As the data revolution wades on, nearly every organization with a data estate (large or small) is looking toward the future. Attention is being focused on how AI and modern analytical capabilities can help increase revenue, decrease costs and mitigate the associated risks. At the heart of this, or any sound data strategy, is governance.
AI will make data governance concerns central to the CFOs
The extreme reach of AI, both within and beyond an organization, will put traditional data governance programs to the test. Traditionally overseen by the CTO, CIO or CDO and with merely nominal approval from the CFO, these governance programs will now become a central concern for the CFO as they aim to manage risk more effectively.
Simply uttering the words “data governance” tends to bring shivers down the spines of executives, analysts and developers alike. The demand for data governance, however, is driven by customers (internal and external), shareholders and regulatory agencies. Ironically, the same people who shudder at the very idea of data governance are the same people who can benefit the most in the long-term.
Data governance is necessary for a better outcome
With the idea in mind that data governance is a “necessary evil” of the modern data estate, companies need to find ways to implement programs that deliver the largest impact and most minimal disruption. Simply implementing data governance as a check box on a list of annual goals has proven unfruitful time and again.
According to Gartner, “by 2027, 80% of data and analytics (D&A) initiatives will fail”. This fact only adds to the perceived uphill battle a company faces in their journey to achieve data and analytics programs.
So how does an organization become the 1 in 5 that gets to the other side of the chasm?
Implementing a data and analytics program that sticks
The answer to the successful outcome for a D&A program lies in strategy, planning, execution — and perhaps most important — a sound change management program. At the start of any good plan, organizations must align the strategic goals with the task at hand.
Strategy first: Understand the data governance playing field
There is no one-size-fits-all model for data governance. Every organization has its own unique DNA to consider. Current data and analytics maturity, competing priorities, skill gaps, business buy-in, executive sponsorship, trepidation from previous failed attempts, modernization efforts and evolving corporate strategies are just some of the many variables that need to be considered when developing a sound program.
It’s imperative for any data governance leader and practitioner to take a step back and understand the playing field before introducing a change with a pre-conceived reputation to an organization. Failure to do so will only result in further harm done to the program and long-term resistance to future attempts.
Data governance and AI: Identify the use case
As it pertains to AI capabilities, companies must first start with the proverbial “why?” Is there a burning use case that needs to be solved for? Is there a new regulatory mandate that needs to be adhered to? Can automating a manual, arduous task that has been burdening the organization save significant time and money and allow for new growth opportunities? Perhaps it’s just developing a sound data literacy program to help teams work more cohesively with one another.
Identifying a use case that aligns with corporate goals, associating the true ROI for it and obtaining buy-in and sponsorship should be the first step to develop or augment a sound data governance program.
Understand the challenge at hand: Questions to consider
With a use case (or cases) agreed upon, it’s next important to understand how to apply a governance model to satisfy the ask, both during development and for long-term nurturing and maintenance.
Some questions that should be considered include:
- Is there an existing governance program in place to apply to the solution?
- Is the data necessary to solve for the use case available and of high quality?
- Are there data owners, stewards and subject matter experts available to help understand and curate the data?
- Are the pieces in place to support, monitor and enrich this solution moving forward?
- What are the data privacy and/or security considerations needed to successfully and safely deploy this new solution?
- How can data lineage and transparency be measured to assure customers, regulators or other consumers of a sound answer (especially important for AI solutions)?
- Do the consumers of the solution have a common data dictionary and catalog from which to comprehend the data they’re reading?
- For AI solutions, has the data and solution been vetted for ethical considerations, bias and privacy?
Data governance alone will not be able to solve many pitfalls that may be associated with the above considerations. However, the data governance team should be identifying potential gaps and bringing them to the attention of the proper stakeholders for remediation.
Apply the right data governance model for your organization
In 2024, most data governance experts and practitioners will tell you that a centralized data governance model is not adaptable to modern-day agility requirements (certain exceptions withstanding). Determining which model is right for the organization (i.e. decentralized, hybrid, hub and spoke), however, is of importance to any effort and is paramount to adoption across the organization.
For example, consider the breadth of knowledge a customer service AI chatbot may need to have at its disposal to be effective. Also consider that the data must be constantly vetted and curated at each source to make sure the parts form a sound holistic solution. Each of these sources need ownership and stewardship for accountability. If any of the pieces break down, the more the risk of the entire solution losing trust from its users. Assigning the owners, the stewards, and the other roles consistently within a governance organization and giving them a place to collaborate and determine standards is instrumental.
Start small. Test, rinse, repeat.
For an organization to introduce a data governance program with complete success on the first try is unrealistic. Companies should go into any data governance program accepting that trial and error will be a part of the journey.
For those with a more mature data and analytics environment but little to no formal data governance program, the methodology is fundamentally different from those starting in a place of nascency. Look to find champions for the program and apply a “right size, right fit” approach. For adoption, it’s key to allow for governance to enrich any solution without impeding progress along the way.
How Senturus can help institute well-considered data governance
Senturus has over 20 years of experience implementing enterprise data and analytics solutions built on a strong governance foundation. We take a pragmatic approach to technology, keeping business value top of mind and delivering in phased stages with minimal impact to business.
Whether the ask is more strategic in terms of establishing or re-defining a data governance program or requires more tactical tasks such as a data literacy (catalog/dictionary) effort, master data management or data privacy assessment, Senturus can help.
If your organization is considering implementing AI to its data and analytics strategy or could use guidance to improve existing oversights, get in touch with us. Senturus can help you institute a well-considered program that aligns with your strategic goals and ensures long-term success.
About the author:
Greg Frasca is a Sr. Solutions Architect at Senturus. An accomplished data and analytics professional with more than 20 years of experience, he leads our data governance and strategy practice.
