Answers are grounded in this paper and cite their sources.
Managers often track metrics they believe can potentially predict performance outcomes and help them improve decisions. However, it is unclear how to best select such predictive metrics out of a wide range of candidate metrics. This study develops and demonstrates an analytical approach to metric selection. First, delete metrics that show too little or too much variation in univariate tests. Second, reveal leading performance indicators with pairwise tests. Third, quantify how much each leading indicator explains performance with econometric models, preferably from different research traditions. Fourth, select the best set of key leading performance indicators by assessing their predictive validity in a holdout sample. Finally, use the selected set of metrics and estimation model to perform what-if analyses for proposed courses of action. The authors demonstrate this analytical approach for the leading national brand and the composite of store-brands in a fast-moving consumer good category.
No takes yet. Share an insight, caveat, or question.
Pauwels et al. (2016) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: