This paper recounts the process of undertaking a randomised controlled trial which was designed to examine the effectiveness of an intervention for socially isolated older people aged 75 years and over. It describes the reasons for early cessation of the study and raises the implications of this outcome for policy, practice and research. The intervention under investigation was designed to alleviate loneliness and foster companionship. It involves participants being linked with a small group of others through a teleconferencing system with each group being facilitated by trained volunteers. There was a requirement to recruit and train a minimum of 30 and a maximum of 60 volunteers over 1 year to facilitate 20 friendship groups to meet the number of older people required to be recruited to the study. Problems with recruiting and retaining the volunteer workforce by the voluntary sector organisation, who were commissioned to do so, led to the study closing even though older people were recruited in sufficient numbers. The paper draws upon analysis of various data sources from the study to identify the potential reasons. The discussion raises considerations regarding the extent of infrastructure required to deliver community services to vulnerable user groups at scale, identifies some of the issues that need to be addressed if such volunteer-initiated services are to be successful and informs future research programmes in this area.