This paper reports empirical findings from a European research project that explores the ecosystem services (ES) concept as a boundary object in local land use planning. Our theoretical starting point is grounded in classical Pragmatism: we posit that the uncertainties and knowledge gaps that emerge when the ES concept is taken beyond its mainstream interpretation can motivate stakeholders to question their settled beliefs about how they live from, with and in their local environment. This allows a conception of ‘ecosystem valuation’ as a cognitive social process based on the interpretation of signs. A sign is something, for example a parcel of farmland, which stands to somebody, for example a visiting tourist or a local grower, for something, for example a sense of place and history, in some respect or capacity. Such a semiotic perspective can be grounded in real-world land use planning by considering that actors’ settled beliefs about their environments are embedded in the institutions (norms, rules, shared strategies) that govern their actions. Understanding these beliefs (rather than ‘measuring preferences’), how they are formed and how they might change, can therefore inform local land use planning in practical terms. We reframe and trial the ES concept as a “boundary object” at the nexus of science, policy and practice, addressing three questions: (i): can the ES concept help to better articulate and demonstrate public benefits from peri-urban agri-ecosystems? (ii) what forms of knowledge can be harnessed at the local planning scale to (e)valu(at)e ES?; and (iii) what role do representations play in the process of semiosis? We present preliminary findings from two peri-urban ES case studies in Southern France: the Thau Lagoon south-west of Montpellier; and the Costières de Nîmes south of Nîmes. The Thau case study involved both planners (experts) and broader stakeholders (actors) and addressed the future of agricultural lands under pressure from urban development; the Nîmes case study involved technical experts (experts) only and addressed ecosystem impacts of railway infrastructure development. Our method involved individual interviews and participatory stakeholder workshops. Artefacts employed included lists of services, photographs, satellite-based land uses maps, land cover classifications and dynamic ecosystem matrices linking selected services to specific land cover, land use and land function. Our findings suggest that ES representations (ES maps, typologies, frameworks) can be conceived as shared outputs rather than prerequisite inputs to the participatory land use planning process.