Sustainable Automation in Financial Services: Evaluating Chatbot Efficiency, Equity, and Experience in Costa Rica’s Digital Transformation

Abstract This study explores the impact of chatbot implementation on operational efficiency, sustainability, and customer satisfaction within Costa Rican financial contact centers. It investigates whether artificial intelligence (AI)-powered automation can enhance service delivery without undermining cultural expectations for empathy and relational service. Using a convergent mixed-methods approach, the study analyzed 12 months of operational data alongside customer surveys and focus groups. Quantitative methods included t-tests, multivariate regression, ARIMA time-series analysis, structural equation modeling (SEM), and post-hoc power analysis, while qualitative data were thematically examined for triangulation. Key findings show a 43.7% reduction in Average Handling Time (AHT) and a 26.6% decline in indirect operational costs, confirming the efficiency of chatbot deployment. However, Average Speed of Answer (ASA) did not significantly improve, and 90.4% of users preferred human agents, indicating a gap between technical performance and perceived service quality. SEM revealed that customer satisfaction was influenced indirectly by cost and time efficiency, rather than by direct chatbot interaction. Although the study is limited by its single-institution scope and reliance on self-reported satisfaction data, it offers practical insights for AI adoption in service contexts with strong human-centered norms. Recommendations include developing hybrid AI-human models with dynamic staffing, emotional responsiveness, and clear escalation protocols. The study contributes a scalable framework for sustainable chatbot integration, addressing both operational needs and relational expectations in Latin American and emerging market contexts.