Departing from the classical collaborative knowledge acquisition methodology in expert systems literature, the ‘collaborative value modelling’ framework was developed. It combines two decision support processes: Web-Delphi to elicit individual judgemental knowledge from a (very) large number of stakeholders, and Decision Conferencing (DC) in which the knowledge acquired is then digested by a small group of key-players, to collaboratively develop a widely informed multicriteria value model. This new sociotechnical framework was already tested in real complex evaluation contexts, supported by the Welphi platform and the MACBETH method. This paper focuses on the Welphi process for weighting population-health objectives that informed a subsequent multicriteria DC to construct a European Population Health Index under the EURO-HEALTHY H2020 project. It shows that enhancing MACBETH decision conferencing with an ex ante Web-Delphi process fostered higher participation and collaboration in multicriteria modelling. Furthermore, it shows how the Welphi platform, by collecting participants’ changes of opinion (or judgements) between Delphi rounds, provides a mean to understand the knowledge construction process itself. Current research under the IMPACT HTA H2020 and MEDI-VALUE FCT projects addresses both technical and social challenges, on the one hand, to find ways for improving the platform in order to understand the reasons behind opinion changes and knowledge generated by participants’ interactions and comments, and on the other hand, to explore behavioral aspects within and across Delphi, DC and MACBETH web-based group decision support systems.