Complex public health programmes may suffer from 'teething problems' during their implementation in practice, especially when programmes are developed top-down without the participation of end-users and service providers. Trying out and optimising a new concept in practice with stakeholder's participation may enhance the programme's chance of success. In this article we describe the participatory optimisation process of a German primary prevention programme for stroke caregivers, performed before its full-scale implementation. A Participatory Action Research (PAR) approach was applied immediately after the development of the preliminary programme that was led by the professional research team. Iterative PAR-cycles were composed of: observe, reflect, plan and act. Through PAR a matured concept, containing eight Conceptual Building Blocks, was obtained. The five preliminary building blocks: 'Content', 'Human resources', 'Personalised approach', 'Timing', and 'Setting', were prioritised to be the core blocks, providing the base for individual caregiver support. Three new building blocks: 'Network building', 'Communication', and 'Social safety-net' were designated as facilitating blocks, interconnecting and safeguarding the programme. PAR was found to be vital in systematically detecting the conceptual weaknesses of the preliminary concept and in adjusting its components to the needs of the key stakeholders, before full-scale implementation.