Clara Higuera, a distinguished lead data scientist and project manager at BBVA, recently shared her insights on the transformative impact of AI innovation in the banking sector with Global Finance. The discussion centred on BBVA’s pioneering machine learning (ML) pipeline for early debt recovery and its influence on customer satisfaction and overall operational efficiency.
In addressing the necessity for innovative solutions, Higuera underscored the historical challenges encountered by both customers and banks in debt recovery processes. Traditionally, manual procedures often led to significant delays in providing viable solutions to clients. However, BBVA’s implementation of five machine-learning models has revolutionized this process. By predicting the likelihood of clients encountering repayment difficulties and identifying extended default patterns, the bank can now offer proactive solutions within a significantly shorter timeframe, sometimes as early as within a month, compared to the previous timeline of up to a year.
When questioned about the interventions that the bank can apply when a loan presents challenges, Higuera emphasized the diverse range of solutions offered by the ML models. These include personalized calls from financial advisors to clients and the possibility of refinancing solutions. Furthermore, the models rank clients based on their criticality, enabling financial advisors to effectively prioritize client contact.
The conversation also delved into the interpretability of nonlinear ML algorithms, particularly BBVA’s use of the XG-Boost algorithm. Higuera acknowledged the potential interpretability challenges associated with these models but highlighted the presence of an evaluation and interpretability module in the ML pipeline. This module facilitates visualization and understanding of crucial variables, addressing concerns related to explaining decisions both internally and to clients.
Additionally, Higuera emphasized the importance of building trust and fostering a culture of innovation within the organization, emphasizing the integration of traditional and new methodologies to demonstrate the effectiveness of the ML models. Given the project’s multidisciplinary nature, the seamless integration of data scientists from different departments and the AI Factory unit played a pivotal role in the successful implementation of the ML pipeline.
In evaluating the impact of the five ML models, Higuera emphasized their collective role in preventing clients from transitioning to a more critical state, highlighting the holistic design of the project in aiding clients at an early stage.
Reflecting on the organizational and innovation lessons learned, Higuera emphasized the invaluable contribution of data and AI to organizational success, along with the critical role of effective communication between technical and non-technical teams. Collaboration and teamwork emerged as fundamental values that have played a significant role in BBVA’s AI innovation journey.
In conclusion, Clara Higuera’s insights offer a compelling narrative of how AI and machine learning are transforming the landscape of debt recovery within the banking industry. Through proactive solutions, interpretability modules, and a culture of trust and innovation, BBVA has epitomized the potential of AI to add substantial value to both its operational processes and customer relationships.