Organizations spend millions of dollars on analytical and simulation inventory models to optimize inventory across the supply chain. However, these methods are expensive, difficult to implement and fail to capture all the requirements for defining inventory levels across supply chain. The effect of planned and unplanned equipment downtime, a key factor, is not properly utilized in these models. Many methods use standard statistical distributions, which do not fully capture downtime characteristics of an operation. This may lead to inaccurate computation of inventory sizes. The purpose of this paper is to communicate a new analytical approach of defining inventory levels which is robust enough to consider non-normal distributions associated with equipment downtime. The proposed method is easy to use, less time consuming and can be adapted quickly with changing operational dynamics. The method is based on utilizing equipment performance data to scientifically compute inventory levels. It has been piloted in a major automotive manufacturing facility with very significant results and now is in the plant-wide replication mode.