Risk is difficult to analyse as our data is noisy, uncertainty is high and dependencies are not always obvious. Given that ambiguous and uncomfortable data is easy to ignore, how do we improve our awareness of risk in decisions?