Empirical data on learners' conceptual progression is required to design curricula and guide students. In this paper, we present the Reference Map Change Coding (RMCC) method for revealing students' progression at a fine-grained level. The method has been developed and tested through the analysis of successive versions of the productions of eight cohorts (N = 100 total) of high school biology students groups, involved in a year-long inquiry-based learning design. Concepts and causal links expressed in students' gradually refined explanations of biological phenomena are charted onto reference model maps. Trends within variability in all cohorts are measured by a consolidated Prevalence Index (cPI) counting the occurrence of each item across all versions of the students' explanations. Results of a case study presented reveal great variability in patchwork progressions. Learners' diverse and often surprising conceptual paths challenge the view of learning as a linear process. For example, some items consistently appear later, thereby offering empirical evidence of slow spots that require attention. We discuss possible causes, educational implications, and show that our method offers crucial insight into the process of learning as it happens. We finally argue that RMCC also could become a follow-up tool for interested teachers.