The development of intermediation platforms, which, unlike on-demand platforms, do not intervene in the choice of clients, raises important questions in terms of equal access to employment and discrimination, which are the subject of growing research interest in sectors other than care work: carsharing (Cook et al., n.d.), seasonal rentals (Cui et al., 2020), and the general classifieds market (Ayres et al., 2015). In the absence of a comprehensive database on homecare platform workers' accounts, research struggles to assess their real impact on the labor market. Many homecare platforms of intermediation that connect supply and demand for services have a strategic use of activity data on their site to attract new users and investors. However, it is a crucial issue for research to go beyond the promotional discourses carried by platform actors on equal opportunities of work and employment.

Departing from a database of web scraped profiles extracted from one major homecare service platform in France, this paper analyzes the role of homecare platforms of intermediation in sorting the workforce and dealing with inequalities in access to employment. We assume that far from being neutral in the encounter between supply and demand, domestic and care work platforms organize through their design and infrastructure and algorithm the criteria for selecting worker's profiles. In that sense, intermediation becomes a real pre-selection process. We aim to explore the impact of socio-technical devices on the way workers present themselves and how clients effectively select them.

We first show, through a semiotic analysis of the design of the platform, the way in which it organizes the competition between homecare workers according to three criteria: emotional skills, experience and price. Then we carry out a regression analysis on the 25159 scraped profiles to isolate the characteristics associated with actually working with the platform, which concerns just over 10% of the profiles registered as workers. We demonstrate that platform's choices do have a real impact on selection because the length of experience highlighted by the platform has a significant impact on the chances of finding work on the platform. But clients manage also in some way autonomy in selecting worker profiles on the platform : the social dumping organized by the platform has little effect, as proposing the highest rates increases the chances of finding at least one mission on the platform as well as speaking French as a second language.