Why Data Governance Is Not An Option
March 07, 2015
The great thing about Big Data is more companies are recognizing their data as a valuable business asset. The bad thing is that many still don’t understand the need to have a foundational data governance program to ensure business data is properly gathered, managed and utilized. For those of us who grew up in the world of regulated data, (think big consumer credit databases) I can assure you, data governance is critical.
Let’s assume you are $150M B2B manufacturing business and Big Data has caught your attention. You decide to build a data warehouse and begin dumping in vast amounts of data from sales, marketing, production and financial systems. You get excited about the possible uses of the information but as the warehouse begins to take shape, several issues start to arise. There appear to be data duplications from several systems but no one knows which is correct. There are vast numbers of incomplete records that cannot be attributed to any system in particular. Most importantly, the question of how many active customers exist in the business is largely unanswerable because a clear definition of “who is an active customer” was never established at the onset of the project.
Sound familiar? You’re in good company as a surprising number of businesses fail to establish a governance structure before they begin aggregating data. For those of you in this predicament or even better, trying to avoid it, here are four best practices that should help.
1. Recognition of the asset – it is important that company data is treated as an asset and this means the corner office has to buy in. Executive management needs to acknowledge it and company associates need to treat it as something of value. I have personally experienced both sides of this issue. For example, it is not pretty when the board decides to divest a particular part of the business and the data necessary for due diligence is unavailable or doesn’t exist.
2. It takes a village – data governance is an organizational responsibility meaning all business functions play an important role. I recommend a governance committee with representatives from all functional areas and business units operating within an agreed upon charter. This ensures everyone has a seat at the table and decisions’ regarding what data is captured and how it is used are addressed collectively. As with any committee, it is necessary to have someone in a position of authority in the event agreement cannot be reached on a particular subject. The great thing about governance committees is they can save the business time and money by ensuring that business data is clean and new initiatives are undertaken once. How many times have different business units undertaken the same analysis on the same data set in isolation? This is a gross waste of company resource and can be avoided through ongoing communication within the governance committee.
3. Build an initial data model – data models are dynamic things; always changing as the business grows and evolves. Regardless, maintaining a well-documented data model is imperative to the health of any data program. The data model should not only include the attribute definitions but it should also include source systems, frequency of update and source owner. A member of the governance committee should have responsibility for maintaining the integrity of the model.
4. Set the “active customer” definition at the onset – I recently worked with a client who could not answer two critical questions. What is the definition of an active customer and how many do you have? This question often causes much debate but once a preliminary definition is established a great deal of clarity can emanate from business systems, analytics and reporting. The governance committee should own the definition and any future changes.
As with any business challenge, data governance can seem overwhelming but if done properly, can add substantial value to the business. In my view, if you want to play with Big Data, then you’d better know the basic rules of the game.