INTEGRATING ERP, IOT, AND AI FOR SMART WAREHOUSE AUTOMATION AND LOGISTICS

Abstract

Smart warehouse automation and logistics performance were examined in this study through the lens of closed-loop integration among enterprise resource planning (ERP), Internet of Things (IoT), and artificial intelligence (AI). A quantitative multi-site panel design was applied using 28 warehouse sites tracked across 52 weekly periods, producing 1,456 site–week observations after excluding 20 outage-flagged weeks. Data completeness was high across systems, with ERP 99.3%, WMS/WCS 98.8%, IoT 96.2%, and AI 93.5% field coverage. Integration maturity was measured on a 0–100 scale with a mean of 67.4 and a standard deviation of 11.2, and data quality averaged 0.92 (SD 0.04). Descriptive outcomes indicated inventory record accuracy of 97.1% (SD 1.6), pick errors of 3.4 per 1,000 lines (SD 1.5), order cycle time of 21.8 hours (SD 7.9), exception volume of 148.0 tickets/week (SD 71.0), trailer dwell time of 74.6 minutes (SD 38.2), and on-time shipment rate of 92.3% (SD 4.8). Fixed-effects panel regression models with time-period controls and clustered standard errors showed that higher integration maturity was associated with improved warehouse execution, including higher inventory accuracy (β = 0.031, 95% CI 0.011 to 0.051, p = 0.004), fewer pick errors (β = −0.028, 95% CI −0.045 to −0.011, p = 0.001), shorter order cycle time (β = −0.184, 95% CI −0.239 to −0.129, p < 0.001), and lower exception ticket volume (β = −1.62, 95% CI −2.15 to −1.09, p < 0.001). Logistics outcomes also improved, with higher on-time shipment (β = 0.052, 95% CI 0.026 to 0.078, p < 0.001), reduced trailer dwell (β = −0.92, 95% CI −1.41 to −0.43, p < 0.001), and reduced dispatch delay (β = −0.38, 95% CI −0.57 to −0.19, p < 0.001). Mediation tests indicated partial mediation by data quality; after mediator inclusion, the integration effect decreased for cycle time from −0.184 to −0.081 (p 0.008) and for exceptions from −1.62 to −0.74 (p 0.012), while data quality remained significant for cycle time (β = −0.236, p < 0.001) and exceptions (β = −2.11, p < 0.001), accounting for 45.2%–56.0% of total effects. Collinearity diagnostics were acceptable, with composite-model VIFs of 2.08–2.27 and a maximum condition index of 14.8. Reliability and validity results supported measurement adequacy, including integration maturity reliability (α = 0.89, CR = 0.91, AVE = 0.67) and data quality reliability (α = 0.86, CR = 0.88, AVE = 0.71), with discriminant validity supported by HTMT = 0.78. Overall, the findings indicated that measurable gains in smart warehouse automation and logistics were associated with higher ERP–IoT–AI integration maturity through improved data quality, reduced latency, and stronger performance under volatile, complex, and automated operating conditions.