To turn the vast volumes of data from sensors and data-feeds into information we need to clean the data and manage its uncertainly. Just because there is a data flow does not mean that it is error-free or makes sense in the real world. This short introduction will cover identifying and cleaning outliers, ‘sense-checking’, and how data ‘noise’ can actually be information. Real-world examples include wasps, temperatures hotter than the sun, and publication metadata. A version of this talk was first presented at the Perth Data Science Meetup - 4 October 2018