In this study data-feedback in a cyclic model of data-driven teaching was used to enhance the teaching behavior of students registered in a master course for teachers. Differences between pre- and post-test measures in a simple one-group pre-test post-tests design proved to be significant with effect sizes ranging from d = 0.29 to d = 0.76. Improving teaching behavior in a time span of only six weeks on average is remarkable since earlier studies indicated that it takes over 15 years to master complex teaching skills with a ‘natural development’ of teaching skills of about 25% of a standard deviation.