In 2008, in order to improve the match between the offer of childcare facilities for children under 3 and the demand presented by families, fifty-two family allocations offices (caisses d’Allocations familiales or CAF) responded to a call for tenders issued by the Caisse nationale des allocations familiales pertaining to three randomized evaluations of new family services, included as «early-childhood service offerings » . A participatory assessment protocol accompanied the randomized evaluation of fifteen of the fifty-two CAF, from 2007 to 2010. This article discusses the appropriateness of using randomized evaluations as a tool for assessment, construction or public-policy action.