There has been a great deal of talk in recent years about how companies must adopt a data-driven strategy, however one aspect of doing this is typically forgotten: data governance. One of the key takeaways from Gartner’s Data and Analytics conference this past March was that through 2019, just 10% of companies will require oversight of operational information from line-of-business functions.
There are two primary reasons for this inattention to data governance. The first is that it’s not sexy. That is, it doesn’t demand the big thinking of, say, machine learning. It doesn’t use fancy algorithms. And it doesn’t offer an immediately apparent return on the dollars invested.
The second reason is that the implementation of data governance is often presented in a way that overwhelms company executives. You are advised to convene a panel of data stewards and subject-area experts to create and set forth comprehensive, detailed, and specific policies that you must then disseminate enterprise-wide. Sounds fascinating, doesn’t it?! It’s no wonder that many people want to simply skip this step.
And yet, of all the elements involved in adopting data-driven strategy, data governance is among the most crucial.
All that said, how can executives approach data governance in a simple and headache free manner? An agile-driven data governance program that starts small and grows over time is much more efficient than a top-down approach. Building from the bottom and proceeding cautiously and incrementally will help to ensure not only data consistency, but also more widely spread adoption of best practices. Such an approach can succeed where many, larger data governance initiatives have ended in failure.
Start small. Data governance programs can satisfy a variety of business goals, even if they are small. In some cases, the goal is to comply with such regulatory mandates as HIPAA or Sarbanes-Oxley. It can also enable improved data management enterprise-wide. And perhaps most frequently, data governance offers access to data assets to inform better business intelligence and faster business decisions.
To be of quantifiable use, data must accurately reflect reality. Ideally, the closer the data point is to its source, the greater the likelihood that mistakes will be avoided. The easiest way to achieve this is to push data management and stewardship down to the line-of-business functions. So rather than a process of data governance implementation that begins with an executive-led committee that dictates to end users how data should be managed, it would be better to start with a single data steward whose responsibility stems from their proximity to where transactions occur and includes just one type of data, such as addresses.
Manage through metadata. Metadata gives information the meaning it needs to be useful. In its simplest definition, metadata is “data about data.” So, for example, a data point might be the settling of a claim from an auto insurance company. The metadata might include details about the insured and the claimant, details about the vehicles, other details about communications between the parties, and so on.
But, more importantly, metadata helps to define data points. To take the auto insurance settlement as our example, suppose that one data point is “Claim Paid Date.” Such a data point could indicate one of three things: the date the claim was approved; the date that the check was issued, or the date the check cleared. These could all be valid definitions, but their meanings vary greatly. In order to ensure common understanding of terms, a company needs to, as a critical element of data governance, create what is known as a “business glossary”. With such a document, everyone in the company would have a common understanding of not only what specific data points mean, but, more importantly, when they change and why. This is the most critical aspect of good data governance, helping to ensure that data is accurate, consistent and understood across the enterprise.
Focusing the greatest effort on metadata and the business glossary is a key feature of this agile approach.
In conclusion, iteratively adding to a data governance structure can benefit an organization by building on what works, and adding lines of business based on strategic priority. Just like the benefits of using an agile development process, you can tweak and tune your approach as the governance structure is built out. A truly data driven organization must consider the implications of data quality and consistency to the bottom line; one of the proven ways to do this is through awareness and prioritization that is facilitated through the structure of governance.