Online machinery health monitoring relies on data integrity, so that we can confidently identify subtle change in key failure mode indicators. Some of the most challenging applications are those where the machine is constantly changing speed and may be in a target operating state for only a few seconds. This presentation explores how advances in online diagnostics are enabling in-depth data quality evaluation in real-time, to select which data sets are stored to the database. An actual case study from a steel roughing mill is discussed, showing how online diagnostics can drive improved data quality and enable effective automated monitoring. As industry continues to drive towards big data analytics goals, while also dealing with reduction in experienced practitioners, innovative solutions like this are becoming very relevant.