Dispatch: How China’s JD.com shored up its supply chain during coronavirus

Here, Ella Kidron and Yuchuan Wang from Chinese ecommerce giant JD.com explain the changes it made within its supply chain to ensure it coped with additional demand during the lockdown.

This article first appeared in issue two of the World Retail Congress report The Retail World 2020: Retailing in a time of crisis.

JD.com faces challenge head-on

The outbreak of Covid-19 in China brought widespread challenges to businesses across the country. We had to face this head-on, guaranteeing regular order placement and timely delivery services across China’s nearly 300 cities. The company has delivered 120 million products including 160,000 tons of rice, grains, meat and vegetables to its consumers during the outbreak.

The sudden outbreak of the virus presented four key challenges for JD’s supply chain to overcome:

• A sudden change in product demand resulting in a structural imbalance between supply and demand
• A smaller logistics team than usual due to the Spring Festival holiday, as well as the pressure of closed roads across China
• Ensuring merchant credibility and preventing rising prices of epidemic control supplies
• Managing the challenges brought by the virus’ overall pressure on the economy and insufficient production capacity

Addressing the imbalance

The first problem was in how to address the imbalance between supply and demand. Normally, JD would predict sales and prepare inventory based on historical sales so that inventory would be available where the predicted historical demand is. In general, the demand for masks and other epidemic supplies such as disinfectant and liquid soap is typically low, so inventory in warehouses was also low. Therefore, upstream suppliers maintained a relatively low level of production.

Following the outbreak, the demand for these products skyrocketed, and people attempted to stockpile. If JD allowed people to stockpile freely, then these supplies wouldn’t be allocated sufficiently to meet the real demand.

For example, if we were to allow just a few people in a province with a relatively low number of Covid-19 cases to stockpile all available masks in the market, then those masks might go unused sitting in people’s homes, while the epicentre in Hubei province remained in desperate need of these supplies.

JD’s supply chain scientists used the SEIR (susceptible, exposed, infectious, recovered) model to forecast demand for epidemic supplies to prevent this from happening. The SEIR model assumes that people carry lifelong immunity to a disease upon recovery, and is used to estimate the number of patients, expected patients, and transmission path of a disease in order to be able to forecast the demand for epidemic supplies according to regional population size and to control the demand to ensure that limited materials are distributed to the areas where they are most needed. At the same time, JD’s procurement team mobilised upstream manufacturers to increase production capacity and accelerate supply.

Omnichannel fulfilment

In addition, normally, demand for products like rice, flour and oil is generally higher offline. With the number of people remaining in their homes rapidly increasing due to the epidemic, this demand moved online. Relying solely on the inventory in JD’s warehouses to manage the sudden increase in ecommerce demand of these products would certainly not be sufficient.

To address this, JD-Y (JD’s smart supply chain solutions division) turned to its Omnichannel Fulfilment supply chain innovation program. When a consumer places an order online, the platform matches the order with offline supply closest to the consumer in real-time and then arranges for a courier to deliver to the consumer along the most efficient route.

For example, if a consumer in Beijing places an order for two large 5L bottles of cooking oil and a bag of rice in ordinary times, the products would come from a JD warehouse and then go to a JD delivery station near the customer’s home, from where they would then be delivered to the customer.

Amid Covid-19, instead of relying solely on the stock in JD’s warehouses, the omnichannel fulfilment platform will calculate which brick and mortar store nearest to the customer has the items in stock, and then source the inventory from there instead.

During the epidemic, tens of thousands of offline stores in hundreds of cities have worked with JD to fulfil customers’ orders, and the daily number of orders fulfilled through the program is nearly five times that of a normal day prior to the virus.

Network optimisation

The second problem is logistics capacity. The epidemic occurred during the Spring Festival when some logistics workers were still on holiday. At the same time there were transportation restrictions and road closures, which made transportation and distribution challenging. As a result, some warehouses were unable to supply the cities they used to cover, resulting in numerous delays in fulfilment.

It is also challenging to transport goods from JD’s large regional warehouses to lower-level warehouses such as front warehouses close to cities with limited human resources.

JD-Y’s smart warehouse network platform addressed this issue. ‘Multi-echelon network hierarchical optimisation’ was used to analyse the status of JD’s warehousing network with the updated epidemic information in real-time to calculate and reallocate inventory distribution across the network between cities and provinces. The platform also enables JD’s suppliers to deliver directly to JD’s front warehouses.

During the epidemic period, hundreds of supplies have been supplied directly to JD’s front warehouses instead of going through the entire regional system. As a result, the online availability rate across the country was stabilised at over 95%.

Early-warning model

Another issue was merchant management. With the skyrocketing demand for masks, alcohol and other epidemic materials, some merchants took the opportunity to substantially increase commodity prices, while others attempted to provide false inventory quantities and were therefore unable to distribute products on time after customers purchased them, impacting user experience as well as the reputation of the credible merchants on the platform. It is extremely difficult to discover this type of speculative behaviour manually.

Based on intelligent forecasting and abnormal data detection algorithms, our supply chain scientists launched a risk early-warning model, which identified more than 200 suspected abnormal price rises and overselling actions that the company was then able to address.

Finally, there are the challenges brought by economic pressure and insufficient industrial production capacity at a certain time in the future. During the epidemic, JD.com’s supply chain scientists are studying data on the impact of historical epidemic events, and combined with the data during the novel coronavirus, in order to forecast future trends of various industries.

JD is using this information to actively communicate with suppliers and partners about the future situation, difficulties encountered, and planning and deployment, in order to develop the most effective recovery plan.