The concept of judicial compliance has attracted plenty of attention in the last two decades. Yet, despite the growing scholarly interest, important research questions remain largely unresolved. This is partly due to the persistent use of unsystematic research, built on the cherry picking of cases. The content of only a few well-known judgments has been thoroughly examined, and the rest remains largely ignored by the legal scholarship. The aim of this article is to introduce a sketch of a new three-level approach for improving research on judicial compliance in a multi-level arena. We show how the use of automated text analysis in combination with more traditional legal methods might shed more light on the concept of judicial compliance and judicial dialogues. We explain the procedure of the automated collection of data and their coding and also point out the risks of using automated text analysis when studying judicial compliance. The approach is demonstrated on a single case study of the use of European Court of Human Rights rulings by Czech apex courts. This study assesses how often and in what way the domestic courts engage with the European Court of Human Rights case law.