Data Democratization’s Impact on Users and Governance
David Stodder, senior research director for business intelligence with TDWI, discusses data democratization -- including self-service tools, the increased burdens on users, and data governance recommendations.
- By Upside Staff
- August 3, 2023
In this “Speaking of Data” podcast, TDWI’s Dave Stodder discussed the latest developments in data democratization. Stodder is senior research director for business intelligence at TDWI. [Speaker quotations have been edited for length and clarity.]
Stodder began by discussing what democratization is and what trends might be driving its popularity.
“In essence, data democratization is empowering people with the tools, data, and information so they can be influential in their roles and make good decisions.” This is being driven by the ambitions organizations have -- to improve operational efficiency and effectiveness, reduce costs and eliminate bottlenecks, and improve interaction and engagement with customers and business partners -- the traditional aspects of running a successful business.
“In an older era, spreading business intelligence out among more types of users was more of a top-down process. Now, it’s more of a bottom-up process because younger people are coming into organizations who are more accustomed to using data and have a different set of expectations.” This creates pressure on the organization to provide those employees with the data resources they’re looking for.
According to Stodder, one such resource is the availability of self-service tools.
“In a recent survey, we found that 73% of respondents consider data democratization and self-service functionality either ‘extremely important’ or ‘very important.’ This includes organizations of all sizes, not just large ones,” Stodder explained. “Some of the most innovative self-service tools have come out in the last ten years -- tools with more intuitive interfaces or drag-and-drop functionality.”
With the recent push towards natural language processing and large language models such as ChatGPT, Stodder expects to see a new wave of tools that incorporate these functions in the near future.
The key, he says, is giving users a starting point such as predefined data sets, visualizations, or other types of functionality so that less-experienced users are not stuck staring at a blank slate but can get started more easily.
Impacts on Users
Unfortunately, Stodder pointed out, moving more of these data-related functions to business users means less IT involvement, which can potentially increase the burden on the user to handle tasks such as data preparation and cleansing, and ensuring data quality. Tools such as data catalogs and metadata catalogs can simplify some aspects of data handling, such as locating its source and identifying its lineage. This means democratization becomes about much more than just a front-end tool but rather a full stack of data-handling tools.
A key result of increased user involvement in the nuts and bolts of data is the increased importance of data literacy throughout the organization, Stodder added. “It’s essential for organizations to understand what their current capabilities are and to make a plan to address any stumbling block they’re having.” Training tailored to the full range of user personas, from advanced users to more basic data consumers, will be critical to any data democratization effort.
Impacts on Data Governance
Another critical aspect of a democratization effort is an effective governance program. “Organizations can easily expand their data programs faster than they expand their governance programs,” Stodder explained, “which, given the existing strain placed on governance by regulations and the complexity of the data landscape, can only compound the problems.”
Some of these governance issues can also be exacerbated by the distributed nature of a democratized landscape. “Many organizations are trying to consolidate to a kind of hub-and-spoke model,” Stodder said, “which has been effective for many of them. However, you’re also going to see a lot of discussion about different approaches such as the data mesh or the data fabric as ways to manage distributed data architectures to achieve results faster.”
Stodder went on to describe two broad categories of governance:
- Defensive governance. This covers areas such as data privacy, conforming to state, federal, and international regulations, and preserving data security.
- Offensive governance. This involves the aspects of data management that ensure data is high quality, complete, trustworthy, accurate, and fit for purpose.
Other elements, such as data lineage, contribute to both aspects of governance by providing information about the origin of the data and the transformations that have been carried out on it to enrich and cleanse it. Such tools can also contribute to increased data security.
A Final Word
“Ultimately,” Stodder said, “it comes down to making sure that all users understand that the data owners and stewards are there to serve as guides and mentors who understand all the rules and can help others understand and comply with them.” The other key, Stodder added, is to make sure you always keep sight of the larger business strategy you’re trying to achieve and the specific question you’re trying to answer. These will keep your efforts focused in such a way that they will return value both in the short term and in the long term.
[Editor’s note: To hear the full conversation, replay the podcast episode here.]