Innovation contests are a growing trend among organizations that wish to harness the wisdom of crowds to achieve competitive advantage. Selecting the most promising ideas constitutes a challenge, as such contests generate hundreds or even thousands of ideas. In this context, it is increasingly important to use IT tools to support raters in the convergence process. Thus, it becomes essential to understand the decision processes associated with this task to develop platforms, which will nudge raters towards improved choice accuracy. Considering this goal, we conducted an online experiment in which 190 participants eliminated the least promising ideas in presentation modes with either high (2 ideas/screen) or low (30 ideas/screen) decomposition of information load. We found that higher decomposition of information load leads raters to acquire more information on ideas and exert more judgements. In turn, more judgements to eliminate ideas improved choice accuracy. Our findings add to the growing academic knowledge base on idea selection processes and how IT platforms can be designed to ensure successful convergence processes.