In the world of big data, stored manufacturing data creates opportunities and challenges. Typically, manufacturing data has an informative time stamp. Time stamps must be rationalized from various data sources. The work process and physical relationship between the variables can be defined. Prediction is rarely sufficient; models should be interpreted to inform improvement efforts. Data in manufacturing has traditionally depended on designed experiments and aggressive observational studies. However, these miss out on the larger inference space offered by stored data. Further, these dynamic studies are entirely dependent on subject matter knowledge, science and direct experience. Big data can access all the wealth of experience stored in it. This presentation will go through a series of examples, giving practical tips on how to seize the opportunities and avoid the pitfalls.