Within Business Intelligence (BI) systems, an industrial Key Performance Indicator (KPI) is a measurement of how well the industrial process in the organization performs an operational activity that is critical for the current and future success of that organization [1]. The industrial leading indicators are one type of KPIs that present key drivers of industrial business value, are predictors of future outcomes. Thus leading indicator discovery is critical to success of the industrial objectives. There are some challenges in leading indicator discovery. The traditional approach depending on domain experts’ experiences is labor-intensive and error-prone. In addition, because the time shifts between industrial KPIs are vague and often inconstant for variability of business concerns, the correlation between them cannot be correctly calculated using the traditional distance functions. In this paper, we propose a semi-automatic system with an iterative learning process for discovering leading indicators to help trace anomalies and optimize the industrial objectives. Finally two industrial case studies are conducted by applying the proposed methods in the production printing application. The proposed system has two key differentiations and novelties: (1) the semi-automatic framework uses temporal data mining techniques combined with domain knowledge to enable timely access to KPI analysis, and anomaly tracing; and (2) an iterative learning method continuingly uncovers the “root” leading indicators along with the changes of business environment.