When systems satisfy but users struggle: explaining the legacy system and satisfaction paradox during professional internships

Purpose Legacy information systems remain embedded in the administrative infrastructure of many public-sector organisations, yet little is known about how they shape the satisfaction of students who encounter them through mandatory work-integrated learning. This study examines how legacy system quality affects administrative practice quality and student satisfaction during internships. Design/methodology/approach An explanatory sequential mixed-methods design was used. Survey data from 78 office administration students who had completed internships were analysed using partial least squares structural equation modelling. The five respondents with the lowest satisfaction scores were then interviewed to help explain the quantitative results. Findings Legacy system quality predicted administrative practice quality (β = 0.842) and satisfaction (β = 0.432). Administrative practice quality predicted satisfaction (β = 0.469) and partially mediated the effect of system quality (indirect β = 0.395), accounting for 47.8% of the total effect. Nevertheless, the least satisfied students described server outages, repeated manual data entry and work that “feels manual”. Research limitations/implications The cross-sectional, single-programme design limits causal inference and generalisability, and two constructs showed borderline discriminant validity. Expectation heterogeneity is proposed as an untested explanation for the divergence between the quantitative and qualitative findings. Practical implications Public-sector host institutions should prioritise server stability and cross-platform data integration, and programmes can prepare students to work with imperfect systems. Originality/value The study situates legacy systems within public-sector work-integrated learning and offers an explanation of how positive average satisfaction effects and lived frustration can coexist.