How Shawbrook is building trusted data foundations for AI

About Shawbrook

Shawbrook is a specialist UK savings and lending bank serving established businesses, property professionals and individuals. Its model combines specialist human expertise with digital, technology and data capabilities to deliver the flexibility, speed and certainty its customers value.

The challenge

As Shawbrook's data estate grew, managing data quality through informal reports and spreadsheets became increasingly difficult. People across the organisation could flag that data looked wrong, but the process depended heavily on individuals who knew the data well. Recent mergers and acquisitions added further complexity.

At the same time, the scale of the environment continued to increase. Around 100 analysts write SQL queries across roughly 2,000 Power BI dashboards. In a regulated bank, incomplete or inaccurate data can affect more than an internal report: it can flow into regulatory submissions, business decisions and, increasingly, AI-generated answers.

Shawbrook needed a more systematic way to detect issues, establish ownership and create a continuous feedback loop between the data team and the business.

The solution

Shawbrook introduced Bigeye to automate data quality monitoring and make issue management operational. When Bigeye detects a problem, the relevant teams can be notified, an owner can be assigned and progress can be tracked through to resolution.

The implementation has also helped Shawbrook broaden the conversation beyond data quality. Continuous monitoring, lineage, profiling and the ability to identify and tag Critical Data Elements are informing wider governance policies, runbooks and workflows. The team is also comparing manual reconciliation processes with Bigeye for selected use cases as it considers how the platform can become a trusted source for data-quality evidence.

Catching a regulatory reporting risk early

One of the clearest examples came from Shawbrook's monthly MLAR-D regulatory reporting, which includes information about loans and the status of applications. Shawbrook placed a Bigeye control on the field used to record a 'dead date' - the point at which an expired application cannot be revived.

Bigeye identified 22 applications that had reached expiry status without a dead date. If those records had been included unchanged, Shawbrook could have reported incorrect information to the regulator.

The alert did more than identify an anomalous field. When the data team investigated with the business, they found that underwriters were expected to run a wizard as part of the application workflow. That made it possible to address the underlying process rather than simply correct the affected records.

The example shows how data quality can become a practical governance mechanism: detecting an issue early, tracing it back to the process that created it and giving the right team clear accountability for fixing it.

From data quality to governance

As visibility has improved, Shawbrook has also gained a clearer view of where further work is needed, including legacy schemas and sensitive-data visibility. The team is using profiling and AI-recommended data quality rules to focus its effort, while building a hierarchy of Critical Data Elements that can support internal governance and future regulatory conversations.

What began as a data-quality initiative is therefore evolving into a broader governance foundation, with Bigeye increasingly positioned as a core part of Shawbrook's data-management approach rather than a tool used only for regulatory reporting.

Creating trusted foundations for AI

The same foundations are becoming more important as Shawbrook expands its use of AI. Its AI engineering team has been proving what is possible, while the wider business is beginning to ask how AI-generated results can be trusted, when an answer may be less reliable and what evidence is needed to certify the data behind it.

Shawbrook is working towards Databricks and Genie workspaces through which users can ask questions of data rather than relying only on dashboards. With stronger monitoring and governance in place, the team has greater confidence in the datasets it makes available for these experiences. The controlled approach is also helping Shawbrook build confidence with information security while continuing to innovate.

The impact

• Earlier detection of data issues before they affect regulatory reporting or downstream decisions.
• Clearer ownership and accountability for resolving data-quality problems.
• A stronger working relationship between the data team and business stakeholders.
• Growing visibility into lineage, sensitive data and Critical Data Elements.
• Greater confidence in the trusted datasets that will underpin AI-driven querying and decision-making.