Google trends as an early signal in international flower trade

Abstract

This study investigates the predictive potential of Google Trends for U.S. flower imports, examining whether fluctuations in online search volumes can anticipate changes in import patterns. Monthly flower import data from 2004 to 2023 were analyzed alongside four Google search indices using a comparative evaluation of time series models, including ARIMA, ARIMAX, and SARIMAX. The analysis reveals clear predictive patterns, emphasiz- ing the importance of incorporating seasonality into forecasting. Results indicate that the SARIMAX model, which explicitly accounts for seasonal components, consistently outperforms simpler models, achieving a mean absolute percentage error (MAPE) of 7.04%. Among the four search indices examined, only the distribution index—comprising terms related to florists and delivery services—demonstrated statistically significant predictive power. This finding suggests that consumer interest in flower accessibility and delivery strongly influences import dynamics. The study contributes original evidence supporting the use of Google Trends as an early indicator within the international flower trade. By integrating online search behavior into forecasting models, stakeholders can gain timely insights into market demand, potentially improving supply chain management, inventory plan- ning, and operational responsiveness. Overall, this research highlights the value of digital search data as a practical, cost-effective tool for anticipating trends in global floral markets and informing strategic decisions.