Reliable, secure, self-service access to trustworthy, clean data is at the center of alleviating the data teams burdened by data requests. While self-service data access is straightforward in theory, implementing such a data strategy requires a deep understanding of business user needs coupled with a complete reimagination of data architectures, from data governance to data delivery, support networks, and more.
Wayne Eckerson, president, Eckerson Group, led the annual Data Summit pre-conference workshop, “Data Democratization: How to Build a Self-Service Strategy That Empowers Business Users and Eliminates Data Bottlenecks,” discussing practical, time-tested approaches to data democratization, helping to eliminate data bottlenecks while effectively anticipating business needs.
The annual Data Summit conference returned to Boston, May 14-15, 2025, with pre-conference workshops on May 13.
“Self-service feels so simple, it’s a win-win,” said Eckerson. “If our users are happy, the data team is happy.”
While rapid, agile access to data is a great idea, “if you don’t pair it with standards and governance, it all falls apart.” With too many reports, conflicting data, and increased costs and risks, governance becomes crucial in ensuring that self-service data democratization is able to realize its potential benefits.
Eckerson likened the relationship between self-service and governance as that of a fast car and brakes: Would you drive a Maserati without brakes? Likely not; governance is the brakes to an ultra-fast self-service data framework.
“You have to do everything else right before you can deliver self-service,” added Eckerson. “The first step is to know thy users. If you don’t know your users…you’ll miss the mark.”
Understanding your users and assigning them “personas” to define their behaviors, typically falling between casual data users to more intensive, enables organizations to tailor their data services and foster a more granular sense of self-service. Eckerson further divided organizational personas into more specific roles:
- Data consumers that want personalized, interactive dashboards to consume reports
- Data explorers that want customizable, interactive dashboards, sometimes modifying reports
- Data analysts that want analytics workbenches, creating reports from scratch
- Data scientists that require Python, Notebooks, Sagemaker, etc. to create ML/AI models
“Self-service for each of these personas is different, which is why you need to create an inventory,” explained Eckerson. This inventory organizes the classification information of users, including skills level, access level, as well as the amount of training or support that will need to be offered. These charts—which assign personas paired with names, titles, and contact information—allows enterprises to know exactly who they’re serving.
The right operating model helps unify and federate these users into a consistent framework. Divided into three parts, a good operating model consists of:
- Enterprise data team, which builds and maintains the enterprise data platform and all shared asserts, delivering strategic enterprise solutions
- Domain-based development, which understands business needs, building and maintaining domain-based data products and solutions
- Business domains, which specify business needs, consuming and governing data
“Nothing [your users ask for] should be a surprise if your operating model is working properly,” Eckerson noted.
Regarding architecture, Eckerson emphasized the importance of an “MVP” approach—or minimum viable products—in order to first build something simply workable. Proving its ability to function, then improve upon that architecture, helps build momentum and funding, discover true requirements, and avoid costly failures.
Returning to the importance of data governance, “doing an assessment right off the bat for your data governance program is not a bad idea,” said Eckerson. He further offered the following tips for establishing a robust data governance structure for self-service:
- Find a top executive who is committed to data.
- Piggyback on a major defensive or offensive initiative.
- Focus on one dimension of data governance, such as conflicting reports or data quality.
- Find a leader to take charge.
- Form a cross-functional team to support the leader and guide the initiative.
The right project management is another significant piece of self-service data democratization, helping to:
- Allocate scarce development resources.
- Gain a consensus on how to allocate resources.
- Prioritize and align development efforts.
- Require business and technical leaders to talk.
“There is a lot of proactive work at a strategic level…to make sure things stay on track, which helps you stay ahead of project requests,” said Eckerson.
Product management—not project management—is another aspect of data democratization that helps centralize a variety of information to better propel self-service data frameworks toward success. Data products are a subset of data assets, according to Eckerson, maintaining unique characteristics that may differ from data assets. Successful data products are:
- Targeted to and audience
- Designed for broad sharing and reuse
- Packaged with rich metadata
- Systematically governed
- Managed by a product team
- Continuously funded and improved
- Consolidated and accessed through a data marketplace or data store
Many Data Summit 2025 presentations are available for review at https://www.dbta.com/datasummit/2025/presentations.aspx.