This paper presents a subsectoral analysis of historical occupational data. We describe a new way of coding and producing occupational data, using an occupational graph rather than specific schemes. We then employ compositional data analysis and new metrics derived from properties of ternary plots to offer a more detailed and complex and dynamic image of the relationship between economic growth and structural change. As a result of this, it highlights a series of core sub-sectoral patterns, which were until now often overlooked by traditional sectoral analysis or invisible due to higher levels of data aggregation. We find that longer historical series confirm Rodrik’s pattern of ‘premature deindustrialisation’ (Rodrik 2016), including the rapid movement towards tertiary subsectors for later developers and the diminishing gender gaps in structural change. We develop a typology to identify which subsectors are most correlated to economic growth and discuss what that may indicate in terms of theories of structural change. The empirical examination of economic development adds historical depth to ongoing academic and policy-oriented discussions on the nature and contribution of structural changes to economic development.