We present a user study where subjects are asked to solve fact-finding search tasks, where a visual clue is provided in addition to text. Search tasks with visual clues are often
encountered in real life, for example when we come across an object, such as a ower or a landmark, that we would like to know more about. Our study reveals that these tasks are
much harder to solve than the text-based ones, and users employ very different strategies for solving them. Furthermore, even though image search is extensively used to solve visual search tasks, there are several features that an ideal image search engine must possess before it can be effectively used for solving such tasks.