Automatic Product Classification for Partyboss with PowerSuite.ai
Sector
Retail
Products
20,000+
Employees
20 to 50
For webshops with a large and dynamic assortment, a clear product structure is essential. In this case article, you’ll learn how Partyboss automates product classification using PowerSuite.ai and an AI-driven classification environment while ensuring consistency, speed, and scalability.
Background
Partyboss is the e-commerce company behind Feestkleding.nl and Carnavalskleding.nl. The company focuses entirely on party and carnival clothing, complemented by a wide range of accessories.
The assortment consists of tens of thousands of items and grows rapidly, especially during peak periods such as carnival and holidays. Due to these constant changes, new products are regularly added and existing items are updated.
The Challenge.
- Time-consuming manual product classification
Product classification was largely done manually. For each item, employees had to determine which product group, line, and subline it belonged to. This process was labor-intensive and time-consuming, especially given Partyboss’s large and growing assortment.
- Faster time-to-market
During busy periods like carnaval, the assortment expands quickly. New products need to be available online as soon as possible. Partyboss therefore looked for a solution that could speed up the classification process without compromising quality or control.
- Inconsistent product classification
Because classification was done manually, differences in interpretation arose. Similar products were sometimes categorized differently, especially when it came to subjective attributes such as themes and subthemes. This resulted in a less consistent product structure.
Our solution.
- AI-driven product classification
PowerSuite.ai proved to be the best-fitting AI solution for Partyboss’s challenge. The system analyzes product descriptions and images to determine where a product belongs within the webshop structure.
- Machine learning and automatic information extraction
The solution uses machine learning and Large Language Models to recognize patterns in existing product data. Product texts are automatically analyzed, extracting relevant attributes from descriptions and using them for classification.
- Multi-label classification model
The AI model can assign multiple labels to a single product. This means an item can, for example, be categorized under both “sailor costume” and “pirate costume.” This leads to a more complete and consistently structured product catalog.
The result
PowerSuite has empowered Partyboss to....
Realize accurate product classification
Product groups are now automatically classified with 99.6% accuracy. Lines and sublines are also assigned very reliably, resulting in a consistent product structure.
Accelerate time-to-market for new products
Because classification is largely automated, new products can be published online more quickly. This is especially important during peak periods when the assortment grows rapidly.
Set up a scalable process for assortmentmanagement
Most of the classification work is now handled automatically. This saves employees time and allows them to focus on improving and optimizing the product structure rather than performing manual data entry.
