In group-living animals, an individual’s fitness is predicted by non-random interactions with other group members and social network analysis has become a powerful tool to study these interactions. We experimentally studied the social network structure in group-foraging subsocial spiders that naturally live in kin groups but accept immigrants. Spiderlings were individually marked and we observed interactions during six foraging trials in groups comprising (i) siblings, (ii) siblings with two non-siblings, and (iii) assorted spiderlings. In this foraging context, we found a higher social network structure in sibling groups compared with assorted groups or sibling groups containing two non-siblings. We asked whether non-siblings in the treatment containing mostly siblings and two immigrants are excluded or less connected, which would explain the overall reduced social network structure of the whole group. We found that non-siblings were not generally excluded but that their presence negatively affected the network structure of the whole group. The connectivity of foreign individuals in this treatment was moreover predicted by their size relative to the other group members with very small and very large spiderlings being well connected. Our findings support the idea that siblings have an advantage over unrelated individuals and that the social network structure may play a role in the evolution of social behaviour in spiders.