本研究探討跨國企業ERP系統導入後，總部標準化治理與海外據點在地作業需求之間的系統變更決策問題。當ERP系統由總部推展至海外據點後，不同地區使用者可能因作業流程、制度環境與實務經驗差異而提出變更需求。若企業缺乏一致且可說明的需求價值判斷機制，具跨區域共用價值的在地創新可能被低估，而局部在地化需求亦可能因缺乏邊界檢核，增加系統複雜度與後續維運負擔。 本研究以半導體製造業A公司為個案，採用質性個案研究方法，透過訪談、參與觀察與文件分析蒐集資料，並運用限制理論之思考流程工具，包括現況樹、衝突圖、未來圖、必要條件圖與行動轉換圖，分析跨國ERP系統變更決策中的不良效應、核心限制與管理衝突。研究發現，A公司ERP系統變更決策的核心限制在於「跨區域需求優先權判定缺乏公認標準」，導致部分具共用價值的在地需求容易被視為單一地區需求而遭到延後或排除。為回應此問題，本研究提出ERP系統變更決策流程模型，將需求評估延伸至跨區域價值、流程節省度、數據透明度、系統穩定度與培訓適用性等構面，並透過需求分類、優先排序與開發路徑判斷，協助企業區分全球標準需求、具共用價值的在地創新、局部在地化需求與應審慎排除之需求。;This study explores ERP system change decision-making after ERP implementation in multinational enterprises， focusing on the tension between headquarters-driven standardization and local operational needs at overseas sites. When an ERP system is extended from headquarters to foreign subsidiaries， users in different regions may propose change requests due to differences in business processes， institutional environments， and local operational experience. Without a consistent and explainable evaluation mechanism， local innovations with potential cross-regional value may be underestimated， while local-specific requests may increase system complexity and maintenance burdens. This study adopts a qualitative case study approach and examines Semiconductor Manufacturing Company A. Data were collected through interviews， participant observation， and document analysis. The Theory of Constraints(TOC) thinking process tools， including the Current Reality Tree， Evaporating Cloud， Future Reality Tree， Prerequisite Tree， and Transition Tree， were applied to analyze undesirable effects， core constraints， and managerial conflicts in cross-national ERP system change decision-making. The findings indicate that the core constraint is the lack of a commonly accepted standard for prioritizing cross-regional requirements. As a result， local requests with shared value may be delayed or excluded. To address this issue， this study proposes an ERP system change decision-making process model that evaluates change requests through cross-regional value， process time savings， data transparency， system stability， and training applicability. Through requirement classification， prioritization， and development path selection， the model helps distinguish global standard requirements， local innovations with shared value， local-specific requirements， and requests that should be carefully excluded.
