The influence of delivery riders' work experience on potential delivery riders' app downloads: insights for HRM in the gig economy

Purpose This study examines how rider-generated app store reviews serve as recruitment signals in the gig economy and how their meanings vary across algorithmic governance architectures, shaping platform attractiveness among potential delivery riders. Design/methodology/approach Using 273,460 reviews (2017–2025) from China's delivery-rider apps, the research relates review-derived work-experience signals to daily app downloads, treated as a proxy for early-stage attraction. Drawing on signaling theory and an adapted Attraction–Selection–Attrition (ASA) lens, this paper conceptualizes platform–signal fit as the alignment between governance-produced work-experience cues and how prospective riders plausibly decode those cues under uncertainty. Work-experience topics are extracted from rider reviews via Latent Dirichlet Allocation (LDA), attraction-sensitive associations are estimated using quantile regressions, topics are consolidated into four signal dimensions (Perceived Efficiency, Autonomy, Economic Return, and Algorithmic Risk Transparency) through Principal Component Analysis (PCA) and Confirmatory Factor Analysis (CFA), and a modified Analytic Hierarchy Process (AHP) is applied to translate cross-quantile regularities into prioritized human resource management (HRM) guidance. Findings Findings reveal the cross-architecture asymmetries. Perceived Efficiency is negatively associated with attraction on batch-order platforms but positively associated on point-to-point platforms, indicating semantic bifurcation. Algorithmic Risk Transparency is positively associated across architectures and is most concentrated under mainstream (mid-quantile) attraction conditions. Originality/value The study contributes a theory-consistent, non-causal mapping of platform–signal fit and a replicable workflow for converting review-based signals into testable, segment-sensitive recruitment interventions.