THB Verhoef, a leading supplier of genuine engine parts for diesel, gas, and hydrogen engines in maritime shipping, was facing the challenge of minimizing the lead time from quote request to invoicing. With an international customer base spanning into the hundreds, achieving a more efficient response to inquiries is essential. The initial step towards achieving this is directed at the sales department, which receives a significant volume of emails on a daily basis.
Currently, there is a delay in responding to these emails, slowing down the entire process. A critical challenge for THB Verhoef lies in the fact that the sales team spends significant time manually sorting through emails and then assigning them to the appropriate personnel. To address this inefficiency, Team Datacation was tasked with developing a solution that automatically categorizes incoming emails based on their type (order, quote request, or other), customer type, and engine type.
Within this complex challenge, we collaborated and communicated extensively, maintaining clear lines of communication for smooth alignment between THB Verhoef and Datacation. To create an intuitive solution, we implemented a temporary method to track the natural sorting behavior of employees. These data were then used to train a text classification model that could accurately determine the context of emails. Furthermore, we leveraged THB Verhoef's deep domain knowledge to recognize engine types, after which we applied our Natural Language Processing skills (NLP) to refine the model.
Our final solution comprises several components. Firstly, we developed a simple text classification model to identify the type of email. Additionally, we created a more advanced model that implements business logic to determine the customer type based on email content. For recognizing engine types, we employed a combination of textual analysis and domain knowledge, resulting in a highly accurate classification of 98.5 percent.
