Popular Plastic and Sanitary Fittings: When a Good Model Leads to a Bad Decision
A data-driven model promised to cut defects at Popular Plastic and Sanitary Fittings. Instead, its recommendation made things worse. The situation raises a critical question: when is a good model ready to guide a real business decision?
At a glance
Industry
Plastics Manufacturing
Setting
Injection-Moulding Production Plant
Core Challenge
Using data analytics to reduce production defects and scrap
Abstract
Popular Plastic and Sanitary Fittings: When a Good Model Leads to a Bad Decision examines the promise and pitfalls of data-driven decision-making in manufacturing operations. Facing persistent production defects in its injection-moulding process, Popular Plastic sought to improve quality and reduce scrap costs through regression analytics. Using historical production data, the company's process engineering team developed a statistically significant model that recommended lowering barrel temperature to optimize performance. The recommendation was quickly incorporated into standard operating procedures, reflecting growing confidence in analytics-based management.
However, the anticipated improvements failed to materialize. Defect rates increased, quality-related costs rose, and several business decisions were influenced by analytical conclusions that later proved incomplete. The case reveals how omitted variables, hidden operational practices, correlated process inputs, and insufficient model validation can undermine even statistically convincing results. It highlights the challenges organizations face when translating analytical insights into real-world operational decisions and demonstrates the risks of relying on historical data without adequately considering production realities and domain expertise.
Through the experience of Popular Plastic, students are invited to explore the relationship between statistical modeling, operational knowledge, managerial judgment, and decision governance. The case raises critical questions about model reliability, data quality, cross-functional collaboration, and the appropriate role of analytics in business decision-making. It provides a practical opportunity to examine how organizations can balance quantitative evidence with human expertise to build more effective and trustworthy data-driven strategies.
Key Topics: Data-Driven Decision Making, Business Analytics and Regression Analysis. Manufacturing Operations Management. Injection Moulding Process Optimization, Data Quality and Model Validation, Statistical Modeling and Managerial Decision-Making, Operational Risk and Process Control, Evidence-Based Management, Cross-Functional Collaboration and Organizational Learning, Analytics Governance and Business Performance
All rights reserved. © 2026 Nikkei Business Lab Asia. No part of this publication may be copied, stored, or transmitted in any form. Copying or posting is an infringement of copyright.
Disclaimers:
(1) Regarding Case Study Content: This case study is based mainly on secondary data and analysis of publicly available information unless otherwise stated, and is intended solely for educational purposes. Any opinions expressed by the author(s) are designed to facilitate learning discussion and do not serve to illustrate the effectiveness of the company. Additionally, banner images and logos used in the case study are intended for visualization in an educational setting and it is not used to represent or brand the company. For any dispute regarding the content and usage of images and logos, please contact the team.
(2) Regarding University Affiliation and Titles of Authors: The university affiliation and titles of author(s) seen in the case study is based on their affiliation and title during the time of publication. It may or may not represent the current status of said author(s).
