What is data governance?
Data governance is the practice of organizing processes, standards, and responsibilities to enable the company to know where its data is, how it is used, whether it is protected and how fully it meets data quality criteria (accuracy, completeness, reliability, relevance, and timeliness). It describes who can take what actions with what information, and when, under what circumstances, using what methods.
By implementing best practices for data governance, organizations reduce privacy and security risks, improve response to regulatory requirements, and enable more advanced data analytics and data science initiatives.
The goal of data governance is to manage data as a strategic asset.
To ensure the success of the data governance program, it’s also important to understand what data governance is not:
Data governance is not data management. While pursued as part of data management, data governance is only one — albeit a central — component of a data management strategy. The latter is an overarching term covering 10 knowledge domains, including Data Architecture, Data Modelling, Data Storage & Operation, Data Security & Privacy, Data Integration & Operability, Document Management, Master Data Management, Data Warehousing and BI, and Metadata Management (discover more about data management strategies here).
Data governance is more than master data management (MDM). Similar to data governance best practices, MDM also plays a key role in ensuring users’ access to the same current and complete datasets, but MDM controls only master data, i.e., the organization’s key business data about customers, products, suppliers, locations, etc.
Data governance is more than metadata management. Metadata management centers around data catalogs that use metadata (data that describes data) to deliver a searchable inventory of the company’s data assets. Data cataloging was the main focus of data governance initiatives in their early days and continues to be marketed as the answer to data governance, but it is not. Data catalogs lack many essential data governance capabilities, like lineage tracing or data quality checks.
Data governance is more than data stewardship. Data governance is about how decisions are made and how people and processes are expected to behave in relation to data. Data stewardship is the hands-on execution of data governance initiatives on a day-to-day basis (read more below).
What are the key business drivers behind data governance?
According to data governance professionals participating in a 2022 Zaloni survey, today’s investment in data governance is primarily driven by an effort to improve data quality (74%) and get faster insights from data analytics/BI (57%). In the 2022 State of Data Governance and Empowerment Report released by ESG and Quest Software, 41% of respondents ranked data quality as the top driver for applying data governance best practices against 37% who cited “improving data security.”
On the one hand, increasingly more organizations want to democratize access to quality data to drive better decisions.
On the other hand, advances in machine learning solutions have fueled interest in the revenue-generating potential of big data now successfully leveraged for business impacts. This trend is particularly strong among customer-focused companies that are struggling to get new insights as fast as possible to remain competitive.
How do leading organizations implement best practices for data governance?
Introducing best practices for data governance takes a lot of effort because you need to instill rules and procedures around data access, consistency, and uptake of new data sources. This means holding a lot of meetings with stakeholders, rather than doing purely technical work.
The second biggest challenge is finding an optimum balance. When promoting best practices for data governance, you need to give business users the right level of flexibility that will be enough for exploring data but not enough for messing things up.
Regardless of how difficult, frustrating, or costly data governance is, it is something that any organization has to deal with eventually if it wants to become data driven.
Drawing on ITRex’s extensive data experience, we’ve collected a few useful tips on how to get started with data governance best practices. Jump in.
1. Identifying the most painful problems for business
The first step would be to do an assessment of the organization’s data governance readiness to discover the pain points that are holding the company back. You need to prioritize them and build a high-level plan to work around priority use cases. Companies at the beginning of their data governance journey usually hire an external consultant to do this job (drop us a line, if you need an experienced data consultant).
For instance, you can even start building off from your CFO’s frustration over different financials they get for the same historic period every time they ask for them. Or the company’s painful problem can be an inefficient business process for discovering outdated transactions, or meaningless report generation, or siloed systems that lock important data away from users who need it quickly for forecasting patterns.
Identifying the quick wins of introducing data governance best practices and setting specific benchmarks will give you a compass and help you with the next step (see below).
2. Getting buy-in
Data governance has to start from the top to ensure role clarity and empowerment, so executive buy-in is key.
This part will be hard because data governance is rarely perceived as a profit center.
Just telling the C-suite that you want to improve data quality won’t be enough. You need a clearly defined business case built around the pain points identified in the first step. This case should promise value creation.
