Customer perceptions of AI-driven customer service: evidence from Somalia

Abstract

Artificial intelligence-driven customer service is transforming relationship management through automated interactions, yet its success hinges on trust, service quality, and satisfaction—especially in fragile contexts. This study offers Somalia-specific evidence on how these factors shape satisfaction and recommendation readiness, while testing digital literacy as a moderator. A cross-sectional survey was conducted in major urban centers among users of AI customer service in telecommunications and mobile banking. Using a bilingual questionnaire, 353 valid responses were collected and analyzed via covariance-based structural equation modeling. Results showed moderate overall perceptions: transparency was the weakest trust facet, privacy concerns were prominent, efficiency drove service quality, and personalization lagged. Satisfaction exceeded that of human services. All hypotheses were supported—trust and service quality predicted satisfaction, which in turn predicted recommendation readiness. Satisfaction partially mediated the effects of trust and service quality, and digital literacy strengthened the satisfaction–recommendation link. The findings indicate conditional AI acceptance in Somalia, where functional performance drives satisfaction more than abstract trust, but transparency and privacy remain barriers. Hybrid service models, transparency-by-design, localized personalization, and digital literacy investments are critical to equitable adoption and avoiding digital stratification.