The iterative process of ethnography not only constructs theory, but its methodology should embody theory. Developing a theoretical framework often demands adjustments in methodology, to leverage previous work and to avoid assumptions compounding through the magnification of blind spots. New theory in policy-engaged ethnography has emphasised the importance of analysing interactions between teachers. However, such analyses have not yet emerged in research on education in Japan, partly because of enduring conceptualisations of teachers as homogeneous. Methodological tools that do exist are grounded in anglosphere data and remain unsuited to research in Japan's schools. As such, participants are positioned amongst, but not in relation to, other teachers and school administrators. Yet, taking teacher heterogeneity seriously enhances the salience of selection bias. The same teachers who are enthusiastic about participating in the research are often those who are enthusiastic about the topic of the research. What is difficult in school ethnography is procuring, and preferably ensuring, participation from teachers who are "not" enthusiasts toward the topic of the research. Designing research with both the heterogeneity of teachers and the specifics of the local context in mind can provide more diverse data with potential to inform analyses of micro-politics in Japan's schools. In reference to a five-year research project and grounded in data from Japan, the paper proposes some procedures to access a diverse range of teachers based on a theory developed to conceptualise teachers' orientations towards policy work.