The metaverse offers new possibilities in collaborative working, especially in training, as it can reduce costs, enhance safety, and allow learning without affecting the real world. This study examines the impact of the immersion of the training environment (low to high: computer, metaverse, and real world) on subjective and objective performance (mistakes, time, and long-term memory), perceived cognitive effort, and enjoyment—based on an adaption of the cognitive fit and task-technology fit theories—for training assembly tasks to be performed in the real world using LEGO models. Our lab experiment (n=30) shows no significant differences in performance exist, except that training in the metaverse lasted the longest, the perceived cognitive effort for the metaverse and the real world was the lowest, and enjoyment was particularly high for the metaverse. Thus, in addition to performance, other soft factors are important for the success of a technology, emphasising the human factor in learning.