Personalization Versus Privacy in Digital Marketing Examining the Interactive Effects of Data Transparency Perceived Intrusiveness and Consumer Trust on Purchase Decisions

Abstract

The increasing adoption of artificial intelligence and data-driven personalization in digital marketing has transformed consumer engagement, product recommendations, and purchasing behaviour while intensifying concerns regarding personal data exploitation, algorithmic surveillance, and perceived privacy intrusiveness. Although personalized marketing improves information relevance and consumer convenience, excessive personalization may undermine consumer trust and negatively influence purchase decisions. This study proposes a novel Privacy–Transparency–Trust Optimization (PTTO) algorithm to investigate and predict the interactive effects of marketing personalization, data transparency, perceived intrusiveness, and consumer trust on purchase decisions. The proposed framework integrates Privacy Calculus Theory, the Stimulus–Organism–Response framework, and Trust Theory within a hybrid analytical architecture combining Partial Least Squares Structural Equation Modelling (PLS-SEM), Extreme Gradient Boosting (XGBoost), Random Forest, and Explainable Artificial Intelligence using SHapley Additive exPlanations (SHAP). The PTTO algorithm introduces a trust-sensitive interaction mechanism that models nonlinear relationships, conditional mediation, and moderation effects while optimizing purchase-decision predictions subject to transparency and privacy-intrusiveness constraints. The empirical design combines structured consumer survey responses with consented digital marketing interaction records to evaluate perceived personalization benefits, transparency perceptions, intrusiveness, trust formation, and observed purchasing behaviour. Model performance will be assessed through stratified cross-validation and independent holdout testing using accuracy, precision, recall, F1-score, area under the receiver operating characteristic curve (ROC-AUC), Brier score, and calibration error. Comparative experiments will benchmark PTTO against Logistic Regression, conventional PLS-SEM prediction, Random Forest, XGBoost, and a multilayer perceptron neural network. Graphical analyses will include moderation plots, response surfaces, SHAP feature-attribution diagrams, ROC curves, calibration plots, and comparative algorithm-performance charts. The study will test whether data transparency strengthens the positive relationship between personalization and consumer trust, whether perceived intrusiveness weakens purchasing responses, and whether consumer trust mediates these competing effects. The proposed framework is expected to identify circumstances under which personalization enhances purchase decisions without generating disproportionate privacy concerns. Its contribution lies in integrating interpretable behavioural relationships with privacy-constrained machine learning to establish an empirically testable framework for transparent, trustworthy, and consumer-centred digital marketing. Research status: The PTTO algorithm is a proposed research contribution. Its comparative superiority remains a hypothesis to be tested using empirical data; no performance values are presented as established findings.