AI bridges the gaps between custom and standard taxonomies
If you are an SAP MM user, you are probably quite familiar with the Material Groups (MG) categorization field, which is mandatory in any Material Management transaction from Master Data to Purchase Requisition, Orders, Plant Maintenance, and beyond. Moreover, linked to MG, there are the Category Management structure, the budget allocation, the approval workflows, and last but not least, all relevant reporting and spend visibility are related to MG. So, what are Material Groups; they are spending categories organized in a hierarchy that can be structured as a tree or, in other words, as a taxonomy. Custom taxonomies are quite common, particularly within large multinational companies. They are often built top-down, aligning with the organizations’ internal evolution and specific needs. The common characteristic is a variable granularity, meaning that categories are very specific and precise in certain branches related to the core business, and they are generic for other less strategic areas. Another characteristic is the inherent ambiguity of certain categories that are related to the destination of usage of a certain material and not by their very “nature.” This ambiguity, combined with the difficulty of manually choosing one MG among hundreds or even thousands, drives the categorization of the same or similar materials into different categories depending on the criteria applied by the person doing the data entry. As a result, you have materials (especially MRO) that are categorized in mixed buckets, and random errors relating to the intrinsic confusion between nature and usage destination. Errors that can be over 50% of the PO lines, making Spend Analysis and Material Master Data nonreliable.
Simultaneously, there is a global shift towards adopting standard taxonomies to establish a universal language for communication. Standard taxonomies like UNSPSC (United Nations Standard Products and Services Code) mainly follow a bottom-up categorization approach, where products and services are classified based on their inherent attributes and characteristics, leading to a more standardized and transparent classification system. This is the reason why UNSPSC is the classification structure adopted by the major global marketplaces and Source-to-Pay (S2P) platforms.
Well, the first reaction you may have been why do not change the MG structure with UNSPSC? The short answer is that it is a complicated task to do in SAP MM, impacting many transactions and processes. So, to make possible integration between SAP MM and S2P platforms like SAP Ariba a or Coupa , it is necessary to map the Custom taxonomy with UNSPSC. Bridging the gap between a top-down and a bottom-up taxonomy poses significant challenges since each taxonomy is built with different logic; therefore, finding relationships between categories one-to-one is often impossible. If you have a situation one-to-n, it is like in chess when you are in check and must choose a move to escape. Here, there are only two moves, and both lead to checkmate: I either accept the error of selecting one of the “n,” or I choose a more generic category that unites the two taxonomies and end up with categories that are difficult to analyze because they are mixed. This situation hinders the seamless integration of Procure-to-Pay (P2P) systems, resulting in a lack of transparency and clarity in spending management and visibility.
However, the emergence of advanced technologies, particularly AI categorization engines, has revolutionized this landscape. These technologies effectively r esolve the longstanding dichotomy between custom and standard taxonomies, enabling both to coexist and categorizing elements in parallel and independently.
This breakthrough empowers organizations to categorize data easily and accurately in diverse taxonomies. It streamlines the interoperability between internal systems and enhances compatibility with external partners and global marketplaces. Ultimately, the harmonious coexistence of custom and standard taxonomies facilitated by AI technology paves the way for enhanced operational efficiency, improved data visibility, and more seamless integration into the global business ecosystem.