How to credibly achieve theoretical saturation while qualitatively analysing millions of observations is a non-trivial question. Facing such an issue, methodological advancements in social sciences propose externalising the inductive pattern retrieval to Machine Learning algorithms with humans acting as evaluators. In contrast, this article offers an alternative solution to the problem of qualitatively analysing large data. In particular, it presents a methodological framework that combines traditional qualitative methods with Machine Learning applications to enhance and scale human-led analysis. In so doing, it presents an iterative workflow, deepening opportunities and challenges and showing prospects for theory building, evaluation, and testing.
