Inside AIB’s Data-Driven Legal Panel Review Process

According to a recent Brightflag article, legal departments are increasingly under pressure to optimize performance, control costs, and demonstrate measurable value. At Allied Irish Banks, Head of Panel Management Steve Cummins shows how a data-driven legal panel review process can deliver on all three fronts. 

The review was prompted by market shifts and the recognition that the existing panel, which had been in place for four years, no longer aligned with AIB’s needs. Cummins and his team sought to modernize service delivery, maximize value from external counsel, and establish clear expectations around financial management. To do this, they formed a cross-functional review team that combined senior leadership with staff who regularly worked with outside counsel, ensuring that both strategic and practical perspectives shaped the process.

The team requested detailed submissions from law firms on pricing, resourcing, and value-added services, while integrating spend and performance analytics to assess historical performance. This combination of forward-looking proposals and backward-looking data provided a comprehensive framework for evaluation. The review also included a reset of billing guidelines, adopting best practices, and tailoring them to enforce accountability and transparency.

The outcomes have been significant: improved cost control, better service alignment, and stronger relationships with their legal panel firms. More importantly, AIB now has a repeatable process to benchmark firm performance and maintain accountability. A key achievement has been achieving 98% billing guideline compliance, a metric Cummins emphasizes as critical for predictability and cost efficiency. 

For Cummins, the most important lesson of this data-driven legal panel review process is collaboration. By fostering transparency and challenging established practices, AIB has created a review model that legal teams elsewhere can adapt to ensure both value and continuous improvement.