Algorithmic Intentions, Material Consequences: Personalization in the Subscription Economy

Abstract

Many subscription box services market their products as uniquely tailored to each consumer’s individual preferences. This personalization typically begins with an on- boarding quiz where consumers articulate their tastes and preferences, creating the expectation that each subsequent delivery will reflect these stated choices. While 85% of businesses believe they are offering a personalized service, only 60% of consumers agree. This disconnect suggests that “personalization” often functions more as marketing rhetoric than as a substantive product feature. The implications of failed personalization extend beyond poor customer satisfaction; they also drive significant waste: from discarded products and excessive packaging to carbon emissions generated by unnecessary manufacturing and shipping. This thesis investigates whether companies truly deliver on the promise of personalized selection. The research employs a two-part methodology: (1) a controlled 2x2 factorial experiment comparing expert vs. non-expert participants who receive either preference-aligned or preference-misaligned boxes, and (2) qualitative interviews to explore participants’ perceived satisfaction, product fit, and personalization. The experiment was carried out on two different subscription boxes from distinct product categories: IPSY, which delivers a curated beauty box, and MistoBox, which ships subscribers a hand-picked bag of coffee. The findings illuminate the intersection between consumer experience, algorithmic design, and environmental impact, offering ways in which personalization strategies can be optimized to serve consumer and sustainability objectives. Notably, in the experiment, over 80% of IPSY products went unused, highlighting the urgent need for more effective personalization models to reduce waste and improve product relevance.