A common phenomenon is crowdsearch, i.e. when a group of agents is invited to search for a valuable physical or virtual object, e.g. creating and patenting on an invention, solving an open scientific problem, searching for a vulnerability in softwares, or mining for a nonce in proof-of-work blockchains. We study a binary model of crowdsearch in which agents have different abilities to find the object. We characterize the types of equilibria and identify which type of crowd guarantees that the object is found. Sometimes even an unlimited crowd is not sufficient. It can happen that inviting more agents lowers the probability of finding the object, which may also happen when non-strategic agents are added. We characterize the optimal prize and show that having one prize (winner-takes-all) maximizes the probability of finding the object but this is not necessarily optimal for the crowdsearch designer.