Technological innovations have enabled the emergence of online labour market platforms, empowering individuals to penetrate the world of traditional offshoring and challenging localised labour market dynamics. A great number of workers thrive at online platforms and embrace these tools to find customers for their businesses, to counterbalance market fluctuations, and earn wages above the local average. However, online labour market workers are also known to suffer numerous drawbacks, such as precarious working conditions, unpaid work, and severe fragmentation of jobs into tasks that limit skill use and development. Yet, our understanding of what causes this divergence in experiences is limited. Adopting propositions from the labour market segmentation literature, I show that, similarly to offline markets, online labour markets are composed of structurally delimited segments with different social processes governing the allocation of work. Using unsupervised clustering techniques from network science, I show that the clustered skill topology constrains mobility between segments in online platforms. I also show that this segmentation explains large differences in the earnings potential of individual workers. Together, these results provide a new explanation for the persistence of diversified experiences in online labour markets and inform strategies for future research of online platforms as highly segmented labour markets.