Category Optimization For Incoming Goods Batches

Challenges and Industry

Our client, a large retailer from the Middle East, with over 40 brick and mortar stores and an ecommerce store selling a wide range of goods, from clothing to toys and home décor, was struggling with operations management. On a daily basis, the client had to deal with a large amount of different categorization from various suppliers. These categories couldn’t be matched or combined precisely, and often did not correspond to the client’s needs. For instance, suppliers often categorized T-shirts differently. While one had it under “T-shirts”, another categorized the product under “Tops” and so on. This made it rather difficult for the client to automate categorization.

The challenge for GFAIVE was to build a solution that would automatically provide a unified categorization for incoming batches, so that the retailer would be able to save valuable time and money on processing and categorization.



  • In the first phase of the project, as a PoC, the GFAIVE team optimized the toy sector in the client’s categorization. As a result, the client got a standalone script that automatically classified and categorized any new toy SKUs  and a one-time report containing detailed explanations and insights on each of the toy categories. The client worked with this new toy categorization for a few weeks and was already able to see the benefits of GFAIVE’s solution and estimate the value that machine-learning based categorization would bring to the organization.

  • In the next phase GFAIVE’s team built a solution based on machine learning algorithms. The algorithms found correlations and patterns in data that a human could not recognize and thus provided very accurate and meaningful categorizations. The solution defined categories in an easy-to-understand format; based on the client’s customers’ behavior and preferences. For example, a category “Women’s jeans” was transformed into two separate categories - “Women’s jeans casual over 40+” and “Women’s jeans 30+” because the behavior of the customers that buy these categories was very different. Another example of restructuring was merging categories that were formed as a result of different supplier categorization.

  • The solution provided the optimal categorization allowing the client’s current software tools to perform much more accurate demand forecasts and distribution plans. The solution categorized accurately and quickly, in an automated manner all new items (SKUs) based on descriptions/categorization from suppliers.



After GFAIVE’s category optimization tool was integrated into the retailer’s systems (without having to change existing IT infrastructure), the following results were observed:


  • The client was able to control inventory on a much more detailed level through accurate categorization. This led to a 10% decrease in inventory holding costs.

  • Having a unified categorization tool that automatically (and thus, quickly) categorizes large volumes of incoming data on new SKUs, saved the company over 300 man-hours/week, as the employees were now able to focus on other relevant tasks. 

  • When the results of our solution were integrated into the client’s existing distribution planning software, they got a more accurate and better understanding of  where and in what proportions to send goods from bought batches. This led to an increased ability to maximize sales as customers were able to get the products they wanted, which in turn increased revenue by 5%.​

© 2021 by GFAIVE - ML demand forecasting for fashion retail. Privacy Policy

  • Instagram - White Circle
  • GFAIVE LinkedIn page
  • GFAIVE Facebook page
  • Twitter - White Circle
  • YouTube - White Circle