The construction industry has abundant data but lacks sufficient actionable information. Deriving such information is crucial for addressing one of its primary challenges: insufficient productivity. This paper presents a workflow to extract, analyse and present available productivity data on sites in a consistent manner. It was demonstrated using data from structural frames of 11 buildings, totalling 116 floor levels. Seventy-five percent of the levels were found to spend over 3 worker-hours per m2 of floor area on average. Additionally, most buildings exhibited cubic learning curves, showing performance improvements in mid-levels due to worker familiarisation. However, unlike previous studies, all the buildings demonstrated significantly poorer labour productivity towards the end. This might be due to not minimising the workforce proportionally to reduced work, coupled with complexities in logistics and design on higher levels. While having a learning curve does not signify good performance, its shape was found to reflect labour management practices on projects.