Financial institutions are leading the way in strengthening their data governance capabilities as artificial intelligence reshapes the industry, research by the Enterprise Data Management Association (EDMA) found.
The study, published in the international organisation’s annual Global Data Management Benchmark Report, found that financial organisations scored the highest, and beat all all other industries, in their data governance policy rollouts. Almost half of institutions surveyed were assessed as having “achieved” or “enhanced” their governance capabilities – the two highest ratings in a six-tier evaluation.
The annual study, which has been conducted since 2014, tests companies against the EDMA’s Data Management Capability Assessment Model (DCAM). It is intended as a best-practice performance benchmark across eight data management “components”, including business, data and technology architecture, data strategy and management strategy and analytics management. New capabilities, such as data management communications an metadata management, have been incorporated into the methodology.Regulatory Incentive
The EDMA study, which surveyed more than 435 organisations across more than 50 countries, attributed the financial industry’s advances to the strict regulatory codes that it is obliged to follow.
“Financial institutions consistently outperform non-financial sectors across most DCAM components, driven by regulatory pressure, formal governance expectations, and disciplined operating models,” the report’s authors wrote.
Attention has been focused on financial institutions’ data governance structures as they lead adoption of AI in their operations and analytics. That’s become even more critical with the advent of agentic AI, which automates multiple workflows and features less human interaction.Good data governance will substantially help ensure the accuracy and quality of data that feeds AI models, but it is now also vital for training data, model behaviour, life-cycle management, observability and continuous monitoring, the report stated.
“Effective AI requires well-defined data, trusted metadata, lineage, controlled access and resilient architectures,” the report’s authors wrote. “When these elements are weak or absent, AI systems inherit and amplify underlying data issues, resulting in biased outputs, operational instability and heightened regulatory exposure.”
Developing Capabilities
The report described financial institutions as being in the “developing” stage of using data management for AI adoption, ahead of other industries. This, it stated, reflected how firms were investing large amounts of capital but transitioning slowly.
The importance of governance for AI was stressed in an op-ed piece published this week by Data Management Insight, written by Shai Popat, managing director, product and commercial strategy, financial information, SIX.
“Many firms experimenting with agentic AI will already have encountered hallucinations despite using high quality data inputs,” Popat wrote. “For agentic AI to meet the high standards required for safe use in financial institutions, robust governance frameworks are essential.
“For decades, financial firms have developed governance procedures to minimise the impact of human error, whether fat-finger trades or an extra zero entered into a system,” he added. “The same discipline is now needed for AI, ensuring outputs move from plausible to defensible and from interesting to usable.”
Advances Tempered
While financial companies beat every industry in all of the components of the survey, on only handful were they assessed as having reached the higher tiers of achievement.
They scored highly in the data strategy component, which describes the “vision and value for data” within an organisation, reaching the fourth tier, demonstrating that managers and users are competently overseeing a strategy and are ready to put those processes into operation.
But within the rest of the components the industry often only nudged into the third tier, which denotes organisations that still have capabilities in the developmental stage. Its lowest scores were seen in analytics management and business data knowledge.
“Even within financial services, most organisations remain short of best practice achievement, demonstrating that regulatory compliance alone does not ensure full institutionalisation,” the report stated.
The DCAM benchmark is used by half of all the companies surveyed that operate an industry standard data management model, including almost two-thirds of financial companies, the report stated.
Subscribe to our newsletter
