Analisis Klasifikasi dan Laten Trait Loyalitas Nasabah Bank Syariah Indonesia Berdasarkan Kualitas dan Efisiensi Layanan Digital

This study examines customer loyalty regarding the low Google Play Store ratings and slow error-resolution of the BYOND BSI application compared to competitors. It aims to map and classify the latent traits () of customer loyalty among Islamic Banking students at UIN Sumatera Utara with a total population of 853 students. Using Slovin's formula, a minimum sample of 90 respondents was established. Employing Item Response Theory (IRT) via the Generalized Partial Credit Model (GPCM), latent trait estimation () was conducted after verifying unidimensionality and the Test Information Function (TIF). The instrument yielded an optimal TIF  within the range of , establishing four loyalty categories where high digital quality and service efficiency correlate with high loyalty, and vice versa. However, anomalous segments were identified: loyal customers despite poor perceived service, and disloyal customers despite good perceived service. This reveals that loyalty is also driven by external factors such as Sharia compliance and switching costs. Consequently, technical service quality alone is insufficient for customer retention, leaving users vulnerable to competitor switching.