From Linear To Nonlinear: Pls Sem And Ann Analysis Of Ai Personalization And Brand Engagement In Muslimah Fashion E Commerce

Wening Ken Widodasih (Universitas Pelita Bangsa)
Kurbandi Satpatmantya BR (Universitas Pelita Bangsa)
Wirajannastho ABR (Universitas Pelita Bangsa)
Felina C Young (Philippine Women University)

Kasta: Jurnal Ilmu Sosial, Agama, Budaya dan Terapan
Vol. 6 No. 3 (2026): August

This research investigates brand engagement as a mediator between AI personalization and marketing outcomes in Muslimah fashion e commerce (Jabodetabek), with religiosity as a moderator. A survey of 412 consumers (18–45 years) who bought modest fashion online in the last six months was analyzed using PLS SEM and an artificial neural network (ANN). Four AI components—interaction (?=0.324), information (?=0.281), accessibility (?=0.192), and customization (?=0.358)—positively drive brand engagement (R²=0.672). Brand engagement completely transmits AI’s effects to repurchase intention (indirect ?=0.452), customer lifetime value (?=0.394), and word of mouth (?=0.378). Religiosity only moderates the engagement repurchase intention path (?=0.182, p<0.01). The ANN identifies brand engagement as the most influential predictor (100% normalized importance). This study extends the S O R framework to a religiously sensitive e commerce setting and introduces combined PLS SEM and ANN analysis to Islamic marketing literature.