Excel-based Automated Testing: Reducing Manual Validation Time

Executive Summary

This report covers an automated testing framework that turns time-consuming manual HS code validation into a fast, scalable process. The solution tackles the challenge of validating large datasets of product descriptions against expected HS codes, cutting validation time from over a month of manual work to just 4 hours of automated processing. The system processes Excel files containing product descriptions and expected HS codes through a FastAPI endpoint. Users upload their validation datasets in Excel format, and the system automatically processes each entry, compares results against expected outcomes, and generates validation reports. In the first run, the system processed 11,000 product descriptions in about 4 hours a task that would have taken more than a month manually. This is a productivity increase of over 180x, completely changing what's possible for large-scale HS code validation testing. The solution eliminates the confusion and inconsistency of manual testing while providing standardized, repeatable validation. This framework enables ongoing quality checks for HS code classification systems and supports rapid iteration of classification algorithms.

Figure 1.1. Example output from the HS code classification system. The automated Excel testing shows a sample product description alongside the system's recommended 10-digit HS codes, Custom Clarification, Optimized Product descriptions, and LLM Analyzer results. Additional columns indicate whether the classification was successful and list any detected errors.

Introduction

Large-scale testing is essential for validating the accuracy and reliability of HS code classification systems. As these systems process thousands of product descriptions daily, ensuring consistent and accurate classification is critical for maintaining compliance and operational efficiency. Traditional validation has relied on manual testing—both time-consuming and prone to human error. Manual validation typically involves subject matter experts reviewing individual product descriptions, determining appropriate HS codes, and comparing results against system outputs. This requires significant expertise, attention to detail, and time, making it impractical for large-scale validation. The complexity grows exponentially with thousands of products. Maintaining consistency across multiple reviewers and extended time periods becomes nearly impossible. The need for large-scale testing became clear as HS code classification systems evolved to handle bigger datasets and more complex product descriptions. Organizations need to validate system performance across diverse product categories, test edge cases, and ensure accuracy improvements don't introduce new errors. Manual approaches can't scale to meet these requirements within practical timeframes.

The Excel-based automated testing framework addresses these challenges by standardizing validation through automated workflows. The system accepts Excel files with two columns: product descriptions and expected HS codes. This simple format lets users prepare test datasets using familiar tools while the automated system processes thousands of entries systematically. The framework processes each product description through the HS code classification system, captures the results, and performs detailed comparison analysis against expected outcomes. This ensures consistent evaluation criteria, eliminates human bias, and provides thorough reporting on system performance across large datasets. By automating validation, organizations can perform large-scale testing regularly, validate system improvements continuously, and maintain confidence in their HS code classification accuracy at scale. This is essential for organizations processing high volumes of international trade transactions where classification accuracy directly impacts compliance and costs.

The Problem

Manual testing of HS code classification systems created significant bottlenecks that limited the ability to validate system performance at scale. The traditional approach required human experts to manually review each product description, determine the appropriate HS code, and compare results against system outputs—a process that was both time-consuming and inconsistent. For large validation datasets, manual testing became impractical. Processing 11,000 product descriptions through manual validation would take more than a month of dedicated expert time, making thorough testing expensive and lengthy. This extended timeline prevented regular validation cycles and delayed system improvements. The manual process also introduced confusion and inconsistency. Different reviewers might interpret product descriptions differently, leading to varied expected results. Human fatigue and attention drift over extended testing periods further compromised accuracy and reliability.

Additionally, manual testing lacked standardization and repeatability. Each validation cycle required starting from scratch, with no systematic way to track improvements or regressions across different system versions. The absence of automated documentation made it difficult to identify patterns in classification errors or measure progress over time. These limitations meant that clients faced a choice between thorough validation and practical timelines. Most organizations were forced to accept limited testing coverage due to resource constraints, creating uncertainty about system performance across diverse product categories and edge cases.

The Solution

The Excel-based automated testing framework eliminates the inefficiencies and limitations of manual validation. The solution centers around a FastAPI endpoint designed for large-scale testing scenarios, accepting Excel files with a standardized two-column format containing product descriptions and expected HS codes.

The automated system processes each row in the uploaded Excel file systematically, passing product descriptions through the HS code classification system and capturing detailed results. The framework performs comparison analysis between system outputs and expected codes, generating detailed validation reports that highlight accuracy metrics, error patterns, and performance statistics.

The solution maintains the familiar Excel interface users already understand while adding automation behind the scenes. Users can prepare test datasets using standard spreadsheet tools, ensuring the validation process integrates smoothly with existing workflows and requires no specialized technical knowledge.

The automated processing pipeline handles large datasets efficiently, processing thousands of entries in hours rather than weeks. The system maintains detailed logs of all processing activities, enabling thorough analysis of both successful classifications and errors. This tracking provides insights that would be impossible to gather through manual testing.

The framework also ensures consistency and repeatability across validation cycles. The same Excel file can be processed multiple times to track system improvements, and standardized reporting formats enable easy comparison of results across different testing periods or system versions.

Results

The Excel-based automated testing framework delivered major improvements in validation efficiency and capability. The most significant achievement was processing 11,000 product descriptions in approximately 4 hours, compared to the estimated month or more required for manual validation. This represents a productivity improvement of over 180 times the previous manual approach.

The automated system eliminated the confusion and inconsistency problems that plagued manual testing. Every product description was processed using identical criteria and logic, ensuring consistent evaluation standards across the entire dataset. The systematic approach removed human bias and fatigue factors that previously compromised validation accuracy.

The framework enabled testing coverage that was previously impractical. Organizations can now validate their HS code classification systems across diverse product categories, test edge cases systematically, and perform regular validation cycles without significant resource investment. This supports continuous improvement and quality assurance.

The standardized Excel input format proved highly accessible to users, requiring no specialized training or technical expertise. Clients can prepare validation datasets using familiar tools and processes, making the automated testing framework immediately useful without additional learning curves or system integration requirements.

The detailed reporting and logging capabilities provide clear visibility into system performance patterns. Organizations can now identify specific areas for improvement, track progress over time, and validate that system enhancements deliver expected benefits without introducing new classification errors.

Conclusion

The Excel-based automated testing framework successfully transforms HS code validation from a time-consuming manual process into an efficient, scalable automated capability. By reducing validation time from over a month to just 4 hours for 11,000 product descriptions, the solution fundamentally changes the economics and practicality of comprehensive system testing. The framework addresses the core challenges that previously limited validation efforts: time constraints, inconsistency, and lack of scalability. The automated approach ensures consistent evaluation criteria while enabling testing coverage that would be impossible through manual processes within practical timeframes.

Processing 11,000 product descriptions demonstrates the framework's capability to handle real-world validation requirements at scale. This automated testing capability enables organizations to maintain confidence in their HS code classification systems through regular validation cycles. The familiar Excel interface ensures immediate usability without requiring specialized training or complex system integration. Users can leverage existing spreadsheet skills to prepare test datasets, making automated validation accessible to organizations regardless of their technical sophistication. This automated testing framework enables continuous quality improvement for HS code classification systems. Organizations can now validate system changes quickly, test new algorithms thoroughly, and ensure classification accuracy across diverse product categories. This transforms validation from an occasional, resource-intensive activity into a routine quality assurance process that supports ongoing system optimization and reliability.