Occupants affect energy consumption in buildings by contributing internal heat gains, increasing internal carbon dioxide levels and adapting their behaviour. Estimated occupancy schedules are used in building energy models for regulatory compliance purposes and when empirical data are not available. Metadata, such as personal location data, is now collected and visualised on a global scale and can be used to create more realistic occupancy schedules for non-domestic facilities, such as large retail outlets. This paper describes a protocol for extracting and using freely available metadata to create occupancy schedules that are used as inputs for dynamic simulation models. A sample set of twenty supermarket building models are used to demonstrate the impact metadata schedules have when compared with models using the estimated schedules from regulatory compliance. Metadata can be used to create bespoke occupancy profiles for specific buildings, groups of buildings and building archetypes. This method could also help reduce the gap between predicted and actual performance. In the example models, those using the regulatory compliance schedules underestimated heating demand by approximately 10% and overestimated cooling demand by over 50% when compared to models using the metadata schedules. Although this work focuses on UK facilities, this methodology has scope for global application. Crown Copyright (C) 2017 Published by Elsevier B.V. All rights reserved.