DATAVERSITY

Through 40 years in this business, I’ve found data usability to be a critical, but overlooked, data quality dimension. Data can be accurate and valid, but utterly unusable. This issue has never been more important. I recently met with a data lead who had this reaction to a long list of problems in his processed datasets: “Oh, we don’t have to worry about those. Users know about them. They have workarounds.” Sure they do. It’s a surprisingly common mindset on data teams, and I can understand why.

For about two and a half years, Uber calculated its own take by deducting its cut from the gross fare, including taxes and fees, rather than the net. The data flowed on schedule. The schemas were valid. The row counts were normal. Every automated check a modern data team runs would have shown green throughout, because nothing about the data was structurally wrong.

With 88% of today’s organizations using artificial intelligence (AI) in at least one business function, it’s easy to think that most business initiatives are being transformed into AI-enabled tools. The truth is that enterprise-level organizations report differently. Enterprises are still largely in the AI piloting stage, with one-third reporting that they have begun to scale AI programs only within the last several years, and a mere 7% of enterprise-level organizations reporting full deployment.

A Step Ahead is a quarterly TDAN column published by DATAVERSITY. This column explains how to perform data mapping using AI and large language models (LLMs) and how to include these capabilities in real-world data integration, migration, and analytics workflows.

A well-known public service announcement aired on American TV starting in the 1960s. The message: It’s 10 p.m., do you know where your children are? Today’s environment now compels enterprise CIOs to ask a similar question about their data. Data has become one of the most important assets for a company, and as AI scales across enterprises, it’s becoming even more critical to business success.

The rush to deploy large language models (LLMs) and generative AI agents has brought the enterprise to a critical crossroads. Chief data officers (CDOs) are no longer valued merely for the volume or cleanliness of the data they govern; they are now the architects of the cognitive context fueling corporate intelligence.

Part 1 of this series looked at why the conceptual-to-physical knowledge chain breaks before it ever reaches a system: It is funded as a project rather than a sustained practice, and it has no authority to bind the teams who build systems to follow it. Both are, at root, data governance failures: Nobody owns the decision, approves the definition, or pays for its upkeep once the project that produced it has closed. But an organization can fix both.

Artificial intelligence has quickly become a boardroom priority. Across industries, CEOs and executive teams are asking how AI can improve productivity, reduce costs, strengthen decision-making, and create new sources of competitive advantage. Organizations are making significant investments in AI platforms, infrastructure, and talent while accelerating efforts to move from experimentation to enterprise-wide adoption.

Data Governance Was Built to Answer a Different Question Data governance has matured around a familiar set of objects and practices: glossaries, data dictionaries, catalogs, lineage, data domains, quality rules, policies, roles, responsibilities, and controls. These mechanisms have delivered real progress. They help organizations understand what data means, where it resides, where it came from, how it was transformed, who is responsible for it, and whether it meets defined quality thresholds.

One of my favorite moments in any data governance initiative comes much later than most people expect. It isn’t when the governance platform goes live or when the first data stewards are assigned. It isn’t even when the business glossary reaches a milestone everyone has been working toward for months. It’s the moment people begin trusting the definitions.