Predictive Maintenance: Adoption Barriers and Perceived Value

Abstract

Predictive maintenance is increasingly presented as an important application of artificial intelligence in industrial production, with the potential to reduce unplanned downtime, improve maintenance planning, and strengthen operational efficiency. However, despite its technological potential, predictive maintenance has not become an established standard across goods-producing firms. This thesis examines how those firms perceive the value of predictive maintenance and what factors influence its adoption. The study is based on a qualitative research design, using semi-structured interviews and written responses from nine respondents with roles related to maintenance, operations, condition monitoring, and industrial technology. The empirical material was analyzed through reflexive thematic analysis in order to identify recurring patterns in how predictive maintenance is understood, evaluated, and discussed within organizational contexts. The findings show that predictive maintenance is mainly valued in production-critical settings where downtime is costly and early intervention can improve operational stability. However, this perceived value is highly context-dependent and does not automatically lead to adoption. Several interconnected barriers influence implementation, including data availability, uncertainty regarding return on investment, limited organizational slack, and challenges related to trust, interpretation, and human expertise. The findings further indicate that predictive maintenance requires coordination between maintenance, production, IT, and management, as adoption involves changes in routines, responsibilities, and decision-making processes. The thesis concludes that predictive maintenance adoption should not be interpreted as a purely technological decision or as a direct consequence of perceived value. Instead, adoption is best understood as an alignment problem. Successful implementation depends on the interaction between perceived value, organizational capabilities, and socio-technical integration. This perspective helps explain why predictive maintenance, despite its recognized potential, often remains limited to evaluation.