Purpose Despite growing research to increase the effectiveness of online idea generation, few studies have investigated the impact of the task instructions on individual creativity. This research scrutinizes how the originality of submitted ideas varies according to two different types of task instructions: unbounded task instructions as opposed to bounded ones (defined as suggestive or prohibitive). Design/methodology/approach This research analyses to what extent the degree of boundedness of task instructions impacts the originality of the ideas produced on a global online idea generation platform called eYeka (www.eyeka.com). This thesis is an empirical study of more than 1,600 ideas produced by ordinary participants and three types of designers (graphic designers, media designers, and graphic & media designers). Findings First, the results show that suggestive task instructions significantly decrease the score of originality of ideas compared to unbounded task instructions. However, providing unbounded task instructions, where participants can freely generate ideas does not seem to be the optimal strategy, as prohibitive task instructions result in higher originality scores for the generated ideas. Second, it is demonstrated that the impact of task instructions depends on the design skills of the participants. Indeed, even if designers do not consistently produce more original ideas than do ordinary participants, in this study the graphic & media designers group demonstrate better performances. More specifically, in the case of bounded task instructions (i.e. suggestive and prohibitive), no other differences are found between the designers and the ordinary participants with the main exception of graphic & media designers who have lower originality scores with prohibitive task instructions. Finally, graphic & media designers are found to produce more original ideas than the ordinary participants for the case of unbounded task instructions. Research limitations/implications Implications: This thesis allows bridging a gap between the fixation effect theory and the literature about the organizational aspects of crowdsourcing. Additionally, this research compares the performances of ordinary participants with that of designers. It demonstrates that, depending on the task instructions, the performance of ordinary participants and designers differs significantly. Suggestions for future research and limitations in the research process: Generalizability would require running the same exercise with (1) other forms of task instructions or Abstract Guidance version Date: 20/01/2015 Updated: 10/02/2015 Page 2 of 2 crowdsourcing purposes and (2) other Internet platforms. Further research around the influence of task instructions in crowdsourcing effectiveness offers vast opportunities. Some of these potential avenues for investigation are: (1) play on the purpose of the task instructions, (2) test task instructions in other sectors than consum