Traditional health, safety and environment (HSE) reporting communicates the ‘what’ but not the ‘why’ of safety events. Organisations’ operations-data footprints are growing in volume and velocity but data are often siloed and can be of poor quality. This results in an inability to connect the dots and see through the ‘noise’, to identify patterns of high risk behaviour and root causes of high risk incidents to fully realise the true value of available data and deliver well informed decision making. Deloitte has been working with large organisations across the energy and resources industry, connecting traditional HSE data with contextual data, including employees, contractors, rosters, timesheets, training, and environmental and operational data to surface insights that would otherwise be hidden. By applying exploratory machine learning techniques to these datasets, the sector can gain new insights that were previously ‘hidden’ in data siloes. Drawing on lessons learnt, the paper explains how predictive analytical techniques can enable organisations to identify groups of employees at the highest risk of incidents and, critically, what differentiates these groups, to design tailored interventions and optimally allocate finite resources to manage HSE risk. The paper also describes key factors found to be driving high severity or repeat incidents and details how data conventionally used for asset management and operations optimisation can be analysed alongside HSE data to characterise potential control failures. The outcome is a framework that can be applied to provide continuous controls monitoring of material risks and critical assets.