Data Governance Best Practices

If you ask ChatGPT for a description of data governance best practices, you’ll get the following list, all solid concepts:

1. Establish Clear Data Governance Policies
2. Define Roles and Responsibilities
3. Ensure Data Quality
4. Implement Data Classification and Access Controls
5. Create a Data Governance Framework
6. Enable Data Lineage and Metadata Management
7. Ensure Compliance and Risk Management
8. Foster a Data-Driven Culture
9. Leverage Technology and Automation
10. Monitor and Continuously Improve

What stands out to me about this list is how much it implies organizational and management actions and how technology adoption is a component of a governance effort, but only as it supports policy and agreement and management.  This focus on organization is a core theme of all we have written about governance recently.  Put simply, if your idea of “doing governance” is buying a catalog system, then you’re probably not going to meet any of your business goals and the program will eventually die on the vine.

Our List

With the above in mind, we provide some ideas on governance practice that cut across these 10 classic activities and can be followed no matter what level of governance you are implementing.

1. Incremental implementation wins the race
2. Set Realistic expectations
3. Establish a long-term vision & leadership commitment
4. Govern the whole data lifecycle – but it’s most effective & efficient to ensure that data is good from birth 
5. Determine good metrics to track most important goals

Know Your Why

Like an inverted Anna Karenena quote, unhappy data governance efforts fall into many of the same traps and errors, but successful ones are varied as the organizations they serve. Truly understanding your organization’s motivations for data governance can go a long way toward avoiding common pitfalls like abandoning an effort part way through because a reasonable return on the investment can’t be determined. As we noted in Part 1 of our LinkedIn governance series, you need to be able to answer why data governance is valuable for your organization, and then tailor your efforts, investments, tooling, and success metrics to those specific goals. 

Ideally, this should be in service of a larger data strategy for the organization, but also focused on making measurable inroads on a manageable scope of that strategy. Which of the data problems are you focused on? Do you need to track regulatory compliance efforts more efficiently, or improve data quality in key BI tools, or more quickly understand the downstream impact of upstream data changes, or something else entirely? Each of these deserves a different approach, and they all will need more than data cataloging. 

As part of that vision of what a good data landscape looks like – and why it’s valuable to your organization – you’ll have the core of your justification for investing in data governance, and the best ways to track what return on that investment looks like for the organization. 

You Can’t Discover What You Don’t Search For

Unless you have a perfect understanding of your data landscape – and who does, really? – start your data governance effort with a comprehensive data discovery project. As we discussed last time, many organizations will turn to data governance to discover their data landscape for them; while governance is a good function to lead this effort, discovery is a distinct project, and a small subset of what data governance should do.

Documenting data models in a data catalog and a business glossary that captures core business concepts can be a valuable part of discovery, but discovery should include more, such as 

• identifying pain points, 
• data quality analysis, 
• assessment of overall data practices 
• identification of key data metrics to track
• grading the organization with a data management maturity assessment, and
• alignment with overarching business goals and strategy. 

Most essentially, data discovery should have a clear end point, where data governance requirements are produced, and consensus on the goals, metrics, and strategy for the data governance program is achieved and put in writing – in short, where you agree on the why of your governance effort, and then treat that why as your guiding star.

Find the Right Level for You

Determine your strengths and weaknesses across the various core data management disciplines. Over time continue to measure your organization’s growth and maturity (or lack thereof). 

There are many standard models with slight differences in focus (CMMI-DMM, EDM Council DCAM, IBM’s, Stanford’s, Gartner’s and more) and some organizations branch off from these to generate their own; many organizations also apply this kind of rating to their data with a medallion architecture.

Whatever model you use, it’s most important that you use one, and continue to use it as part of your tracking metrics over time. In particular, maturity models that distinguish among data management disciplines and focus on dimensions relevant to your data problem help refocus investment and data governance effort over time, and offer warnings when efforts are slipping in one area or another. 

Consider also what “fully governed” means for your organization and what your target maturity level is. By default, most organizations think they need to aim to get top marks in every maturity category; in the CMMI-DMM this would be that the organization believes and acts as if “data is seen as critical for survival in a dynamic and competitive market” in their approach to every data management discipline. 

But not every organization is (or should be) a data-fueled tech company: Data–and data governance–that is “good enough” to support your core business might realistically be a better fit at a middling maturity (if still likely a few notches above where you are now). Having more modest, realistic data maturity goals in alignment with the core business can help avoid unrealistic ambitions that set data governance up for failure.

Know when you need help

Especially for smaller organizations or ones shorter on existing data resources or maturity – know when you need help. This might take the form of adding a data governance role, finding someone inclined to the work and training them, or bringing in outside expertise (perhaps even consultants, such as ourselves!).

–Monica Fulvio & Matt Petrillo