The mutual impacts between public participation and social capital has been thoroughly investigated, but little is known about the dynamics of social capital during a participative process. To fill such a gap, an ad-hoc monitoring system of social capital was implemented along an actual participative procedure about the future of an University site. Results show that: i) social capital evolution can follow non-monotonic patterns; ii) different dynamics of social interaction may occur that impact the quality of participation; iii) different participatory tools may lead to different dynamics of social capital; iv) repeated measurements of social capital generate a memory effect which reduces the variation of social capital itself. We also describe relevant drawbacks (resource consuming activities, reduced number of participants, etc.) which may reduce the applicability of the proposed approach. We conclude discussing when an on-going monitoring process is worth to be implemented, and we also describe some relevant side results for those interested to the issue of public participation: i) the announcement of the procedure already generated relevant individual learning; ii) the individual interviews were recognized as an essential learning moment, so we suggest including them into the design of any participative process.