Amazon Pharmacy Increases Forecast Accuracy and Reduces Manual Efforts Using AWS Supply Chain

Amazon Pharmacy is licensed in all 50 states to fill and ship prescriptions directly to customers. To do this successfully, Amazon Pharmacy must forecast demand across the entire journey of a prescription—from order intake to fulfillment. Prior to AWS Supply Chain, different teams managed separate forecasting processes, which resulted in longer cycle times, higher resource needs, and limited data granularity. AWS Supply Chain Demand Planning provided the flexibility to generate forecasts at a granular daily level, a capability that enabled Amazon Pharmacy to improve planning agility and meet customer demands effectively.

AWS Supply Chain Demand Planning automates numerous manual tasks such as data entry, calculations, and adjustments, eliminating non-value-added efforts. It leverages machine learning (ML) to analyze historical sales and current data, like open orders, to create accurate forecasts and dynamically adjusted models. This improves forecast accuracy, reduces stockout and limits excess inventory risks.

Amazon Pharmacy now utilizes daily forecasts, applying ML to detect correlations, trends, and seasonality factors previously too complex to determine manually. AWS Supply Chain automatically refreshes forecasts with the latest data and publishes them to downstream and upstream processes, mitigating handoff errors and accelerating critical data refresh cycles.

Amazon Pharmacy emphasizes two key forecasts: T-1 (one week out) and T-5 (longer-term). The T-1 forecast guides immediate staffing decisions to support customer demand, while the T-5 forecast facilitates capacity planning for labor adjustments. The advanced forecasting capabilities of AWS Supply Chain, powered by machine learning, have substantially improved Amazon Pharmacy's forecast accuracy as measured by the key metric of Mean Absolute Percent Error (MAPE). Using AWS Supply Chain, Amazon Pharmacy has seen a significant reduction in its MAPE, demonstrating the solution's effectiveness in generating precise forecasts.

MAPE is a widely used statistical measure to evaluate the accuracy of forecasting methods. It calculates the average absolute percent error between the forecasted values and the actual values, expressed as a percentage. MAPE provides a relative measure of how far the forecasts are from the actual values. A lower MAPE indicates better forecast accuracy. In demand planning and inventory management, MAPE is commonly used as a key performance indicator (KPI) to evaluate and compare the accuracy of different forecasting models or methods, set acceptable forecast error targets such as the industry standard MAPE target of 10 percent or less, and monitor forecast accuracy over time to identify areas for improvement. Amazon Pharmacy achieves a daily MAPE of 5 percent using AWS Supply Chain, indicating strong performance and that the forecasts have a high level of accuracy. This accuracy in projecting prescription unit demand increases inventory agility and enables precise operations planning, ensuring the right staffing levels at the right time to effectively serve customers at the lowest cost.

According to Sachin Pahuja, senior manager of S&OP and Labor Planning at Amazon Pharmacy, "AWS Supply Chain has enabled us to easily integrate our forecasting processes and reduce time-consuming manual tasks, resulting in substantial time savings of approximately 5 hours per week. We now have better identification of correlations, seasonality, and trends, which has significantly improved our forecasting accuracy."

By eliminating repetitive manual tasks, AWS Supply Chain delivers substantial time savings and enhances data reliability. The streamlined process decreases the need for corrective actions, allowing more time for value-added analytics. Amazon Pharmacy has reduced manual adjustments to the forecast models due to increased trust in data accuracy, further reducing effort and manual time.