‘Law-as-data’ is often understood in terms of computational data. This article reframes it around a research-centred notion of data, broadening the methodological possibilities available to empirical legal research. This approach transcends the qualitative/quantitative divide, identifying common challenges that arise when law is treated as research data. Methodological tools from the social sciences and examples from recently published socio-legal scholarship are used to demonstrate how law presents unique challenges at the stages of conceptualisation, operationalisation and coding. These challenges stem from factors such as the non-commensurability of legal texts, the fluidity and contested meaning of legal notions, the difficulty of systematically gathering legal data and the risk of bias towards data-driven research rather than law- or reality-centred inquiry. However, strategies and best practices can be developed to address or mitigate these risks. Legal academics are particularly well suited to this task, given the high level of expertise required to effectively leverage law as research data. The article argues that a true methodological naturalisation of law as research data, which embraces the unique characteristics and limitations of law as a data source, would greatly benefit legal studies by diversifying its methodological toolbox while preserving its specificities.