This paper is inquiring a collaboration between a teacher team and researchers in improving science teaching using Clarke and Hollingsworth’s (2002) model of teacher professional growth. We focus this presentation on how the domain of consequences in terms of expected salient outcomes becomes the motor for a shared reflection in this collaboration. The collaboration is at the same time a vulnerable local project, which is shaped in the cross-fire of other projects. The empirical design includes a year-long series of teacher-researcher meetings, recorded by audio and sometimes video. The setting was that of a Swedish secondary school (ages 13-15). These meetings focused the planning and development of lessons within a specific school science area – the human body – making use of both teacher experience – for example in terms of different ways of conducting “good teaching” in this area – and research outcomes – for example the idea of organising purposes (Johansson and Wickman, 2011). The aim and outcome of the inquiry is 1) the explication of conditions for (un)successful collaborations between teachers and researchers, and 2) an understanding of what shapes the local practice and its current and future form. The use of the theory of practice architecture (Kemmis & Grootenboer, 2008) leads to an identification of key sayings, doings and relatings done within and between the meetings, and that together comprise the nexus of actions that make up the teacher-researcher practice. For example, we show some of the negotiations and renegotiations on the use of a diagnostic instrument, starting out from the external domain, changing from instrumental and summative to a powerful formative tool, having concrete meaning in the local domain of practice, and becoming a key doing relating the practice of the planning meetings to the practice of the classrooms.