A Case Study: Product Classification Automation

When working with our customers, our initial goal is to integrate into the existing operating structure. Simultaneously, we start to think out-of-the-box in solutions that will impact our daily job. All the while keeping compliance with the relevant regulations in mind. In this case study, the CATTS team developed a cloud-based application that automated, among other processes, the cross-population of classification codes between different regions.

In this article, you will read about how this tool enabled our team to execute more efficient and accurate classification operations.

The Challenges and Inefficiencies: Manually Copying and Pasting Data

This customer has an extensive database of products used across multiple countries and regions. Often product classifications are requested for new products that have only minor differences from existing ones, such as a different color, or other small deviations. Of course, having an experienced team, many products are easy to classify based on historic knowledge, but when dealing with large volumes, we still pay attention and time to every request. We analyzed the existing classification process and found that at least a third of the effort was spent on manually copying and pasting existing country codes, rationale notes, and other data into relevant fields.

The Optimization Concept

The product classification system used by the customer allows for data extracts and uploads, which meant we had the opportunity to create efficiencies outside the system. We invented a script that replaced the copy-and-paste activity for one product classification assigned to the product ID level to one EU country and automatically assigned that code to the other EU countries. We then expanded that logic for cross-population between different regions. The same product reference ID is often sold in multiple regions and if we have a classification code from one region, we can sometimes directly translate the 10-digit HS code to other regions. We found existing tools often only do this until a 6-digit HS code level. By creating cross-population tables for different regions, we could match the 10-digit code from one region to the possible 10-digit code used in others.

Developing a Cloud-Based Solution

The goal was for our classification experts to run these processes in the background with minimal interruption of their work. Since CATTS has developed cloud-based tools for optimizing other global trade processes, we have the structure and team in place and developed an enterprise-grade solution. Thus, we developed the Furthermore, the cloud-based algorithm was written to be universally applicable, allowing additional country combinations to be added without having to write an entirely new script. Finally, this also opens the door for additional customers to be able to benefit from this development.

Results: Time Savings and Improved Compliance

After running the application for the first four months, it automatically populated data for between 2,500 to 10,000 unique part numbers each month, saving us more than 100 hours per month. This includes the added audit process to check the outcome from the auto-population tool and ensure compliance along the process. In addition to the time savings, our automated system has also helped to reduce the potential for errors and penalties associated with incorrect HS code classification. By taking the guesswork out of the process, our system ensures that goods are properly classified and can move through customs without delays.

Next Steps: Expanding Correlation Tables and Decision Trees

The next phase is to continue building and expanding correlation tables to add more countries and products. The CATTS trade content team helps us ensure that the content is updated as regulatory changes are published. Additional logic or data connections can be further developed on the cloud-based platform. For those cases where a code in one region can lead to multiple outcomes, we are developing a decision tree, so the classification analyst can choose between the possible outcomes and speed up their work. Another future development could be the ability to define rules based on structured hierarchies of product categories or families; or, if products have a threshold percentage of similarity in the product description of previously classified products, an outcome with a weighed certainty would be presented.

In Conclusion

Overall, the automation of cross-populating HS codes has been a valuable addition to our international trade operations, providing significant time savings and improving the accuracy and consistency of the classification of our customers’ goods. It also provides more future perspectives for our customers to benefit from. Can you see yourself benefiting from this type of automation, or do you have other ideas or challenges you would like to work on with us? Reach out and let us know!