Evaluation of learning collaboratives (LC) needs to account for not just outcomes and context, but also the mechanisms participating teams use to implement and normalise new practices. Normalisation process theory (NPT) mechanisms-, , and -were used to do a constant comparison coding of transcripts of weekly calls between team coaches and mentors during a 9-month LC to implement team-based primary care in 13 health centres. Both the positive and negative (eg, lack of ) use of normalising mechanisms, as well as when they occurred over time, were noted. Findings suggest that normalising mechanisms are not linear, but work concurrently in real time, in a recursive fashion and in negative and positive ways. Clarity of purpose () became clearer as teams met regularly, and optimised team relational work and commitment to using a shared quality improvement process (). Similarly, the concurrence of and likely refined each other. It took 3-4 months for most teams to establish sufficient and , and to access actionable data. Nine months was not enough time for some teams to both implement and change using data. A separate analysis indicated that prominent topics of discussion were interactions within the team, its relationship with the larger organisation, and difficulties accessing data and determining its reliability. Teams which experience sufficient positive aspects of normalising mechanisms are able to tolerate the unevenness and negative aspects of normalising change to succeed.