JD.com Order Transfer, Inventory Management, and Fulfillment Workflow Overview
This article explains JD.com's order transfer process, inventory hierarchy, support relationships, order transfer and planning systems, fulfillment workflow, and risk control mechanisms, illustrating how millions of orders are allocated, scheduled, and executed across multiple warehouses and channels.
JD.com processes millions of orders per minute from various channels (PC, app, WeChat, etc.) that are initially pooled in the OFC (Order Fulfillment Center) pool. The order transfer step determines which warehouse will produce each order, acting as a core distribution plan.
Order transfer is essentially a distribution mechanism that decides the target warehouse and production schedule, linking directly to inventory status.
Inventory Levels
1) Primary Inventory : Early stage where each major city (Beijing, Shanghai, etc.) had its own central warehouse.
As order volume grew, this model became insufficient.
2) Secondary Inventory : RDC (Regional Distribution Center) serves as a central/comprehensive warehouse, while FDC (Front‑line Distribution Center) acts as a pre‑positioned warehouse. Example: Jinan and Tianjin are FDCs.
JD.com now operates seven major regions (Beijing, Shanghai, Guangzhou, Chengdu, Wuhan, Xi'an, Shenyang). An order from Jinan first checks local stock; if unavailable, the Beijing warehouse supports it.
3) Support Relationship
Major cities hold most stock for fast‑moving items, while long‑tail items are supported from larger hubs (e.g., Beijing) to smaller regions (e.g., Jinan), enabling efficient coverage of both high‑frequency and low‑frequency demand.
Key Objectives of Order Transfer
The goal is to create a production plan that quickly routes orders to the appropriate production system, simplifying complex flows for in‑stock orders.
1) Order Transfer Service System
Defines how, where, and when an order is produced, serving upstream OFW (Order Fulfillment Workflow) by receiving order data and invoking middleware for transfer.
Scope: self‑operated and POP (Platform‑Owned Products) orders; responsibilities include fast‑track processing for non‑reserved in‑stock orders.
Dependencies: OFW, delivery configuration, inventory, large‑appliance scheduling, middleware.
2) Order Planning Engine System
Creates production plans that meet delivery promises while optimizing cost and efficiency.
Focuses on non‑in‑stock and reserved orders, handling complex scenarios such as multi‑warehouse, FDC, and parallel inventory.
Scope: self‑operated and POP orders; dependencies include delivery center configuration, inventory, product data, large‑appliance scheduling, Promise, VPR, middleware, and OFW‑COI.
How Transfer Works
The process first checks inventory; unlike earlier splitting steps that ignored stock, transfer aligns orders with available inventory and determines the production location.
JD distinguishes between "available stock" (ready for production) and "reserved stock" (pre‑stocked but not guaranteed).
Front‑end inventory is SKU‑based, while OFC inventory is order‑based.
Order Fulfillment Workflow
After an order enters the system, it is received, split, and passed to downstream services (e.g., split service, transfer service). The workflow includes:
Receiving orders from pipelines and middleware.
Gathering production data and controlling business flow before warehouse entry.
Pushing order data to nationwide warehouses.
Handling re‑splitting, cancellations, modifications, returns, and status callbacks.
Providing data to promise and BI systems, as well as daily operational tools.
Real‑time monitoring and alerting of OFW operations.
The system has two main functions: order information back‑propagation and downstream transmission.
Order Status Back‑Propagation
Real‑time status (e.g., printing, packaging, outbound) is sent to ERP, triggering inventory, outbound, tracking services, and customer notifications.
Non‑Status Business Data Back‑Propagation
Details such as package count, weight, carrier, and invoice are sent to WMS, TMS, and DMS.
Risk Control
Risk control prevents malicious gift‑abuse (e.g., exploiting promotions by ordering a high‑value item with a zero‑price gift that is shipped from a different warehouse). The system links the main order and the gift order so that if the gift is received, the associated main order can be cancelled, mitigating fraud.
IT Architects Alliance
Discussion and exchange on system, internet, large‑scale distributed, high‑availability, and high‑performance architectures, as well as big data, machine learning, AI, and architecture adjustments with internet technologies. Includes real‑world large‑scale architecture case studies. Open to architects who have ideas and enjoy sharing.
How this landed with the community
Was this worth your time?
0 Comments
Thoughtful readers leave field notes, pushback, and hard-won operational detail here.