The importance of information governance and privacy
Information governance is in a critical state of flux. Traditional, monolithic approaches to this vital data management prerequisite are swiftly becoming outdated by the advent of distributed architectures, regulatory compliance concerns, and broader adoption rates of data as a strategic asset.
Centralized, top-down methods to information governance—as well as self-service, decentralized paradigms— lack the flexibility and holistic rigor, respectively, to effectively reduce data’s risk while boosting its enterprise merit. Instead, they simply enlarge the need for a modern governance model, primarily predicated on context and mutability, to fit the infinite circumstances of users, source systems, needs, and applications, while still formalizing how data are used.
Gartner characterizes this increasingly relevant variety of information governance as adaptive governance, in which its greatest constant is its ability to embrace different circumstances and tailor governance principles to support them.
“Adaptive governance is a way of saying context-centric governance,” explained Malcolm Hawker, former Gartner analyst and Profisee head of data strategy. “Meaning, in one context your definitions could be one thing, and in another they could be another. How marketing looks at data is different than how finance or legal does. Marketing looks at data through the lens of how people buy. Finance looks at it through the lens of how people pay.”
Shifting definitions, governance protocols, and practices to meet the contextual demands of the day requires refining staples of data modeling, data discovery, data quality, data stewardship, metadata management, and regulatory compliance to achieve these gains. Success bolsters the enterprise with a newfound propensity for diminishing risk while increasing monetization opportunities for data-driven processes—which is the very reason for utilizing data.
Although many people consider data modeling part of data preparation (for integrating data) instead of information governance, the former consideration is just one aspect of data modeling. There are also conceptual, logical, and most importantly, domain or subject area models that are foundational to the governance rules about how data are used. These comprehensive data models include business concepts and the various terminology or taxonomies that describe them. “For any data governance, for how people want to use and touch data, you need to have a set of rules, and standards and practices, and you need to have them collected, collaborated upon, and owned,” said TopQuadrant CEO Nimit Mehta.
Subject area models formalize these governance essentials in business terms. This way, information governance mandates involve “a couple of different ontologies or taxonomies of metadata,” said Matt Vogt, Immuta VP of global solution architecture. This metadata is influential for data provenance, data cataloging, data privacy, and more. Adapting these governance realms for cross-department data sharing, for example, hinges on combining models—or creating ones that evolve. One way to do so is “to bring different knowledge graphs together for dynamically addressing different contexts and intents,” said TopQuadrant CTO Ralph Hodgson. “Different roles have different needs and graphs.” As Hawker indicated, these graphs and their ontologies contain varying definitions for specific business units, even for the same terms.
Combining domain models or knowledge graphs provides a composable foundation on which adaptive data governance thrives. Data modeling also furnishes monetization opportunities, such as not just defining the term customer, but defining relationships among customers to successfully market products to an entire household, for example. “This is a modeling exercise: what data nerds call a join,” Hawker noted. Consequently, organizations should expand their metadata to include data assets and their connections. Hodgson termed this extension of metadata as “meta relations: the meaning in relationships between things.”
Because of its close ties to data modeling, metadata management is the epicenter of any effectual data governance program. It’ s integral to simply understanding data, in relationship to governance concepts, so that data can adapt to different circumstances and uses. “What helps us imagine what you can do with data is that description,” said Starburst VP of data mesh consulting Adrian Estala. According to Estala, metadata takes on different connotations for different users. End users care about its origin and data quality repercussions; data engineers care about lineage in terms of transformation and integration. The data value owner has the most meaningful lens. “They can tell you the details for where data came from,” Estala mentioned. “Why it was created, how it was created, and what its original purpose is.”