Last week, I participated in a roundtable during a conference in Paris organized by the French branch of DAMA, the data management international organization. During the question/answer part of the conference, it became clear that most of the audience was confusing data management with data governance (DG). This is a challenge my Forrester colleague Michele Goetz identified early in the DG tooling space. Because data quality and master data management embed governance features, many view them as data governance tooling. But the reality is that they remain data management tooling — their goal is to improve data quality by executing rules. This tooling confusion is only a consequence of how much the word governance is misused and misunderstood, and that leads to struggling data governance efforts.

So what is “governance”? Governance is the collaboration, organization, and metrics facilitating a decision path between at least two conflicting objectives. Governance is finding the acceptable balance between the interests of two parties. For example, IT governance is needed when you would like to support all possible business projects but you have limited budget, skills, or resources available. Governance is needed when objectives are different for different stakeholders, and the outcome of governance is that they do not get the same priority. If everyone has the same objective, then this is data management.

Another mistake for understanding governance: seeking consensus-based decisions between parties. A long-time head of data governance at one of the largest European utilities companies recently told me how she has made the mistake of trying to find consensus for every decision. At this point, she’s tired and wants to change her job. This search for consensus is a mistake in a changing world where speed and agility are the new normal. The goal of effective governance must be to guide multiple decisions that together indicate a reasonable direction. For the previous IT governance example, each project asks for a decision, but together, they should be in the allocated budget. This is another aspect of the transformation of data governance in enterprises; it requires more agility for each project but at the same time harmonizes the consolidated consequence of these unitary decisions.

So what are the recommendations for differentiating data governance from data management?

·         It’s critical to identify early what are the objectives (and potentially confusing objectives) of different stakeholders (quality versus costs and risks versus agility).

·         Don’t try to arrive at consensus on these objectives. Be aware of the gaps between them but also of how they relate to each other. For example, seeking 100% quality (a common objective for some data) could waste a huge amount of money without a reasonable return.

·         Identify the KPIs associated with each objective and the limit threshold. A sign of bad data governance is showing only one metric addressing only one objective.

·         Craft a preliminary metric target specifying a reasonable compromise for each project.

·         Begin to consolidate the multiple project metrics into a simple representation showing the direction of their multiple unitary decisions.

Using these multiple metrics and consolidating the direction to go in, you will get governance and ultimately achieve operational excellence.