It is now common knowledge that as many as 40% of the control loops in most industrial processes have considerable potential for improving control performance by reducing variability. Because of the large number of control loops in an industrial plant, controller performance monitoring is indispensable, but equally important is how to prioritize their maintenance. It is well known that variance reduction in a loop occurs by transferring variability to other variables or loops. The focus of this study is to propose a methodology to prioritize loop maintenance based on the potential improvement of each loop and the variability transfer among them. The central point of this work is the Variability Matrix (VM), an array that shows the impact of performance improvement of a given loop on the whole plant. Based on the VM, a methodology to translate this array into a potential loop economic benefit metric is also introduced. The VM can be quantified in the ideal scenario where plant model and controller are available and also when they are not, thus allowing the application of these ideas in industry. The efficacy of proposed methodology is illustrated by successful application to two case studies.