There is a large breadth of research studying the manner in which conversers work together to coordinate information and reach a mutual understanding. However, much of the research tends to focus on standard two-party dialogues, with relatively less emphasis on multi-party dialogues. This study examined whether interlocutors in multi-party conversations take longer and encounter more difficulties when collaborating on a common perspective than in two-party conversations. I recruited 36 participants in groups of 3 to take part in a tangram matching task, in which a director described geometric images to one or two matchers. Using a within-participants design, I compared participants’ performance in a two-party context versus a three-party context. The results demonstrated that directors initially described tangrams for longer, matchers were less accurate and more perspectives were offered in the three-party condition than in the two-party condition. Yet, matchers spoke and collaborated to the same extent in both conditions. These findings therefore suggest that addressees in two-party dialogues find it easier to collaborate with speakers on a common perspective, while multi-party dialogues may require more effort from speakers to make themselves understood to addressees, as more perspectives compete to be mutually accepted.