When algorithms meet emotions: Understanding consumer satisfaction in AI companion applications

AI companion applications are transforming how people form and maintain relationships in the digital world, with millions of users now engaging in emotional and social interactions with AI agents. Understanding what drives user satisfaction becomes crucial as these applications become increasingly integrated into users' daily lives. Drawing on Orlikowski's practice lens theory and employing text mining and hierarchical clustering methodologies on user reviews enhanced with focus group discussions, this study identifies two key determinants of user satisfaction: Functional Capability Perception and Affective Social Attunement. The analysis of 156,637 user reviews and insights from diverse participants reveals that satisfaction emerges through users' simultaneous negotiation of technical proficiency and emotional boundaries in their AI interactions. Functional capabilities positively influence satisfaction, while deeper emotional engagement creates a paradoxical effect where users become more sensitive to AI limitations. The study demonstrates how societal conditions shape these dynamics, with evaluation criteria evolving from simple acceptance to deeper interpersonal connections, reflecting changing attitudes toward human-AI relationships. Our mixed methods approach uncovers the contextual factors and usage patterns that shape how users integrate these technologies into their daily routines and emotional ecosystems. These insights advance our understanding of human-AI relationships while providing practical guidance for developers on creating adaptive systems that respond to evolving user needs. By understanding these complex dynamics, stakeholders can develop AI companions that enhance human relationships rather than attempting to replace them.