Joshua Burkhow, Chief Evangelist at Alteryx
As AI moves from experimentation to implementation, data readiness has emerged as a defining factor in success. The challenge is that many financial institutions are still working with data environments that were not built to support AI at scale. Our research shows that 51% of banks and insurance organisations rate their data maturity as moderate or low, while 47% identify poor data quality as a barrier to achieving higher levels of maturity.
Investment in AI continues to accelerate across financial services. Lloyds Bank, for example, plans for AI to support all customer interactions by 2030, while new industry-specific offerings such as Chat GPT for Financial Services reflect growing momentum behind AI adoption. Yet many organisations lack the underlying foundations needed to move from experimentation to enterprise-wide deployment.
Siloed data, legacy infrastructure, technical debt and inconsistent governance continue to erode trust in analytical outputs while creating obstacles to responsible AI adoption. In an industry where AI-driven decisions can influence lending, fraud prevention, and customer outcomes, the quality of those decisions depends on the quality of the data behind them.
By connecting data across the organisation, modernising core systems and strengthening governance practices, firms can create the conditions for AI to deliver value with confidence. This enables organisations to move beyond isolated pilots and apply AI more effectively across the business.
Poor data quality is the biggest blocker
Data quality remains one of the biggest determinants of AI success in financial services, yet many organisations are still operating without a reliable foundation. Banks are dealing with legacy tech in their data stacks, while one in five have limitations that prevent them from innovating.
Many financial services firms are attempting to scale generative AI on top of data that is incomplete, duplicated, inconsistent or siloed. The result is often inaccurate or misleading outputs, which can quickly be amplified across the organisation. At the same time, valuable context remains trapped in legacy systems and spreadsheets, including institutional knowledge and undocumented business rules. This not only affects model performance but can also limit an organisation’s ability to trace, explain, or justify decisions, creating governance and regulatory challenges.
This drives a second obstacle to AI adoption. When AI-generated outputs are unreliable, confidence in the technology begins to erode across the business. Senior leaders become more cautious about approving AI-driven transformation initiatives, while frontline teams are less likely to rely on AI insights in their day-to-day decision-making.
Moving away from a ‘centralisation-only’ mindset
To address these challenges, many financial institutions have accelerated investment in cloud-based data environments designed to bring together disparate information sources and improve consistency across the organisation. Establishing a more unified data estate is a critical step in strengthening data quality and creating the foundations needed for effective AI adoption.
Yet consolidation on its own is not enough. In many cases, accessing and analysing data within these platforms still requires technical expertise, whether that’s extracting information from underlying systems or working with query languages such as SQL.
The result is that AI initiatives can become concentrated within specialist data and technology teams rather than the business functions where operational decisions are made. This reflects a long-standing model in which analytics is driven primarily through IT-led programmes. While that approach may have worked for previous generations of data projects, AI demands much closer involvement from the teams that understand business processes, regulatory obligations and customer outcomes.
Whether it’s onboarding new customers, assessing credit risk, investigating suspicious activity or managing compliance requirements, those closest to the process are often best placed to identify opportunities for AI and determine how success should be measured. For organisations seeking to evolve into truly intelligent enterprises, data readiness extends beyond building a centralised data platform. It requires giving business teams secure, governed access to trusted, context-rich information so they can help shape AI-powered processes, apply appropriate controls, and embed operational expertise into decision-making workflows. In financial services, where accountability, transparency, and auditability are essential, this becomes even more important.
At the same time, centralisation does not automatically solve the issue of trust. Stronger data foundations can help improve the reliability of AI outputs, but organisations still need clear oversight of how data is accessed, transformed and used within AI systems. Without robust governance and enforcement mechanisms, concerns around transparency, control and risk can continue to limit AI adoption, regardless of how modern the underlying data architecture may be.
Building the layers needed to support business logic
Data lakehouses bring together large volumes of enterprise data within a single environment, creating a stronger foundation for AI. However, centralising data alone does not address many of the challenges that have historically slowed adoption.
An independent analytics layer can help bridge that gap, allowing teams to work directly with governed data without complex extraction processes. This makes it easier for business users to develop practical AI use cases grounded in trusted, high-quality information. These platforms also help organisations embed the business rules and operational knowledge that underpin financial services, from risk controls and compliance requirements to fraud detection and customer onboarding. By applying these rules consistently across AI workflows, organisations can improve reliability, transparency, and governance.
Many analytics tools also provide low-code interfaces, enabling business teams to build and refine AI workflows without specialist technical skills. This makes it easier to validate processes, review data dependencies, and ensure AI applications align with business objectives before deployment. Together, data lakehouses and modern analytics platforms can help financial institutions move AI from experimentation to production. But long-term success depends on more than technology alone. It also requires investment in data literacy, clear governance frameworks, and a culture that supports responsible AI adoption at scale.
What’s next?
As financial institutions increase their investment in AI, progress will depend on their ability to address the data challenges that continue to hinder adoption. Without reliable, accessible and well-governed data, organisations risk becoming stuck in a cycle of pilots and isolated use cases that never scale. Building trust in AI outputs starts with strengthening the foundations beneath them.
When robust data architectures are combined with user-friendly analytics capabilities, business teams are better equipped to develop AI applications that are both operationally relevant and compliant with regulatory expectations. The institutions that view data readiness as a strategic business priority, rather than a technical exercise, will be best positioned to unlock AI’s full potential.