Here in the blog we like to discuss current topics, often inspired by experiences we have working with clients, or meeting with prospective clients, or even sometimes in the (data) news. At least twice in the past couple of weeks we’ve had occasion to talk about ways to “hook” data stewards, and we wanted to explore what that might mean in more detail than these brief conversations allowed. Also, we’ve been trying to spotlight data stewardship in this space – check out our spotlight page for some of our previous posts.
Our approach to data governance and data intelligence has evolved over time, but it has always featured the idea of meeting people where they are in terms of their data needs and use. Data governance isn’t, or shouldn’t be, an extra set of tasks you do when you have time; it should be integrated into your existing data practices. And, in fact, some form of data governance probably already is part of your regular practices—it’s just that many activities aren’t recognized as data governance, and also that many data activities aren’t fully completed in our haste to move on to the next task.
Data stewards have, or if they’re new they may be in the process of developing, deep knowledge of how data is used in their domain’s operations. They make key decisions about what data is gathered, where and how it is stored, with whom it can be immediately shared, whether it is sensitive, and how and by whom it might be regulated. For any given data domain in your organization, it’s probably immediately obvious who these people are, whether data steward is part of their job description or even part of the terminology used in your organization.
Many of the questions people, especially data consumers, have about data can best be answered by a data steward, or a steward’s designee. Where did this data come from? What does it mean when we call some data point by this name? Is the data in this summary accurate? And many of the follow-up actions once these questions are asked are also best taken or delegated by data stewards, whether that’s clarifying terminology, or investigating a potential data quality issue, or confirming that data wasn’t damaged or malformed as it traveled across multiple systems.
Just because someone’s a data steward doesn’t mean they know everything about data, however. Too often we see data stewards burdened with assignments that are only tangentially related to their expertise and their data.
- Requests for access to data their office collects come in element by element, person by person. Data stewards don’t know the whole story behind these requests, and aren’t in a position to do much research. The heuristics available to determine whether to grant access, and to what, tend to be suboptimal.
- Data stewards can provide explanations about how they and their teams use certain data, but it’s unreasonable to expect them to know much about how the same data is used downstream, or in other offices, and it’s especially unreasonable to expect them to anticipate that usage and to make changes in the way they collect and store data in order to accommodate that other use case.
- And while some data stewards may have robust business intelligence or data querying skills, most are not analytics experts, and yet they are frequently expected to provide feedback or approval on products for which they don’t have context or full background.
To be clear, those tasks do represent a certain kind of data stewardship. In our webinars, we talk about (a minimum of) four categories of data stewardship, including subject matter expertise, accountability, gatekeeping, and support/triage. Your domain data stewards typically excel at the first one, but too often we see them dragged down by the other ones.
- If you have a formal data governance structure, your data stewards probably report up to trustees or guardians, which is ideally where you’d like to see the accountability for data reside. We might think of this as a strategic stewardship, whereas the people actually called data stewards would focus more on managing data for operational uses.
- Gatekeepers determine levels of access to tools and data sets, and their work would be informed by domain stewards, who provide data classification, who identify anomalous or poor-quality data, and who could be consulted on thorny or complicated questions about access and usage.
- If you’re well-staffed and well-organized, you might be able to use your existing help desk, or to stand up a kind of data help desk, where users with questions could submit them for triage, evaluation, and potentially a referral. There might even be a role for agentic AI here, which your organization is probably already piloting or considering. (See our post from December 2024 about how AI might handle routine tasks.) Some of those referrals would go to functional stewards, but many of the questions that now go to our core data stewards could easily be handled elsewhere, freeing up stewards’ time to focus on other concerns.
Let's return to the conversation about hooking data stewards. By hooking we mean something like building intrinsic, long-term, consistent engagement with a range of valuable data governance activities and practices. We work with data stewards from all manner of data backgrounds across a variety of organizations, but still we consistently hear that their data governance experience has consisted of fulfilling compliance requirements, performing box-checking busy work, repeating manual tasks that could be automated or mechanized, and answering questions when there could be easily discovered answers in a self-service knowledge base or other resource. That work isn't interesting or challenging, it doesn't drive the business forward, and it certainly doesn't position data as a strategic asset.
So how might we build this long-term, meaningful engagement?
One method might involve rewriting job descriptions. Many employees are motivated by and receive satisfaction from fulfilling job requirements, and so making data stewardship a formal responsibility might build up some additional engagement. However, there's always the possibility that adding specific data stewardship duties to a job description, particularly if those duties are not forward-looking, might result in people fulfilling the letter rather than the spirit of data stewardship. And what data stewardship involves today might be vastly different in a short time: are you willing to commit to reviewing and rewriting these job descriptions regularly and responsively?
