, fourth President of the United States, have to do with your data? Truth be told; almost nothing. However, Madison did have some profound insights on the power of governance. Witness this quote: “Knowledge will forever govern ignorance; and a people who mean to be their own governors must arm themselves with the power which knowledge gives.” These words, spoken at the very start of the Nineteenth Century, could have been aimed directly at data governance.

Data Governance is a set of policies and principles that guides the rules around cradle-to-grave lifecycle of data. PluralSoft has been part of governance committees in multiple organizations and some of key lessons learned are listed below.

Strategy

  • Data is a Corporate Asset to drive business performance optimization strategies. Corporate strategies do effect organizational change. Therefore, the governance charter should be aligned with strategic business objectives. The three common modes that are often used are “goal-driven”, “user-driven” or a “data-driven” approach. Of these, goal driven approach is recommended as the organizational goals permeate to all people and processes.

People

  • Personnel should demonstrate Healthcare domain expertise beyond just technology expertise to conceive, design and help their organization implement a sustainable, long-term Business Information Strategy.
  • To gain the most leverage out of an enterprise wide Information Strategy, the importance of accountability cannot be over emphasized. The organization should have cross-functional (Executive, Administrative and Operational) execution arms that constantly interact and collaborate on the build-out and sustenance of the program.
  • Dedicated resources are mandatory
  • Reward performance through soft incentives such as skill upgrades and career path enhancement; and hard incentives such as innovation bonuses.

Business Rules/Data

  • Business process model of Payer and Provider lines of business should be understood to better understand data lifecycle of core data elements serving information for strategic decision making.
  • Rationalization of source data definition and business rules governing the data element’s life cycle (from cradle to grave), especially for those data elements that overlap across multiple source systems, must be performed.
  • Rationalization / Standardization of business vocabulary through adoption of applicable healthcare standard nomenclature such as LOINC, SNOMED, NIH’s UMLS Meta-Thesaurus.

Process

  • A process for definition of Key Performance Indicators must be established and governed.
  • Clear definition of “what is” and “acceptable level” of data quality with approved “data lineage” is mandatory to achieve the vision of “single version of truth”. It is very important to scope out the “fitness of purpose” for defining the acceptable level of data quality for information analyses.

Metrics

  • Definition of Key Performance Indicators, the monitoring of their performance and utility, their ownership, and finally its change management is critical to delivering accurate and trusted business performance measures to stakeholders.
  • Having SMART (Specific, Measurable, Achievable, Realistic, Timely) business goals and performance requirements for key business processes will ensure SMART KPI definition.
  • KPI development / definition require high levels of business stakeholder involvement and senior executive sponsorship.

Applications

  • Do not underestimate end-user training (various user functional categories and levels of decision making) when delivering BI applications. Time spent on Training and Support is inversely proportional to effort required for ongoing support and maintenance of applications.

Architecture

  • A prerequisite to the design of any iteration of the enterprise data warehouse is a thorough understanding of the business processes and available data sources.
  • Source system Data Profiling is a key pre-requisite step before publishing an enterprise logical data model.
  • Adopt a model driven development approach using a single database design tool, so that logical transformations are well understood by the designers of physical data model as well as data quality analyst and ETL developers.

Not too bad for a Statesman and President who served over 200 years ago!