In England, biscuit is a cookie or cracker and in North America, biscuit is a small, typically round cake of bread. Also, in England a muffler is a scarf or wrap worn around the neck for warmth and in North America, it is a part of a motor vehicle's exhaust system, serving to muffle the sound of the vehicle. What's the point? Imagine an interaction between an individual and sales rep in a clothing store in North America asking for a muffler. It can be very frustrating. And the cause of frustration arises from the fact that the same word with the same spelling means something totally different for these two individuals. This scenario however could be attributed to the fact that same word could mean different things in different parts of the world.
For all of you working in the world of data and analytics, regardless of which industry vertical you are working in, it is not uncommon to see inconsistencies in the master data. Master data is a term used to define critical data for business. Common examples of master data include Customer, Product and Geography. Unlike the example, we discussed above where the impact was a potential loss of sale, the cost of having inconsistent master data can be very detrimental and can lead to poor business decisions if left unresolved. Let us consider an example to further elaborate this. Merger and strategic partnerships between businesses is very common these days. When two manufacturing companies decide to go for a merger it also means a merger of their IT applications and more specifically for this example, their data applications. Both organizations could be maintaining their own material definitions, recipes, codes and other relevant information. Most likely this master data would not be the same across the two companies merging together and if this difference is not identified, data lineage is not identified then without a complete impact assessment this would most certainly lead to misreported data like misreported volume, inventory levels etc. ultimately leading to poor business decisions.
Impact to the ultimate consumers of data results in complaints like - I can't trust this data, numbers do not look the same across different reporting applications, I am missing volume for product(s), wrong product is showing up as out of stock etc.
There are even more serious repercussions in the financial and healthcare industry where compliance and regulatory restrictions are of utmost importance.
It may seem that taking care of such things should not be such an uphill task; however, in the reality most often it is one of the most challenging task for most of the companies.
Even today, in spite of great deal of advancement in the area of technology, having a comprehensive data governance strategy remains a daunting task.
Data governance (DG) by definition refers to the overall management, accessibility integrity, and security of the data employed in any organization.
Typically, when the customers start thinking about a data governance strategy or framework, they are already in pain and would elaborate at length of their symptoms. The first step would be start mapping unique symptoms with unique problems. A couple of simple and very common examples:
Symptom - "It takes forever for my IT team to assess the impact of changing a column in my ERP system to the downstream applications."
Potential Problem - "Lack of a data lineage document"
Symptom - "Different reporting applications are showing different revenue numbers for the same product"
Potential Problem - "Lack of standard and definitions, poor master data management"
A robust data governance strategy would also include the formation of a governing body or a council which defines the standards and definitions, owns and maintains them.
Different organizations in the different verticals of industry could be at a different maturity level when it comes to Data Governance. TDWI Maturity model can also applied to Data Governance maturity for an organization. Diagram below provides some standard indicators as organizations move up the maturity model. Even though the focus areas would be distinct across different industry verticals, the first step towards establishing a Data Governance framework would be a thorough analysis and assessment of the current state; this is no different than how any other project would be approached.
This has been a challenging area for majority of the organizations all along and the technological advancements of the recent times are making it a first priority for all organizations.
Emphasis on having an all comprehensive Data Governance model is not just stemming from the need to mitigate risk and avoid litigations, it also results in significant cost savings by removing duplication of effort, faster turnaround for changes involving multiple applications and taking the overall efficiency of the organization up by several notches.
In years to come, implementing Governance frameworks would continue to be at the top of the list for all CIOs and Risk Management officials and can no longer be deferred to a more opportune time for implementation.
About the Author
Deepti Singh has over 15 years of experience implementing Business Intelligence applications for clients globally. She is a Director in the Richmond, VA office with CapTech managing Data Analysis and Governance service offering and has successfully delivered projects in pharmaceutical, financial, retail, oil & gas and consumer packaged goods sectors.