Another approach could be creating something like a data steward community of practice, which could take a lot of shapes, but is probably characterized by an informal gathering of data stewards on a regular basis. Where we’ve seen these, they exist outside (but not in opposition to) formal data governance structures, and they are designed to facilitate the free flow of ideas, collaboration, and shared insight. Team building is probably good in nearly every situation, but there is always the possibility that informal, unstructured gatherings end up unfocused and of limited utility.
Both of those methods require some amount of administrative effort and sanctioning. So we remain firmly on the sidelines: one or both will work for some organizations, and they could well be counterproductive in other organizations. Or, as we used to see a lot on the old internet, your mileage may vary.
Ultimately, to cause lasting change, you probably want to be disruptive, but not too disruptive. You want to leverage existing skills and abilities, but also build up new competencies. You want to expand and enhance those practices that are productive, and shed the ones that drag down performance or are otherwise not cost-effective. So what does that look like?
We introduced the Data Cookbook about fifteen years ago as a way to free up data stewards from repetitive, time-intensive tasks, among them constantly explaining to data consumers why a given metric exists, demonstrating how critical calculations are performed, reaffirming what data-related terminology means (or gently clarifying that their interlocutor's notions were mistaken), and so on. We hoped that the value proposition would be obvious, and that clients would invest a certain amount of extra effort up front – and probably reduce or delay their efforts elsewhere, at least for a time – until the Data Cookbook became an organizational knowledge base where people would go with data questions. Invest time now so as to free it up later for value-creating work.
What we learned over the years was that people recognized the value proposition, but they struggled to let go of other tasks, ingrained expectations, and what we might call organizational baggage.
- Not all people who are accustomed to having data stewards at their beck and call jump at the opportunity to do their own data discovery, to say nothing of self-service reporting or analytics.
- Organizations that have cried wolf for years about data privacy and security have often developed a diminished or blinkered vision of data stewardship, and simply putting a new tool in front of people doesn’t cut it.
- Gaps in data literacy, insufficient training and reskilling opportunities, and the massive growth in the amount and variety of data collected; taken together they can overwhelm individuals and organizations. When it comes to managing and governing data, where do we start?
And data stewards themselves can find it difficult to transition from an environment of tribal knowledge to an environment of public awareness. This is not, in our observation, a sign of craven self-preservation or simple resistance to change; rather, it’s a recognition that organizational channels for information sharing and data discovery don’t natively lend themselves to populating a business glossary, annotating dashboards, writing data quality rules, and so on.
However, if these and other high-value, long-lasting data stewardship activities can be performed under the aegis of ongoing work, or a new and critical data project/initiative, then it’s easier for them to become routine and essential. Moving to a new and more modern ERP? Hook data stewards during data mapping exercises. Thinking about a new SaaS tool? Hook data stewards when they make their case for the new application: what data are you going to capture? how will it affect your business processes? will the new system store sensitive data, and if so, what is the plan for keeping it safe? Publishing new management dashboards? Hook data stewards during a certification process that signs off on the accuracy and usability of the information therein.
If you don’t have sexy new technology initiatives, you might have to be a bit more creative about developing those hooks. We assume that when there’s something exciting going on, at least some portion of your data stewards will volunteer to be involved, and maybe they’re even the drivers. But you can also hook stewards on mundane and ordinary topics, as long as you use the right bait and you cast your line judiciously. Maybe it’s just a business process review, or seeking operating efficiencies. Sometimes there really is an actual audit of your data!
We've seen clients successfully use the Data Cookbook to jumpstart a laggard or nascent data governance program, and we've also seen them use it to re-energize successful but low-voltage governance initiatives. We've not seen anyone start from scratch and do everything! Our successful clients identify what's already working, and use the Data Cookbook (among other tools) to support and extend that work. Where there are roadblocks or bottlenecks, they look to use the Data Cookbook (often in creative and unexpected ways) to get around them.
The key is for people, especially data stewards, to find their data management tools both indispensable and routine. That's our approach with our Data Cookbook, our integration work, our and our consulting practice. We'd love to try it out with you.
Hope you found this blog post beneficial. To access other resources (blog posts, videos, and recorded webinars) about data governance feel free to check out our data governance resources page.
Feel free to contact us and let us know how we can assist.
(Image Credit: StockSnap_HXDODYZKFS_happycouple_smilingdatastewards_BP #B1287)
