The G7 practices forms of external accountability to answer for its behaviour to the outside world and internal accountability to steer the implementation of what it has decided. To follow up on its development-related commitments, it has set up a permanent framework to produce annual public reports on how G7 national administrations have worked together to implement them. Reports under this framework draw on implementation experience, although G7 Leaders never use them to make decisions on how to carry implementation forward or design new commitments. This is because the G7 process is generally discontinuous and because its accountability process is currently not targeted at channelling feedback from implementation experience back into policymaking. The learning potential inherent in internal accountability is thus not fully used. As a result, the G7 is less effective than it could be in implementing its commitments under changing circumstances. In addition, G7 commitments and methodology do not always make it easy for outside stakeholders to check the G7's words against its behaviour, even though this is important for external accountability. Consequently, this makes it harder to have a rational debate about the legitimacy of the G7. This paper suggests ways to improve the accountability practice of the G7 so that it systematically produces learning effects and thereby better supports G7 legitimacy. The G7 can achieve learning effects by underpinning every commitment with an explicit notion of "how it wants to achieve what" and making sure that this notion gets tested regularly against implementation experience. Closing this feedback loop could be a job for G7 portfolio ministers who could make decisions on further implementation, based on the experience set out in an accountability report. Better designed commitments and improved follow-up would also support G7 legitimacy because this would make it easier for external stakeholders to verify G7 actions against its words.