The seemingly straightforward question of optimal pig delivery weight is more complex than meets the eye. Despite abundant research insights, the industry continues to request additional applied scientific decision support on the delivery weight problem. The current objective is to investigate whether and how the complex decision of delivery weight can be reshaped (reframed) into a more tangible and comprehensible system of factors that matter for making the right decision. We used a participatory decision problem analysis, which resulted in modelling blueprints that incorporate factors prioritised by stakeholders for determining optimal delivery weights. How to efficiently organise such a ‘problem reframing process’ is case-specific: it depends on the objective, the initial problem understanding by the stakeholders, and their learning potential. Efficient co-learning is a prerequisite for successful participatory problem analysis. Our study reveals that the first step in such a process of ‘problem reframing’ should therefore be to answer the question of how to effectively organise co-learning among stakeholders and researchers, instead of starting with a correct and detailed representation of the problem. Useful guidelines for participatory problem reframing processes are (1) providing sufficient participatory learning steps, (2) having few and clearly defined objectives per learning step, (3) providing adapted learning tools per step, (4) establishing a common language and (5) deliberately choosing stakeholders based on prior knowledge of the problem or its context, potential motivation or incentives to be part of the participatory process step and potential role in up-scaling the co-learning process to a larger group of beneficiaries.

Agricultural policy and advisory instruments increasingly operate in heterogeneous farm populations, yet conventional farm typologies often impose mutually exclusive categories that obscure mixed configurations and boundary cases. This paper develops and demonstrates an empirical approach to constructing a farm business-model typology that maps farm-level indicators onto the business model dimensions and estimates archetypal configurations, allowing mixed membership to be assessed empirically rather than imposed by hard clustering. Using farm-level data from the 2020 Belgian agricultural census ( n = 3,810 Flemish dairy farms), we apply archetypal analysis to identify empirically grounded business-model archetypes and quantify mixed membership. We operationalise value creation, delivery, and capture through indicators of fodder-based composition, diversification, labour and land organisation, and economic size. A four-archetype solution emerges: Grassland Stewards (GS), Integrated Feed Landholders (IFL), Entrepreneurial Diversifiers (ED), and Scale Maximisers (SM). Membership patterns indicate substantial overlap between archetypal poles, with mixed membership prevalent across the population. Sectoral concentration patterns further show that some configurations account for a large share of farms and labour, whereas others account for a disproportionately large share of cows and standard output despite representing a smaller share of farms. The resulting configuration map can inform benchmarking and differentiated instrument design by clarifying where diversity is concentrated, how boundary configurations are structured, and which operating logics face distinct organisational and market conditions.