When Retail Buyers Panic, A.I. Keeps Calm
During the pandemic, algorithms overrode the human tendency to panic buy.
When Retail Buyers Panic, A.I. Keeps Calm
At the E-Commerce Warehouse, a Distribution Dilemma
- September 03, 2024
- CBR - Operations Management
Online shoppers often place two or more orders in quick succession—an initial one, and then a follow-up one when they remember something else they meant to buy. To counter the rising costs this multiordering creates, some e-commerce platforms hold orders before fulfilling them, but this welcomes its own complications. Chicago Booth PhD student Mohammad Reza Aminian, Columbia’s Will Ma, and Booth’s Linwei Xin built a decision-making model to determine in real time which orders companies should hold.
Some orders are straightforward to fulfill, but other orders present a trade-off.
Imagine a retailer that can hold one order at a time for a maximum of three hours. As another customer’s order comes in, the retailer must decide whether to continue holding the first order or dispatch it in favor of delaying the new one.
Three algorithms for holding orders
Orders that have a high chance of consolidation and those that have more time in the system before being dispatched are good candidates for holding. But these criteria can conflict, creating a trade-off. When this happens, retailers can follow one of three algorithms:
- Reward-rate algorithm: Hold the orders with the highest reward rate.
- Remaining-reward algorithm: Hold the orders with the largest remaining reward, which is the reward rate multiplied by the remaining time to dispatch.
- Exponential algorithm: Hold the orders with the largest value from an equation that considers both the reward rate and the remaining reward (calculated by multiplying the reward rate by an exponential function of the remaining time to dispatch).
The best option depends on the retailer’s capacity constraints and the platform’s busyness. But in most cases, the exponential algorithm generates the best performance relative to a benchmark. Here’s how the algorithms perform among platforms with different holding capacity:
Mohammad Reza Aminian, Will Ma, and Linwei Xin, “Real-Time Personalized Order Holding,” Working paper, November 2023.
More from Chicago Booth Review
Why Some Regions See Gas Price Surges After a Storm—And Why Others Don’t
Regional transportation networks can blunt the impact of supply-chain disruptions.
Why Some Regions See Gas Price Surges After a Storm—And Why Others Don’t
Ever Closer to an Optimally Cost-Efficient Assembly-Line Operation
Researchers create a computationally efficient model for assemble-to-order operations.
Ever Closer to an Optimally Cost-Efficient Assembly-Line OperationMore from Chicago Booth
Your Privacy
We want to demonstrate our commitment to your privacy. Please review Chicago Booth's privacy notice, which provides information explaining how and why we collect particular information when you visit our website.