Automation of HS codes assignment for Nestle
Machine-learning models assign HS codes from product descriptions to automate customs classification for Nestlé.
Machine-learning models assign HS codes from product descriptions to automate customs classification for Nestlé.
Challenge
Nestlé needed to automate the manual assignment of HS codes to products. Manual data entry consumed employees’ time and introduced classification errors.
HS codes are an internationally standardized system for classifying traded products. Each item in an international shipment needs the correct code on its commercial invoice to determine import duties and other taxes.
Data
The algorithm was trained using millions of product descriptions and their matching HS codes. Each pair of digits in the code represents a specific level of product classification.
Solution
Artelnics developed a machine-learning model that assigns an HS code based on a product’s description. Training connects descriptions with the corresponding codes and produces assignment rules based on individual words or combinations of terms that distinguish products.
Product descriptions labelled with their matching HS codes.
Automatic HS code assignment using learned classification rules.
Results
The company reduced manual data entry and human errors in HS code assignment, improving efficiency and freeing warehouse employees to focus on more complex tasks.
Illustrations

Millions of labelled descriptions are prepared for training.
The model learns the patterns that connect text with HS codes.
Relevant words and combinations are converted into assignment rules.
New products receive the most appropriate HS code with fewer manual errors.