Imaginal retraining (IR) is an approach‐avoidance procedure that has shown promising results in previous studies. The aim of the present study was to dismantle the efficacy of IR's components in a randomized controlled trial (RCT). We conducted a RCT with nine conditions comprising eight intervention groups and a waitlist control group (WLC). Alcohol craving (primary outcome), consumption, depressive symptoms, quality of life, subjective appraisal, and side effects were assessed online at baseline, post intervention (6 weeks), and follow‐up (12 weeks). The sample consisted of 426 participants (age: M = 47.22, SD = 11.82, women: 50.5%). The intervention groups received instructions for four different components of IR (mood induction, mental avoidance of unhealthy stimuli, motor avoidance of unhealthy stimuli, approach to healthy stimuli) that were each conveyed with or without prior psychoeducation (PE). The intervention was delivered online. At total of 163 individuals (42.9%) used the intervention at least once. No group differences were found for any primary or secondary outcome after Šidák correction. Uncorrected statistics showed effects of significantly decreased alcohol consumption for the approach + PE group in the intention‐to‐treat and the merged motor avoidance group in the per‐protocol analyses at post assessment compared with the WLC. Exploratory moderation analyses revealed that individuals with high visualization skills benefited most. The authors conclude that visualization training and motivational components may increase the efficacy and adherence of IR.