In 2014, Tech Crunch reported Amazon filing a patent for “Anticipatory Shipping”, i.e., getting the package ready for shipping even before a customer orders the package. The coverage on this issue has been minimal.
Today, through a student (ht: PB), here is an article on Economist (subscription required) about German firm Otto using AI and machine-learning to purchase 200,000 items a month from third-party suppliers without any human intervention. The orders are made purely based on machine learning on 3 billion prior transactions and a variety of variables (weather, etc.).
Two important factors stood out from the article.
- Customers are less likely to return if the product arrived within two days. Pre-shipping helps eliminate risks due to shipping delays. (There is also some evidence that customers price-hop when there is a shipping delay, which leads to more returns).
- Customers apparently dislike multiple shipments, instead preferring to bundle items into one shipment. This is perhaps true, but I have not seen hard data on the distribution of shipping bundle preference. On a related note, Amazon sometimes encourages “No-Rush Shipping” by offering a coupon to try Amazon Pantry.