This paper investigates the challenges of explainability in recommendation systems for ecommerce, using the case study of Zeplace., a system developed by ThePlaceToWish. In a context where transparency is both a regulatory requirement (EU RGPD, Raison 71 , 2016) and a strategic factor for building user trust (Theis et al., 2023), we propose an approach that integrates explainability from the system's design phase. The model is distinguished by its exclusive use of declarative data provided by users via a personalized avatar, enabling transparent recommendations that uphold privacy. We examine the technical and ethical challenges encountered during the system's development in an industrial context and discuss the solutions implemented to address them. By offering clear and customized explanations, the system aims to enhance user trust and engagement, particularly in the emotionally charged context of gift selection, which involves both expectations and financial considerations. This work provides insights into the design of explainable and ethical recommendation systems within the e-commerce sector.