The Privacy-Personalization Paradox in AI-Driven Customer Experience Analytics: A Trust-Centred Integrative Review and Responsible Personalization Framework

Abstract

Artificial intelligence increasingly supports personalised customer experiences through predictive analytics, recommender systems, behavioural profiling and conversational interfaces. However, the data collection and inference required for personalisation can also heighten perceptions of surveillance, vulnerability and loss of control. This integrative conceptual review examines how AI-enabled personalisation, perceived privacy risk and consumer trust interact within customer experience analytics. The synthesis integrates privacy calculus, social exchange, psychological reactance and signalling perspectives to explain how consumers evaluate personalisation value against privacy-related costs. A trust-centred framework is developed in which personalisation intensity, data sensitivity, algorithmic inference and recommendation relevance jointly shape perceived value and privacy risk, with consumer trust linking these evaluations to engagement, disclosure willingness, satisfaction, loyalty and continued use. The review also distinguishes actual data practices from perceived privacy intrusion and identifies contextual conditions that may alter this balance. The review further proposes a context-dependent nonlinear relationship between personalisation intensity and trust: moderate personalisation may strengthen trust when relevance and value are proportionate to data use, whereas more intensive personalisation may reduce trust when intrusion, surveillance or loss of autonomy becomes salient. Responsible personalisation is therefore framed around proportionate data use, meaningful transparency, consumer control and privacy-by-design. The framework and proposed personalisation threshold remain conceptual and require empirical testing across AI applications, cultural settings and regulatory environments.