Overview

Amazon has added its SKU‑level restock recommendation engine to the file‑upload path of the Send to Amazon workflow. The change, rolled out in early 2024, means sellers who prefer bulk spreadsheet uploads now receive the same data‑driven quantity suggestions that were previously available only through the interactive list option. This extension helps large‑catalog sellers streamline shipment planning while still retaining control over final numbers.

Key Points

• File‑upload integration — Restock suggestions now appear when sellers choose the spreadsheet route in Send to Amazon, eliminating the need to switch to the list view.
• SKU‑specific guidance — Each recommendation targets an individual product, allowing precise adjustments for high‑volume or slow‑moving items.
• Dedicated template tab — Amazon populates a separate worksheet with suggested quantities, which sellers can edit before re‑uploading.
• Adjustable algorithm inputs — Lead time, restock frequency, economic order quantity and pack size can be customized in FBA settings to reflect a seller’s unique supply chain.
• Designed for bulk sellers — The feature is aimed at merchants managing hundreds or thousands of SKUs, reducing manual calculations and potential errors.

How Restock Recommendations Work

1. Data collection — Amazon pulls four core variables for each SKU: estimated lead time, economic order quantity (EOQ), desired restock frequency, and pack quantity. For example, a private‑label brand importing from China might record a 45‑day lead time and an EOQ of 1,200 units.
2. Historical sales analysis — The system overlays the four variables with past sales velocity, seasonality patterns, and recent inventory levels. A best‑selling kitchen gadget that sold 300 units per week for the last three months will generate a higher suggested shipment than a niche accessory with 20 weekly sales.
3. Quantity recommendation — Using the combined inputs, Amazon calculates a target shipment size that aims to balance stock‑out risk against excess‑inventory costs. In the earlier private‑label example, the algorithm might recommend sending 1,500 units to cover the longer lead time while staying close to the EOQ.