Reproductive performance has recently been a growing concern in cattle dairy systems, but few research methodologies are available to address it as a complex problem in a livestock farming system. The aim of this paper is to propose a methodology that combines both systemic and analytical approaches in order to better understand and improve reproductive performance in a cattle dairy system. The first phase of our methodology consists in a systemic approach to build the terms of the problem. It results in formalising a set of potential risk factors relevant for the particular system under consideration. The second phase is based on an analytical approach that involves both analysing the shapes of the individual lactation curves and carrying out logistic regression procedures to study the links between reproductive performance and the previously identified potential risk factors. It makes it possible to formulate hypotheses about the biotechnical phenomena underpinning reproductive performance. The last phase is another systemic approach that aims at suggesting new practices to improve the situation. It pays particular attention to the consistency of those suggestions with the farmer’s general objectives. This methodology was applied to a French system experiment based on an organic low-input grazing system. It finally suggested to slightly modify the dates of the breeding period so as to improve reproductive performance. The formulated hypotheses leading to this suggestion involved both the breed (Holstein or Montbéliarde cows), the parity, the year and the calving date with regard to the turnout date as the identified risk factors of impaired performance. Possible use of such a methodology in any commercial farm encountering a biotechnical problem is discussed.