Providing estimated delivery time-windows is quickly becoming the norm in modern home delivery. The problem many delivery operations face is that while end-users appreciate shorter time windows, they also hate it when the driver is late (or early) relative to the promised window.
At package.ai, we are embarrassingly passionate about using advanced, data-centric algorithms to optimize last-mile logistics. When Ayal, our algorithms engineer, joined the team, he started looking into the problem of custom-fitting time-windows on a customer-by-customer basis. The goal was to create a statistical model that would maximize the ‘on-time’ rate per a given time-window duration. Specifically, we wanted to see whether we can enable our customer to offer shorter windows than 3-hours for roughly the same ‘on-time’ rate.
Ayal built a model that takes into account spatial and temporal attributes, such as time, date, and route specifications in order to dynamically fit time-windows. After multiple iterations and data manipulations, the new model improved the on-time rate for 3-hour windows by 20%. It was even more impressive for 2-hour windows: an improvement of 40% and the same on-time rate as the traditional 3-hour windows!
The ability to provide accurate AND short time-windows is a big deal for home-delivery operations who care about customer satisfaction and constantly improving service levels (without raising costs). We support such delivery operations with a state-of-the-art technology platform. The beautiful thing is that as more data is collected, our algorithm improves its accuracy and allows our customers to provide even shorter time-windows without compromising the on-time rate. This is just one example of the great promise of big-data and the revolution it is starting to bring to last-mile logistics.
For us at package.ai, calculating optimal time-windows is just the beginning – Jenny, our virtual AI-based operator, goes on to communicate these time-windows to the recipients and autonomously coordinates the delivery with them. When even one recipient changes their schedule, all other deliveries on the route may be impacted which requires further calculations and algorithmic adjustments… But that is a story for a different time 🙂
What’s your time-window story? We’d love to hear what you think about time-windows for home delivery…
P.S – if you’d like us to take a crack at fitting your time-windows more optimally, drop me a note at firstname.lastname@example.org…