A collaborative virtual learning environment (CVLE) is generally understood as a digital shared multiuser world where students can join and move freely. Such a system provides an ideal setting for peer-assisted learning, since it mimics key affordances found in a regular classroom setting whilst lacking many of the constraints. While many studies explore the feasibility, non-functional attributes and learning impact of CVLEs, there is little empirical work dedicated to 3D collaborative learning environments. Given the current metaverse hype, tools and online services for CVLE implementation are bound to become more commonplace. Thereby, prompting educators to leverage the advantages of such teaching interfaces. This paper intends to set the tone for CVLE best practices and learning expectations by striking a direct comparison, both quantitatively and qualitatively, between face-to-face lecturing and a 3D CVLE. A group of 52 students attended four Games and Interactive Technologies Engineering lectures on gamification and serious games before spending three weeks with a 3D CVLE that comprises the same learning material. We quantitatively compared the two teaching modalities on: (a) effectiveness by means the cognitive, affective, and psychomotor (CAP) perceived learning questionnaire; and (b) student preference through a comparative Likert scale course satisfaction survey. Our quantitative results showed no significant difference in perceived learning between the two modalities and an overall lower satisfaction for the 3D CVLE mode. During the qualitative feedback sessions, we collected the reasons for the lower satisfaction scores and were able to conclude that educators who wish to pursue 3D CVLEs should ensure that: (a) their environment has a high usability rating, with minimal bugs; (b) any tasks they wish to incorporate for learning are interactive and contextual to the learning material; (c) customizable features, especially avatars, are present; and (d) the reward mechanic is not a mere accrual of engagement evidence, but impacts gameplay directly.