Trust in Algorithms in E-Commerce Recommender Systems: A Bibliometric Mapping (2012–2025) and a Managerial Playbook for Acceptability

With the increasing integration of artificial intelligence into e-commerce platforms, trust in algorithmic decision-making has become a critical issue. Recommender systems significantly shape consumer choices and influence visibility within digital marketplaces, yet remain largely opaque. This study aims to bridge the gap between algorithmic accuracy and perceived trustworthiness by conducting a bibliometric and topic modeling analysis of 163 peer-reviewed publications (2012–2025). Results indicate a paradigmatic shift from usability-focused approaches toward governance-aware frameworks encompassing fairness, explainability, and accountability. To capture this transformation, the Acceptance Triangle model is introduced, conceptualising algorithmic acceptability across three interdependent layers: trust calibration at the interface level, exposure fairness at the platform level, and accountability mechanisms at the institutional level. The model is further operationalised through the Trust UX Playbook—nine managerial design levers with associated key performance indicators—and a Composite Acceptability Score integrating accuracy, fairness, and complaint reduction. The findings suggest that trust alone may be insufficient for understanding long-term acceptability in e-commerce recommender systems. Instead, the alignment between user experience, market equity, and governance legitimacy is interpreted as an analytically useful condition for conceptualising algorithmic acceptability. This research contributes a structured framework for assessing and designing acceptable recommender systems, offering actionable guidance for designers, decision-makers, and regulatory stakeholders seeking to improve algorithmic transparency, fairness, and accountability in online commerce.