A large distributor of industrial tools
Our client had over 800,000 SKUs in their catalog and were about to expand their digital footprint. All of their processes were in place except that searches on their ecommerce site did not bring up the right products. Additionally, they were unable to implement proper filters on the left hand navigation. This meant that many of the products in their catalog were not accessible by their online customers.
What could the problem be? The product content itself was in good shape - as far as digital assets, titles and descriptions were concerned. They had high quality images, latest user guides and warranties and detailed specification documents. So enrichment was not the problem.
As our team continued with the analysis of the product data, we noticed that there were an unusually large number of products categorized under an L1 category called “Miscellaneous”. Now, we get it - we do need a catch-all for times when new products come in and have to be quickly onboarded so that they are in the digital catalog, but the hierarchies may not support these new products yet. So a miscellaneous bucket is certainly useful.
However, when more than half of the products are under “Miscellaneous”, it becomes extremely concerning. An urgent reclassification of half a million SKUs was required, which, if done manually, would incur prohibitive costs and resources.
Our team examined all the products in the “Miscellaneous” bucket, and quickly identified at least 15 new “Level 1 categories”. Our Taxonomy Management Tool was used to add the new categories into their product hierarchy and seamlessly sync to catalog.
Now that the taxonomy was well-organized, the dataX Auto-Classification tool stepped in. This uses a machine learning model that is trained on most product hierarchies. All we needed to do was fine-tune it to the client’s data. That done, the SKUs were being auto-classified into each of the new L1 categories at an average rate of approximately ~250,000 SKUs per month.
Using a combination of machine learning and human expertise to intelligently automate the process, our client was able to achieve a 70% savings on cost and time, compared to their earlier process that relied heavily on manual effort.