Basically, you should link your data governance best practices initiative to either revenue lost due to bad data and resulting business errors or man hours spent (e.g., by data teams finding, curating, or enabling data). You can also cite the costly risk of regulatory non-compliance or increasing spending on data storage.
Looking further ahead, you might also have a tough battle ahead of you fighting for buy-in from various departments. Your program might be labeled as red tape slowing down processes. Or you might have to deal with frustrated users demanding unrestricted access to everything that your policies won’t allow.
Getting buy-in from across the organization is critical to your data governance initiative. So, polish your storytelling skills.
3. Defining roles
Although each organization can set up its data governance structure differently, there are stakeholders that are most commonly involved:
Chief Data Officer: This is a centralized role primarily responsible for defining policies, standards, and rules around data governance best practices while making sure they are followed for all data projects. This role can be taken by a senior data master or by the chief information officer, who is often seen as a perfect fit.
Steering Committee: This team is often set up within senior management and includes C-suite individuals. It steers the governance strategy with specific outcomes aligned with business needs, defines data domains, assigns leaders, approves funding for initiatives, and oversees the progress.
Data Owners: These are typically senior managers or departmental heads assigned to be responsible for a specific data domain in addition to their primary role. Their tasks range from approving data definitions to reviewing master data management activities, resolving data issues, and providing the steering committee with inputs.
Data Stewards: Their major responsibility is to attest to the quality of data by ensuring data is correctly cataloged, critical data elements are identified, data lineage is documented, datasets are prioritized by impact, and data quality standards are adhered to. They run quality checks, profile data sources, implement security and analytics updates, and sometimes act on behalf of data owners approving access requests.
A clear definition of roles and responsibilities is a fundamental part of best practices for data governance. If no one owns them, then no one will care about achieving the results.
4. Doing the heavy lifting
Data governance is not a one-off project. It’s an ongoing program, and applying data governance best practices is a continuous, iterative process. This process involves complex technical activities, including:
Setting and enforcing the rules: These legislative-like activities focus on defining, enforcing, and reevaluating policies, procedures, standards, and the enterprise data architecture that will guide the organization into the future desired state. They span all stages of the data lifecycle, from data creation to acquisition, integrity, security, quality, and use.
Establishing a single source of truth: For this, companies build both a business glossary and a data dictionary. A business glossary is a list of business terms with their definitions used in the day-to-day activities of the organization. It is compiled in cooperation with all business departments to ensure that everybody uses the same language. Data dictionaries comprise detailed definitions and descriptions of data itself, or metadata, which can be data type, size, default values, constraints, etc. Both are key to ensuring that the company makes decisions on the same data.
Selecting tools: Fundamentally, data governance is not a problem that can be solved through technology, but it’s important to automate as much as possible to save time and resources. There are tools to support the process, including open-source solutions, which can help with data discovery, profiling, curation, classifying sensitive data, visualizing lineage, providing integrated business glossaries and data dictionaries, or generating data quality measurements.
To make the adoption of data governance best practices smooth and efficient, leading organizations also follow two strategies when designing their operating models:
They keep systems as simple as possible. This often means centralizing governance of technology, designing simple architecture, and keeping fewer dashboards and KPIs for easy focus and consistency. They also don’t over restrict data use, however tempting it may be. Creating an intricate web of permissions to give everyone access exactly to the data they require is confusing and delays business value.
They measure progress through metrics. The goal is not only to display the value added by data governance best practices to counter resistance but also to make sure there is continuous improvement. Sample metrics can include contributions to business objectives, risk reduction, operational efficiencies, the use of relevant tools, or conformance to standards and procedures.
5. Engaging users
The success of the data governance program is a company-wide responsibility. It ultimately depends on how end-users are data aware, literate — and excited about data enablement.
To make employees understand the value of quality data, leading organizations offer regular training. If a salesperson onboarding a client has to fill in dozens of table fields they don’t make sense of, why would they care about entering this data correctly? You need to educate users on the value of data governance best practices that is not always tangible to them. At the same time, you should also show how sticking to the rules can improve their personal and business results.
Data governance leaders also use various incentives. Such interventions include rewards for sticking to data quality standards or demonstrating new use cases, along with a punishment for errors uncovered in a periodic audit.
Set clear data governance processes and buy all users into them. Mastering data governance best practices is no easy fit, but your organization can succeed.