Beyond the black box: impact of algorithm explainability, control, and literacy on user evaluations of personalized recommendations and advertising

This study examines how algorithm explainability and user control shape consumer evaluations on social network service (SNS) platforms, with algorithm literacy serving as a boundary condition. Study 1 demonstrates that perceived explainability and control enhance algorithm legitimacy, which, in turn, improves attitudes toward the recommendation system. Study 2 extends the analysis to personalized advertising on SNSs using a 2 (explainability) × 2 (control) × 2 (literacy) between-subjects experiment. The results reveal significant main effects of explainability and control, an explainability × control interaction, and a three-way interaction indicating that highly literate users respond most favorably when both features are high. Across studies, algorithm legitimacy consistently mediates the effects of explainability and control on attitudes. The findings identify legitimacy as the mechanism linking algorithmic transparency features to consumer responses and show that literacy amplifies these effects. Practically, enhancing transparency and user control can strengthen perceived legitimacy and improve advertising outcomes.