Losing track of who is in a conversation, and what is being said, is always a problem especially on audio-only conference calls. This paper investigates how domain-independent social feedback can support such interactions, and improve communication, through the use of audio cues. In particular, we show how an agent can improve people’s ability to accurately identify and distinguish between speakers, reassure users about the presence of other collaborators on the line, and announce events like entry & exit with minimum impact on users cognitive ability.