Data Governance

Requirements for Data Governance: Accountability, Transparency, Guidelines

Making a Data Governance Program successful

Are problems identified and solved in turn? What if unfortunately they are not? If poor data quality is hindering a business process at some point in the operations of a company, and if it is noticed, this does not mean that the matter can be communicated, that the cause can be clarified, and that it can be remedied by IT. Why is this so? Faulty data is often identified too late, its cause is often not investigated at the data management level, and as a result the errors continue to propagate.

Small Mistake, Big Effect

On September 23, 1999, the $125-million Mars Climate Orbiter satellite burned up in the Martian atmosphere. The cause of this failure, which was as spectacular as it was expensive, was incredibly simple. Important values had been entered into the computer in the metric unit of force that was internationally customary – the Newton – but the U.S. satellite builder had calculated these values in the Anglo-American “pounds of force.” In the corresponding field of the navigation software, some- one had simply defined the wrong unit – or rather, a unit about which there was no appropriately documented consensus among those individuals involved. And either someone was not paying attention when entering the value or did not have a clear understanding about unit definitions. The result: the trajectory was wrong, and the probe came much too close to Mars.


With the help of guidelines the company en- sures that it is using the correct data when making decisions. The policy of the telecom- munications company clearly states what parameters should be used to determine who a customer is and how that diers from who a prospect is for the company. And regarding the confusion with respect to units, the policy pro- vides information on what units of weight and measurement are used for finished goods that are handled by the company.

By means of guidelines a company pursues four essential goals:

  1. Avoiding deviations in the evaluation of data (a clear definition of the basis)

  2. Creating a basis for measuring data quality and for checking compliance

  3. Creating access for employees to all information and allowing them to see guidelines as guardrails

  4. Establishing the golden record or single point of truth where everyone in their domain has access to the necessary information

Introduction of a Governance Organization

A solution to consider is the introduction of a program from a governance organization that brings the business units, IT managers, and those responsible for operational data to the same table. The way to start with such an introduction can be split into two phases:

The goals of this approach are:

  • a governance organization is rolled out successfully and is managed appropriately

  • communication between IT and business is improved

  • accountability, transparency, and guidelines are formalized

  • data quality issues are addressed proactively

  • the data governance organization creates value for the company

Phase 1: Inventory

  • Who are the right people?

  • Who is doing what?

  • What are we doing (compared to best practice)?

  • What is not working?

Phase 2: Create transparency

  • Make collected information about data and all
    interrelationships in the company transparent

  • Consolidate existing documentation

  • Establish processes for documentation


One example of a solution is the DATAROCKET Guide tool. It supports the management of governance areas in setting up the governance organization.

An organization chart can be uploaded, and the names of the business responsibilities can be transferred into governance structures. The persons and groups who make the data- related decisions can be defined quickly at the various levels. For use in the technical structures, the tool offers standard structures for IT systems, such as Salesforce or SAP, or it enables the definition of one’s own structures. Finally, the tool also offers workflows for creating guidelines, which can then be transferred to data quality tools after the technical details have been enriched.