Crowdsourcing is an increasingly important approach to pursuing innovation. In crowdsourcing ideation websites, crowd members are often exposed to some stimulus ideas, such as examples provided by companies or peers' ideas. Understanding the effect of being exposed to original stimulus ideas in this context may inform the design of the crowdsourcing process. To test this effect, an experiment was conducted where crowd workers were asked to design a public service advertisement. Depending on the experimental condition, the participants were exposed to an original idea, or a common idea, or no idea. As compared to the absence of exposure, exposure to an original idea decreased fluency, defined as the number of ideas generated by each person, and increased the average originality of ideas generated by each person. By contrast, exposure to a common idea had no effect on either idea originality or fluency. The semantic similarity between the stimulus idea and the first idea generated was higher when the stimulus was common versus original as measured by latent semantic analysis. The implications of these results for research and practice are discussed.