Outlier events happen on a regular basis in any operational environment. Unfortunately many outliers are simply normal behavior which happens infrequently or simple errors (human data entry errors, sensor errors, sampling errors, data compression errors, etc.). All these outliers are typically bunched up along with the anomalous events making it difficult to identify anomalies within a large group of outliers. Organizations have struggled to move from preventive to predictive maintenance due to this natural phenomenon affecting all operational data. In bolder innovative context, outliers handling and data cleaning tasks are slowing down analytics, machine learning, deep learning and AI projects. Our presentation will look at the workflow utilizing PI OCS to uncover outliers, label anomalies and data errors, and enable data cleaning filters and forecasting anomalies moving forward. This is a first step in a successful journey towards predictive maintenance & AI-based projects.