How JD.com’s Smart Supply Chain Powered the 618 Mega‑Sale: Strategies & Algorithms

Overview

During the 618 shopping festival, JD.com’s Y Business Management Department leveraged a data‑driven, intelligent supply‑chain platform to address the challenges posed by the COVID‑19 pandemic and the massive demand fluctuations of the promotion. The goal was to improve inventory efficiency, reduce costs, and enhance the customer experience through coordinated algorithmic solutions.

Pre‑sale Planning and Intelligent Stocking

Before the event, the team implemented collaborative planning and intelligent stocking to ensure adequate product placement across cities. Advanced demand‑uncertainty models provided city‑level probability forecasts for millions of SKUs, enabling real‑time adjustments to procurement and inventory costs.

Demand Forecasting and Replenishment Models

Smart prediction tools generated fine‑grained forecasts, while an end‑to‑end deep‑learning replenishment model—first applied to categories such as dried food and tea sets—optimized stock levels. Dynamic programming with multi‑period planning produced optimal stockpiling recommendations, balancing cost estimates with actual stock‑piling scenarios.

Stockpiling Management System

A dedicated stockpiling management system modeled inventory baseline values and simulated customized cost adjustments. By solving a continuous‑time dynamic programming problem, the system delivered optimal stockpiling quantities, maximizing long‑term profit while minimizing capital risk.

Warehouse Network Optimization

Using a warehouse‑network optimization platform, the team applied exact and tabu‑search heuristics combined with ADMM column generation to produce optimal inventory distribution plans across multiple tiers, constraints, and service levels. This reduced fulfillment latency and cut supply‑chain costs.

Real‑time Inventory Response

During the promotion, a real‑time inventory response system integrated demand spikes, logistics capacity, and regional sales characteristics to dynamically adjust stock levels, ensuring product availability while continuously optimizing procurement and inventory expenses.

Allocation Strategies and Post‑sale Inventory Health

Advanced allocation algorithms identified non‑urgent items for early transfer, flattening demand peaks and improving labor efficiency. Post‑event, a visualized inventory‑cost model identified unhealthy stock, providing data‑driven recommendations for discounting, clearance, or price adjustments to maintain a healthy inventory structure.

Pandemic‑Driven Technical Collaboration

The pandemic accelerated joint development between the supply‑chain platform and ecosystem teams. Upgrades to the core framework—such as multi‑level caching, asynchronous updates, and process aggregation—enhanced system responsiveness. Cross‑departmental “guarantee teams” ensured 24‑hour monitoring and rapid issue resolution.

Framework Upgrades and Business‑Chain Tracking

Key upgrades included:

Baseline framework overhaul to support high‑frequency access to accounts, permissions, and rules.

Formation of a business‑chain tracking squad to coordinate cross‑line issue detection and resolution.

Zero‑level service upgrades that doubled single‑machine performance and passed stress tests for high‑traffic pages.

Pricing Optimization

Price‑health initiatives standardized “real price,” “stable price,” and “low price” strategies across merchants. Data from NPS and sentiment analysis informed price‑adjustment decisions, while the “慧定价” tool leveraged big‑data and AI to refine promotion rules and evaluate effectiveness.

Fulfillment Operations Plan

The fulfillment system introduced a special operation plan for the promotion, including early order transmission, suspension of order cooling periods, and customized split‑order rules to maintain system stability. Simulation tools modeled various rule scenarios, enabling data‑driven decisions that reduced costs and improved delivery speed.

Simulation, Order Pausing, and Capacity Management

Simulation of split‑order rules and real‑time monitoring of paused orders allowed the team to identify bottlenecks, coordinate logistics, and shorten pause durations. Capacity management employed early order transmission strategies to alleviate warehouse pressure caused by pre‑sale order surges.

Conclusion

Through a combination of big‑data forecasting, algorithmic optimization, cross‑functional collaboration, and continuous system upgrades, JD.com’s intelligent supply‑chain successfully supported the 618 promotion, achieving cost reductions, inventory stability, and an enhanced shopping experience despite pandemic‑related disruptions.