Zalando SE is one of Europe’s leading online platforms for fashion, footwear, accessories and beauty. Since its founding in 2008, the firm has grown rapidly and now operates in 25 European markets, serving over 51 million customers. Headquartered in Berlin, Zalando's broad portfolio includes products from more than 7,000 international and local brands, alongside its own private labels.
Technology is at the heart of Zalando's strategy of building the next generation of personalized e-commerce. But this does not stop with consumer-facing activities. This commitment extends to its international trading operations. Complexity and compliance are major challenges in this area, which is why the firm turned to MIC for improving and increasing the automation level of its customs and trade compliance processes.
The Importance of Efficient Tariff Classification for E-Commerce
Customs tariff classification is a key activity for Zalando as a major exporter. This is a critical but notoriously complex process. Accuracy is non-negotiable, as misclassification can lead to delays, fines and compliance issues. To meet this challenge, smart solutions are needed, and this is where MIC's AI solution comes in.
We deployed our AI-based classification tool across Zalando's operations to take its customs tariff classification process to a whole new level. We used increased automation to ensure smoother trade flows and full transparency in classification, while minimizing human error and greatly speeding up the processes. Here's how we did it.
MIC AI Classifier: Automation Meets Accuracy
Zalando had already been using the MIC CCS module to increase the automation level in the customs tariff classification process. This utilizes intelligent algorithms to determine the correct customs tariffs and ensure compliance with customs regulations. Additionally, MIC’s Global Trade Content Service (GTCS) regularly updates customs tariff content, keeping Zalando up to date with evolving regulations.
However, to take the process a step further, Zalando recently adopted the MIC AI Classifier, which provides even greater levels of automation and accuracy. The new AI solution has streamlined the classification process for a vast number of products, dramatically improving the speed and accuracy of the tariff classification.
The MIC AI Classifier leverages cutting-edge artificial intelligence and machine learning technologies like fine-tuned Foundation Models to further increase the automation level of the goods tariff classification process. This ensures unparalleled accuracy, compliance, and efficiency in aligning products with the correct tariff codes.
As well as general available information like harmonized tariff system nomenclature published by each country, binding tariff information and many more, our Foundation Model is also fine-tuned with Zalando's internal information. This includes more than 740.000 articles and 3.8 million images and uses Natural Language Processing (NLP) and Computer Vision to suggest the best matching tariff number every time.
Additionally, the AI Classifier provides a suggestion certainty score for each tariff recommendation, indicating the confidence level in the assignment. This enables Zalando to greatly improve the efficiency of its export operations.
Georg Knopf, Director Customs & Global Trade at Zalando, said: "Thanks to MIC’s AI solution, we’ve tremendously increased the automation level of a complex process and gained speed, accuracy and scalability - turning tariff classification into a real competitive edge."
How MIC's AI Solution Drives Real Results
Zalando classifies up to 540,000 items every month. Before the introduction of the MIC AI Classifier, this demanded significant manual effort. However, the technology has been able to reduce processing time for each classification by an average of 71 percent.
This means that, when fully rolled out across all departments, the system is expected to handle up to 1.3 million items per month - almost a threefold increase. What's more, 49 percent of resources previously allocated to tariff classification have been saved, allowing employees to turn their attention to other value-adding activities within the company.
All this has been achieved while maintaining the highest levels of accuracy. The software delivers correct tariff codes in 97 percent of cases for the top two suggestions and 99 per cent for the top five. Even more impressive is the first-suggestion accuracy rate, which stands at 92 to 93 percent. This gives Zalando confidence in the system and reduces the need for random checks.
Mr Knopf added: "Increasing automation levels in tariff classification has fundamentally transformed our export workflows. What used to be a manual bottleneck is now a strategic advantage. With MIC’s AI-powered classification module, we’ve turned one of our most resource-intensive compliance tasks into a fast, scalable, and reliable process."
Looking Ahead: What's next for Zalando and MIC?
The new classification software clearly demonstrates the potential of Artificial Intelligence in customs processes. For Zalando, it not only relieves operational burdens, but also provides a strategic advantage in international logistics, making the business faster, more precise and more resource-efficient than ever before.
This solution was implemented very quickly thanks to the agile approach used by MIC's team and close collaboration with Zalando throughout. The deployment sets a standard that could serve as a blueprint for the entire industry when it comes to improving essential customs classification processes.
Mr Knopf said: "With MIC’s AI-driven solution, we’ve transformed our tariff classification process: boosting accuracy, cutting processing times and supporting our goal of fully automated exports."
