The Data Governance Institute defines “Data Governance is a system of decision rights and accountabilities for information-related processes, executed according to agreed-upon models which describe who can take what actions with what information, and when, under what circumstances, using what methods.”
The world of connected devices has resulted in data explosion. Each one of us is sharing different types of data through emails, sensors, and smart devices.
In business organizations, coordinated data governance strategies clearly demarcate between ownership and responsibility areas pertaining to the specific data. These boundaries are created based on product categories, geographic regions, market segments, as well as internal business functions.
Let us illustrate a high-profile example to learn about the challenges of data security. The 911 Commission once faulted government agencies like FBI, NSA, and CIA due to unwillingness to share important security data across government agencies and other branches.
Similar challenges are faced by corporations, hospitals, universities, and governments, in relation with making data shareable. This problem may be resolved once enterprises start recognizing data to be a shared asset and treat it like one.
This is why enterprises must focus on implementing data governance measures. A set of policies, practices, and organizational standards will provide a base for deriving valuable insights from organizational data.
The general consensus is that decision-makers want credible, reliable, accurate, predictive and insightful data. What does that mean? How will it translate into efficient guidelines of data management? The best method is to present relevant data to the decision-makers through ‘data lineage’ such that they understand how and why the specific data elements are important.
Data lineage refers to data flow across an organization from its origin, usually with a customer/sensor/external source, through its application and consumption across different stages of the data lifecycle.
The data life-cycle process illustrates various stages where data is transformed, or bad data is discarded. These stages are referred to as basic data governance points to check, validate, authorize, and approve data elements; it may also be used if there is a change in data ownership or responsibility.
Data is a complex and specialized domain, having its own specialists and jargons like metadata, ETL, master data management, Hadoop etc.
Data governance is context-based; it may refer to:
- Organizational bodies
- Decision rights
- Policies, rules , standards, and business guidelines
- Key accountabilities
- Enforcement of information-related activities
Lack of Data Governance
Global enterprises are resorting to data governance measures without understanding its definition properly. From where did this data come? Do you have the right data? Can you use this data legitimately? Do you have any other alternative? How would you transform this data into meaningful business insights? Does the data lack quality? Most importantly, does everyone interpret and understand this data equally?
These are some questions that enterprises must answer before adopting data governance measures in their organization.
Forbes mentions the potential challenges of ungoverned data through the following image:
Effects of Ungoverned Data
- Rorschach Effect: Each person interprets the organizational data in his or her own personal meaning. For instance, if every employee interprets the data displayed on the company dashboard in their own way, it would disturb the consistency of the big picture. One report may define a customer in a completely different way from another customer. Similar effect may arise in the case of expenses or revenue. In addition, every employee may think of a different business model for driving business growth. So, if you do not define data clearly, it may result in free interpretations and self-serving associations.
- Data Brawl: This effect occurs when a person uses the data to arrive at a conclusion which brings disrepute to another person in the organization. For instance, a CFO says sales have declined because only closed deals were taken into account. The VP-Sales mentions that sales are increasing due to a huge number of deals through successful contract negotiations.
The CEO’s response normally will hamper the quality of data, further leading to disrepute and heated arguments, which could have been avoided with the help of efficient data governance.
If you are a line decision maker, you must understand the actual meaning of the numbers and the collected data. However, absence of a common language might hamper data processing and interpretation techniques; every one related to the data must be on the same page and understand its common objective.
Often, employees and decision makers look out for quick answers, to what they consider as simple questions, for instance, how many customers do we have? But, would this yield a successful and reasonable outcome?
No. Actually, this response leads to complex technical explanations. A genuine explanation of why you cannot find answers quickly is often interpreted as an excuse or an evasive argument.
Data governance addresses these issues efficiently. It focuses on developing a policy and strategy including:
- Identifying stakeholders, clarifying key accountabilities, and establishing decision-making rights.
- Establish, monitor, and review polices and standards
- Establish strategies for managing important enterprise data
Key Principles of Data Governance
Following principles assist stakeholders in collaborative measures to address problems related to data, something which is inherent for any organization.
To develop a reliable, sustainable information management strategy, you must have an enterprise-wide and multidisciplinary approach to data governance. A sound data governance strategy must receive support from all levels across the entire organization.
The first principle is clear ownership. You need to form a data governance council to formulate data policies, standards, and procedures for the entire company. Representatives of all business units must be a part of this council.
The second principle is value recognition. However, it is difficult to measure the value of critical business data in terms of dollars, but it is an important business asset. Businesses without proper standards and quality of data do not function well. Data governance efforts understand the importance of business data, and along with recognition from the C-suite, financial support must be provided aptly for the effort and expenses incurred on effective data management strategies.
Third principle refers to effective policies and procedures of data governance. For maximum effectiveness, these policies and standards must cover data silos across the organization and apply to your business as a whole.
A fundamental BI model is reliable and can be followed by every team member. Otherwise, it may lead to data chaos and create problems in delivering quality data and business decisions.
Another principle guiding data governance policies is the quality of data. The source of data should be reliable and trustworthy. Some companies practice human intervention to fix and control data quality. However, with a wide range of data quality methods and tool available, poor quality of data should be a rarity.
Data security, an important element of data governance, is often combined with data access. But data security extends provisioning access to qualified data analysts and includes the business capabilities of securing stored data of your enterprise. Most of the solutions are collecting huge volumes of data using cloud storage methods. While these methods are secured with several layers of data security measures, enterprises must understand the overall business impact of data transfers from cloud solutions to other technologies.
You should be concerned about data leakage when sharing data with external parties. Understand and document how your data is being stored, shared, and safeguarded across various data collection agents.
Privacy concern is a crucial element of data governance; enterprises must inform a customer about what data is being collected and how are they using it. Several best practices are followed within published privacy guidelines; however the best principle is to deliver clear, concise data usage policy offering customers an opt-out function for those customers who do not want to be tracked.
In addition, being aware of and classifying data as anonymous or segment-wise, personal details wise will help enterprises treat each dataset independently. The classification must be governed by technology that goes beyond web analytics and deals with BI, CRM, and enterprise marketing management.
Data integrity is expected of enterprises because modern data collection techniques leverage the outcomes of processed data. Web analytics equates data to a set of actions constituting a single ‘user session. For instance, IBM TeaLeaf associates multiple online activities with a single user session to provide essential context for the collected data.
This data is presented in a way that reveals true insights of the digital environment. Many enterprises separate processed data as ‘raw data’, which is stored in the enterprise data warehouse and analysed later. Beware of inherent risks of storing raw data that may result in complex, incomprehensible data. Data integrity ensures data is used and analysed in the intended form.
Access to enterprise data has become a concern to address quickly. Provisioning access to corporate email accounts of employees. Govern data access when partners share data externally or with other technology partners. Enterprises must be careful when data is shared by third parties, ad servers, data aggregators etc that may compromise the restricted access to digital data.
Along with the data governance principles mentioned above, enterprises must remember that an effective framework helps in efficient management of organizational data. It also improves coordination between departments and provides relevant business insights related to products and different business units.