Automating Procurement Practices Using Artificial Intelligence

Automating spend analysis for procurement practices, our novel three-component classification model processes unstructured spend texts to replicate expert decision-making. Using data from Cranswick PLC, our model improves supplier categorization accuracy and identifies cost-saving opportunities, with projected annual savings of £16–£22 million. Conducting a spend analysis of procurement practices is a challenging task for manufacturers. It requires deciphering large-scale spend data in the form of unstructured texts and identifying opportunities for savings. This process relies on procurement experts’ know-how and is often performed manually, a laborious task often leading to missed savings opportunities. Automating spend analysis through natural language processing and machine learning presents several challenges, such as (i) a lack of true detailed category labels for suppliers, (ii) a lack of sufficiently large sets of training data, (iii) hierarchical taxonomies that vary across manufacturers, and (iv) the reduced accuracy of hierarchical categorization algorithms beyond two levels. Our novel three-component classification model tackles these issues, facilitating the automation of spend analysis and the replication of procurement experts’ decision-making processes. By processing input data composed of unstructured spend texts from Cranswick PLC, a leading UK food producer, our model delivers accurate supplier categorizations that pinpoint areas ripe for substantial savings. This approach not only shows greater accuracy compared with existing benchmark models but also aids in identifying key product categories and suppliers for cost-saving initiatives. By simulating the application, we project that our method could bring annual savings of £16 million to £22 million ($20 million to $28 million) for Cranswick PLC, illustrating the significant advantages of automating spend analysis. History: This paper was refereed. Funding: This work was supported by Innovate UK [Grant 80093].