Analyst Perspectives

Data governance has always been a critical part of the data and analytics landscape. However, for many years, it was seen as a preventive function to limit access to data and ensure compliance with security and data privacy requirements. To fulfill today’s data-driven agendas, many enterprises need an evolved perspective on data governance. The development of new applications driven by artificial intelligence requires a more agile and collaborative approach to data governance—one that automates and accelerates previously manual processes to increase access to data with ongoing requirements for regulatory compliance.

Data governance is integral to an overall data intelligence strategy. Good data governance provides guardrails that enable enterprises to act fast while protecting the business from risks related to regulatory requirements, data-quality issues and data-reliability concerns. I assert that, through 2026, the primary concern for more than three-quarters of Chief Data Officers will be governing the reliability, privacy and security of their organization’s data.

The importance of data governance is well recognized and understood, with almost 9 in 10 participants in ISG’s Data Governance Benchmark Research identifying data governance as important or very important to their organization. The primary benefits of data governance are improved data quality, accuracy of reporting and business intelligence, operational efficiency and enhanced regulatory compliance.

Traditionally, many enterprise data governance initiatives were driven by manual processes reactive to changing data privacy, security policies and regulatory requirements, limiting access to data to ensure compliance with these guidelines. This approach poses challenges for enterprises trying to respond quickly to evolving security threats, competitive concerns and regulations, as well as new opportunities to deliver enhanced efficiency and new business opportunities with the development of AI-driven applications.

AI is heavily dependent on data, so data governance and privacy issues that impact data and analytics also impact AI and generative AI. Additionally, GenAI systems can exacerbate governance risks. As my colleagues Jeff Orr and David Menninger have recently explained, GenAI systems can inadvertently generate harmful or offensive content, so enterprises must guard against toxicity to prevent unintended consequences. GenAI models also learn from historical data, which may contain biases. Enterprises need robust mechanisms to detect and rectify bias during model training and deployment. Plus, enterprises should create and enforce policies that govern the use of sensitive data by GenAI applications, including strong privacy controls and best practices to safeguard against breaches. Failure to address these governance challenges could severely impact an enterprise in terms of damaging its reputation and customer relationships, as well as falling afoul of emerging regulations, such as the European Union Artificial Intelligence Act.

Rather than limiting the use of data, the implementation of well-defined data governance policies and procedures provides a framework that expands access to data, which enables enterprises to make faster decisions by providing a platform for self-service data discovery and analysis with AI. Enterprises implementing GenAI find that governance is an essential enabler of success, with better coordination and governance rated as the number one factor that could improve the value, or time to value, for participants in ISG’s GenAI Market Lens research.

Our research illustrates a gap between awareness of the need for governance in AI initiatives and policies to govern AI and machine learning models. More than 4 in 5 participants in our Data Governance Benchmark Research reported that governing AI and ML is important or very important. Despite that, less than one-quarter of enterprises have data governance policies for AI/ML models (23%) and scores generated by AI or ML models (22%). Using data governance for AI is a work in progress: almost one-half of participants plan to govern scores generated by AI or ML models (48%) and AI or ML models (42%) in the future.

One reason for this is a relative paucity of functionality. As my colleague David Menninger recently explained, the 2024 ISG AI Platforms Buyers Guide research indicated that AI platform software providers have been slow in incorporating AI governance capabilities. In the interim, data governance products, in coordination with AI platforms, can improve trust in data for AI projects. Key data governance capabilities that support AI workloads assessed in the 2024 ISG Buyers Guide for Data Governance include data usage, data lineage, data quality, data security and access control capabilities. As I recently explained, data governance catalogs also enable data stewards, data quality and data governance professionals to define and manage data usage policies, view and manage data profiles, determine and administer data quality rules and define and administer data models and master data definitions.

AI and data governance are symbiotic. Data governance processes and products can help improve AI, but AI also has a role in data governance by automating and accelerating previously manual processes. For example, AI can automatically identify personally identifiable information and other forms of sensitive data, flagging potentially inappropriate use. AI is increasingly incorporated into data quality software to automate and enhance data quality checks, supporting automation of data classification, metadata management and data lineage. 

The use of AI to improve data governance is a work in progress. Less than 1 in 3 software providers assessed in our Data Governance Buyers Guide were graded A- or above for the use of AI for data asset descriptions (30%), metadata generation (26%), data usage recommendations (17%) and data masking or obfuscation (13%). Data governance providers are still developing and testing these features, so I anticipate this performance will improve when we conduct the research for the 2025 Buyers Guide study. In the meantime, I recommend that enterprises evaluating data governance platforms and providers be conscious of how data governance improves AI and how AI can improve data governance.

Regards,

Matt Aslett