Infios uses machine learning to improve delivery date accuracy

Infios has launched a new solution designed to improve the accuracy of delivery date predictions for e-commerce businesses by employing machine learning. Traditionally, e-commerce companies have relied on estimates provided in carrier service level agreements (SLAs), which are often conservative and can lead to lost sales or customer dissatisfaction.

The new machine learning model utilizes actual delivery times collected through Infios' Freight Audit & Payment (FAP) service. This data provides a more precise picture of when packages actually arrive at customers' doorsteps. The model is trained on this data, combined with information on carrier performance and zip-to-zip transit patterns, to generate more dynamic and realistic delivery date estimates.

The issue with traditional SLA-based estimates is their cautious nature, which can result in abandoned shopping carts if competitors promise faster deliveries. Conversely, overly aggressive estimates can lead to missed deadlines and a decline in customer trust. Infios' model aims to balance these risks by more accurately predicting actual transit times, considering factors such as locations, service levels, and seasonal variations.

The model calculates an estimated delivery date (EDD) from three components: processing time, lead time, and actual transit time. Specifically for transit time prediction, the model analyzes patterns in FAP data to identify when actual deliveries consistently outperform or underperform carrier promises. This approach helps businesses offer more reliable delivery dates and improve customer experience while optimizing their logistics.