Reducing data requirements when selecting key performance indicators for supply chain management: The case of a multinational automotive component manufacturer

Abstract

The recent trend towards collecting large amounts of data potentially allows organisations to identify previously unknown data patterns that can lead to significant improvements in their performance. However, carrying on collecting this data over time and across numerous locations is expensive. Consequently, when monitoring performance, organisations can be faced with a dichotomy between continuing to collect large amounts of data or whether to use a much reduced set of data. This is a particular problem with Key Performance Indicators (KPIs). Additionally, too many indicators can lead to difficulty in data interpretation and significant overlaps between the indicators, making the understanding and managing of changes in performance more difficult. In this paper, a novel statistical approach is introduced based on the use of Principal Component Analysis (PCA) to reduce the number of KPIs, followed by TOPSIS (Technique for Order Performance by Similarity to Ideal Solution) for validating the results. It is applied to the case of a multinational automotive component manufacturer where 28 KPIs were reduced to 8. The performance of the original set of 28 KPIs was compared with that of the reduced set of 8 KPIs. The peaks of the two TOPSIS time-series coincided, and there was a high correlation between them. Therefore, having the extra 20 indicators provided little extra precision for the considered time interval. Hence, the approach is a valuable tool in helping to reduce a large number of KPIs down to a more practical and useable number.