Many tasks with interfaces require entering data accurately, for example, entering patient data in an electronic records system or programming infusion pumps. However, human error is inevitable. Paradoxically, results from cognitive psychology suggest that representing information in a poorer quality format increases the likelihood of memorising the information accurately. This is explained by the dual system account of cognition where slower, more effortful but more accurate thinking is invoked via the poorer quality representation. We present two studies where we transfer these results to the domain of data- entry and show that poorer quality format of to-be-copied information leads to increased accuracy in transcription tasks. Moreover, this is not a consequence of the typical speed-accuracy tradeoffs. The results of our novel approach have implications for the design of data-entry tasks in domains such as healthcare.